forked from logzhan/RobotKernal-UESTC
Merge pull request 'main' (#1) from ray/RobotKernal-UESTC:main into main
Reviewed-on: http://logzhan.ticp.io:30000/logzhan/RobotKernal-UESTC/pulls/1LRD_Develop
commit
eb34e7f105
|
|
@ -0,0 +1,403 @@
|
|||
// Copyright (C) 2011 Carl Rogers
|
||||
// Released under MIT License
|
||||
// license available in LICENSE file, or at http://www.opensource.org/licenses/mit-license.php
|
||||
|
||||
#include "cnpy.h"
|
||||
|
||||
#include <stdint.h>
|
||||
|
||||
#include <algorithm>
|
||||
#include <complex>
|
||||
#include <cstdlib>
|
||||
#include <cstring>
|
||||
#include <iomanip>
|
||||
#include <regex>
|
||||
#include <stdexcept>
|
||||
|
||||
char cnpy::BigEndianTest(int size)
|
||||
{
|
||||
if (size == 1)
|
||||
return '|';
|
||||
int x = 1;
|
||||
return (((char*)&x)[0]) ? '<' : '>';
|
||||
}
|
||||
|
||||
char cnpy::map_type(const std::type_info& t)
|
||||
{
|
||||
if (t == typeid(float))
|
||||
return 'f';
|
||||
if (t == typeid(double))
|
||||
return 'f';
|
||||
if (t == typeid(long double))
|
||||
return 'f';
|
||||
|
||||
if (t == typeid(int))
|
||||
return 'i';
|
||||
if (t == typeid(char))
|
||||
return 'i';
|
||||
if (t == typeid(signed char))
|
||||
return 'i';
|
||||
if (t == typeid(short))
|
||||
return 'i';
|
||||
if (t == typeid(long))
|
||||
return 'i';
|
||||
if (t == typeid(long long))
|
||||
return 'i';
|
||||
|
||||
if (t == typeid(unsigned char))
|
||||
return 'u';
|
||||
if (t == typeid(unsigned short))
|
||||
return 'u';
|
||||
if (t == typeid(unsigned long))
|
||||
return 'u';
|
||||
if (t == typeid(unsigned long long))
|
||||
return 'u';
|
||||
if (t == typeid(unsigned int))
|
||||
return 'u';
|
||||
|
||||
if (t == typeid(bool))
|
||||
return 'b';
|
||||
|
||||
if (t == typeid(std::complex<float>))
|
||||
return 'c';
|
||||
if (t == typeid(std::complex<double>))
|
||||
return 'c';
|
||||
if (t == typeid(std::complex<long double>))
|
||||
return 'c';
|
||||
|
||||
else
|
||||
return '?';
|
||||
}
|
||||
|
||||
template <>
|
||||
std::vector<char>& cnpy::operator+=(std::vector<char>& lhs, const std::string rhs)
|
||||
{
|
||||
lhs.insert(lhs.end(), rhs.begin(), rhs.end());
|
||||
return lhs;
|
||||
}
|
||||
|
||||
template <>
|
||||
std::vector<char>& cnpy::operator+=(std::vector<char>& lhs, const char* rhs)
|
||||
{
|
||||
// write in little endian
|
||||
size_t len = strlen(rhs);
|
||||
lhs.reserve(len);
|
||||
for (size_t byte = 0; byte < len; byte++) {
|
||||
lhs.push_back(rhs[byte]);
|
||||
}
|
||||
return lhs;
|
||||
}
|
||||
|
||||
void cnpy::parse_npy_header(unsigned char* buffer, size_t& word_size, std::vector<size_t>& shape, bool& fortran_order,
|
||||
std::string& typeName)
|
||||
{
|
||||
// std::string magic_string(buffer,6);
|
||||
uint8_t major_version = *reinterpret_cast<uint8_t*>(buffer + 6);
|
||||
uint8_t minor_version = *reinterpret_cast<uint8_t*>(buffer + 7);
|
||||
uint16_t header_len = *reinterpret_cast<uint16_t*>(buffer + 8);
|
||||
std::string header(reinterpret_cast<char*>(buffer + 9), header_len);
|
||||
|
||||
size_t loc1, loc2;
|
||||
|
||||
// fortran order
|
||||
loc1 = header.find("fortran_order") + 16;
|
||||
fortran_order = (header.substr(loc1, 4) == "True" ? true : false);
|
||||
if (fortran_order)
|
||||
throw std::runtime_error("npy input file: 'fortran_order' must be false, use: arr2 = np.ascontiguousarray(arr1)");
|
||||
|
||||
// shape
|
||||
loc1 = header.find("(");
|
||||
loc2 = header.find(")");
|
||||
|
||||
std::regex num_regex("[0-9][0-9]*");
|
||||
std::smatch sm;
|
||||
shape.clear();
|
||||
|
||||
std::string str_shape = header.substr(loc1 + 1, loc2 - loc1 - 1);
|
||||
while (std::regex_search(str_shape, sm, num_regex)) {
|
||||
shape.push_back(std::stoi(sm[0].str()));
|
||||
str_shape = sm.suffix().str();
|
||||
}
|
||||
|
||||
// endian, word size, data type
|
||||
// byte order code | stands for not applicable.
|
||||
// not sure when this applies except for byte array
|
||||
loc1 = header.find("descr") + 9;
|
||||
bool littleEndian = (header[loc1] == '<' || header[loc1] == '|' ? true : false);
|
||||
assert(littleEndian);
|
||||
|
||||
// char type = header[loc1+1];
|
||||
// assert(type == map_type(T));
|
||||
|
||||
std::string str_ws = header.substr(loc1 + 2);
|
||||
loc2 = str_ws.find("'");
|
||||
word_size = atoi(str_ws.substr(0, loc2).c_str());
|
||||
if (header.substr(loc1 + 1, 1) == "i") {
|
||||
typeName = "int";
|
||||
} else if (header.substr(loc1 + 1, 1) == "u") {
|
||||
typeName = "uint";
|
||||
} else if (header.substr(loc1 + 1, 1) == "f") {
|
||||
typeName = "float";
|
||||
}
|
||||
typeName = typeName + std::to_string(word_size * 8);
|
||||
}
|
||||
|
||||
void cnpy::parse_npy_header(FILE* fp, size_t& word_size, std::vector<size_t>& shape, bool& fortran_order,
|
||||
std::string& typeName)
|
||||
{
|
||||
char buffer[256];
|
||||
size_t res = fread(buffer, sizeof(char), 11, fp);
|
||||
if (res != 11)
|
||||
throw std::runtime_error("parse_npy_header: failed fread");
|
||||
std::string header = fgets(buffer, 256, fp);
|
||||
assert(header[header.size() - 1] == '\n');
|
||||
|
||||
size_t loc1, loc2;
|
||||
|
||||
// fortran order
|
||||
loc1 = header.find("fortran_order");
|
||||
if (loc1 == std::string::npos)
|
||||
throw std::runtime_error("parse_npy_header: failed to find header keyword: 'fortran_order'");
|
||||
loc1 += 16;
|
||||
fortran_order = (header.substr(loc1, 4) == "True" ? true : false);
|
||||
if (fortran_order)
|
||||
throw std::runtime_error("npy input file: 'fortran_order' must be false, use: arr2 = np.ascontiguousarray(arr1)");
|
||||
|
||||
// shape
|
||||
loc1 = header.find("(");
|
||||
loc2 = header.find(")");
|
||||
if (loc1 == std::string::npos || loc2 == std::string::npos)
|
||||
throw std::runtime_error("parse_npy_header: failed to find header keyword: '(' or ')'");
|
||||
|
||||
std::regex num_regex("[0-9][0-9]*");
|
||||
std::smatch sm;
|
||||
shape.clear();
|
||||
|
||||
std::string str_shape = header.substr(loc1 + 1, loc2 - loc1 - 1);
|
||||
while (std::regex_search(str_shape, sm, num_regex)) {
|
||||
shape.push_back(std::stoi(sm[0].str()));
|
||||
str_shape = sm.suffix().str();
|
||||
}
|
||||
|
||||
// endian, word size, data type
|
||||
// byte order code | stands for not applicable.
|
||||
// not sure when this applies except for byte array
|
||||
loc1 = header.find("descr");
|
||||
if (loc1 == std::string::npos)
|
||||
throw std::runtime_error("parse_npy_header: failed to find header keyword: 'descr'");
|
||||
loc1 += 9;
|
||||
bool littleEndian = (header[loc1] == '<' || header[loc1] == '|' ? true : false);
|
||||
assert(littleEndian);
|
||||
|
||||
// char type = header[loc1+1];
|
||||
// assert(type == map_type(T));
|
||||
|
||||
std::string str_ws = header.substr(loc1 + 2);
|
||||
loc2 = str_ws.find("'");
|
||||
word_size = atoi(str_ws.substr(0, loc2).c_str());
|
||||
if (header.substr(loc1 + 1, 1) == "i") {
|
||||
typeName = "int";
|
||||
} else if (header.substr(loc1 + 1, 1) == "u") {
|
||||
typeName = "uint";
|
||||
} else if (header.substr(loc1 + 1, 1) == "f") {
|
||||
typeName = "float";
|
||||
}
|
||||
typeName = typeName + std::to_string(word_size * 8);
|
||||
}
|
||||
|
||||
void cnpy::parse_zip_footer(FILE* fp, uint16_t& nrecs, size_t& global_header_size, size_t& global_header_offset)
|
||||
{
|
||||
std::vector<char> footer(22);
|
||||
fseek(fp, -22, SEEK_END);
|
||||
size_t res = fread(&footer[0], sizeof(char), 22, fp);
|
||||
if (res != 22)
|
||||
throw std::runtime_error("parse_zip_footer: failed fread");
|
||||
|
||||
uint16_t disk_no, disk_start, nrecs_on_disk, comment_len;
|
||||
disk_no = *(uint16_t*)&footer[4];
|
||||
disk_start = *(uint16_t*)&footer[6];
|
||||
nrecs_on_disk = *(uint16_t*)&footer[8];
|
||||
nrecs = *(uint16_t*)&footer[10];
|
||||
global_header_size = *(uint32_t*)&footer[12];
|
||||
global_header_offset = *(uint32_t*)&footer[16];
|
||||
comment_len = *(uint16_t*)&footer[20];
|
||||
|
||||
assert(disk_no == 0);
|
||||
assert(disk_start == 0);
|
||||
assert(nrecs_on_disk == nrecs);
|
||||
assert(comment_len == 0);
|
||||
}
|
||||
|
||||
cnpy::NpyArray load_the_npy_file(FILE* fp)
|
||||
{
|
||||
std::vector<size_t> shape;
|
||||
size_t word_size;
|
||||
std::string typeName;
|
||||
bool fortran_order;
|
||||
cnpy::parse_npy_header(fp, word_size, shape, fortran_order, typeName);
|
||||
|
||||
cnpy::NpyArray arr(shape, word_size, fortran_order, typeName);
|
||||
size_t nread = fread(arr.data<char>(), 1, arr.num_bytes(), fp);
|
||||
if (nread != arr.num_bytes())
|
||||
throw std::runtime_error("load_the_npy_file: failed fread");
|
||||
return arr;
|
||||
}
|
||||
|
||||
cnpy::NpyArray load_the_npz_array(FILE* fp, uint32_t compr_bytes, uint32_t uncompr_bytes)
|
||||
{
|
||||
std::vector<unsigned char> buffer_compr(compr_bytes);
|
||||
std::vector<unsigned char> buffer_uncompr(uncompr_bytes);
|
||||
size_t nread = fread(&buffer_compr[0], 1, compr_bytes, fp);
|
||||
if (nread != compr_bytes)
|
||||
throw std::runtime_error("load_the_npy_file: failed fread");
|
||||
|
||||
#if 0
|
||||
int err;
|
||||
z_stream d_stream;
|
||||
|
||||
d_stream.zalloc = Z_NULL;
|
||||
d_stream.zfree = Z_NULL;
|
||||
d_stream.opaque = Z_NULL;
|
||||
d_stream.avail_in = 0;
|
||||
d_stream.next_in = Z_NULL;
|
||||
err = inflateInit2(&d_stream, -MAX_WBITS);
|
||||
|
||||
d_stream.avail_in = compr_bytes;
|
||||
d_stream.next_in = &buffer_compr[0];
|
||||
d_stream.avail_out = uncompr_bytes;
|
||||
d_stream.next_out = &buffer_uncompr[0];
|
||||
|
||||
err = inflate(&d_stream, Z_FINISH);
|
||||
err = inflateEnd(&d_stream);
|
||||
#endif
|
||||
|
||||
std::vector<size_t> shape;
|
||||
size_t word_size;
|
||||
bool fortran_order;
|
||||
std::string typeName;
|
||||
cnpy::parse_npy_header(&buffer_uncompr[0], word_size, shape, fortran_order, typeName);
|
||||
|
||||
cnpy::NpyArray array(shape, word_size, fortran_order, typeName);
|
||||
|
||||
size_t offset = uncompr_bytes - array.num_bytes();
|
||||
memcpy(array.data<unsigned char>(), &buffer_uncompr[0] + offset, array.num_bytes());
|
||||
|
||||
return array;
|
||||
}
|
||||
|
||||
cnpy::npz_t cnpy::npz_load(std::string fname)
|
||||
{
|
||||
FILE* fp = fopen(fname.c_str(), "rb");
|
||||
|
||||
if (!fp) {
|
||||
throw std::runtime_error("npz_load: Error! Unable to open file " + fname + "!");
|
||||
}
|
||||
|
||||
cnpy::npz_t arrays;
|
||||
|
||||
while (1) {
|
||||
std::vector<char> local_header(30);
|
||||
size_t headerres = fread(&local_header[0], sizeof(char), 30, fp);
|
||||
if (headerres != 30)
|
||||
throw std::runtime_error("npz_load: failed fread");
|
||||
|
||||
// if we've reached the global header, stop reading
|
||||
if (local_header[2] != 0x03 || local_header[3] != 0x04)
|
||||
break;
|
||||
|
||||
// read in the variable name
|
||||
uint16_t name_len = *(uint16_t*)&local_header[26];
|
||||
std::string varname(name_len, ' ');
|
||||
size_t vname_res = fread(&varname[0], sizeof(char), name_len, fp);
|
||||
if (vname_res != name_len)
|
||||
throw std::runtime_error("npz_load: failed fread");
|
||||
|
||||
// erase the lagging .npy
|
||||
varname.erase(varname.end() - 4, varname.end());
|
||||
|
||||
// read in the extra field
|
||||
uint16_t extra_field_len = *(uint16_t*)&local_header[28];
|
||||
if (extra_field_len > 0) {
|
||||
std::vector<char> buff(extra_field_len);
|
||||
size_t efield_res = fread(&buff[0], sizeof(char), extra_field_len, fp);
|
||||
if (efield_res != extra_field_len)
|
||||
throw std::runtime_error("npz_load: failed fread");
|
||||
}
|
||||
|
||||
uint16_t compr_method = *reinterpret_cast<uint16_t*>(&local_header[0] + 8);
|
||||
uint32_t compr_bytes = *reinterpret_cast<uint32_t*>(&local_header[0] + 18);
|
||||
uint32_t uncompr_bytes = *reinterpret_cast<uint32_t*>(&local_header[0] + 22);
|
||||
|
||||
if (compr_method == 0) {
|
||||
arrays[varname] = load_the_npy_file(fp);
|
||||
} else {
|
||||
arrays[varname] = load_the_npz_array(fp, compr_bytes, uncompr_bytes);
|
||||
}
|
||||
}
|
||||
|
||||
fclose(fp);
|
||||
return arrays;
|
||||
}
|
||||
|
||||
cnpy::NpyArray cnpy::npz_load(std::string fname, std::string varname)
|
||||
{
|
||||
FILE* fp = fopen(fname.c_str(), "rb");
|
||||
|
||||
if (!fp)
|
||||
throw std::runtime_error("npz_load: Unable to open file " + fname);
|
||||
|
||||
while (1) {
|
||||
std::vector<char> local_header(30);
|
||||
size_t header_res = fread(&local_header[0], sizeof(char), 30, fp);
|
||||
if (header_res != 30)
|
||||
throw std::runtime_error("npz_load: failed fread");
|
||||
|
||||
// if we've reached the global header, stop reading
|
||||
if (local_header[2] != 0x03 || local_header[3] != 0x04)
|
||||
break;
|
||||
|
||||
// read in the variable name
|
||||
uint16_t name_len = *(uint16_t*)&local_header[26];
|
||||
std::string vname(name_len, ' ');
|
||||
size_t vname_res = fread(&vname[0], sizeof(char), name_len, fp);
|
||||
if (vname_res != name_len)
|
||||
throw std::runtime_error("npz_load: failed fread");
|
||||
vname.erase(vname.end() - 4, vname.end()); // erase the lagging .npy
|
||||
|
||||
// read in the extra field
|
||||
uint16_t extra_field_len = *(uint16_t*)&local_header[28];
|
||||
fseek(fp, extra_field_len, SEEK_CUR); // skip past the extra field
|
||||
|
||||
uint16_t compr_method = *reinterpret_cast<uint16_t*>(&local_header[0] + 8);
|
||||
uint32_t compr_bytes = *reinterpret_cast<uint32_t*>(&local_header[0] + 18);
|
||||
uint32_t uncompr_bytes = *reinterpret_cast<uint32_t*>(&local_header[0] + 22);
|
||||
|
||||
if (vname == varname) {
|
||||
NpyArray array = (compr_method == 0) ? load_the_npy_file(fp) : load_the_npz_array(fp, compr_bytes, uncompr_bytes);
|
||||
fclose(fp);
|
||||
return array;
|
||||
} else {
|
||||
// skip past the data
|
||||
uint32_t size = *(uint32_t*)&local_header[22];
|
||||
fseek(fp, size, SEEK_CUR);
|
||||
}
|
||||
}
|
||||
|
||||
fclose(fp);
|
||||
|
||||
// if we get here, we haven't found the variable in the file
|
||||
throw std::runtime_error("npz_load: Variable name " + varname + " not found in " + fname);
|
||||
}
|
||||
|
||||
cnpy::NpyArray cnpy::npy_load(std::string fname)
|
||||
{
|
||||
FILE* fp = fopen(fname.c_str(), "rb");
|
||||
|
||||
if (!fp)
|
||||
throw std::runtime_error("npy_load: Unable to open file " + fname);
|
||||
|
||||
NpyArray arr = load_the_npy_file(fp);
|
||||
|
||||
fclose(fp);
|
||||
return arr;
|
||||
}
|
||||
|
|
@ -0,0 +1,321 @@
|
|||
// Copyright (C) 2011 Carl Rogers
|
||||
// Released under MIT License
|
||||
// license available in LICENSE file, or at http://www.opensource.org/licenses/mit-license.php
|
||||
|
||||
#ifndef LIBCNPY_H_
|
||||
#define LIBCNPY_H_
|
||||
|
||||
#if 0
|
||||
#include <zlib.h>
|
||||
#endif
|
||||
|
||||
#include <stdint.h>
|
||||
|
||||
#include <cassert>
|
||||
#include <cstdio>
|
||||
#include <fstream>
|
||||
#include <iostream>
|
||||
#include <map>
|
||||
#include <memory>
|
||||
#include <numeric>
|
||||
#include <sstream>
|
||||
#include <stdexcept>
|
||||
#include <string>
|
||||
#include <typeinfo>
|
||||
#include <vector>
|
||||
|
||||
namespace cnpy {
|
||||
|
||||
struct NpyArray
|
||||
{
|
||||
NpyArray(const std::vector<size_t>& _shape, size_t _word_size, bool _fortran_order, std::string _typeName)
|
||||
: shape(_shape)
|
||||
, word_size(_word_size)
|
||||
, fortran_order(_fortran_order)
|
||||
, typeName(_typeName)
|
||||
{
|
||||
num_vals = 1;
|
||||
for (size_t i = 0; i < shape.size(); i++)
|
||||
num_vals *= shape[i];
|
||||
data_holder = std::shared_ptr<std::vector<char>>(new std::vector<char>(num_vals * word_size));
|
||||
}
|
||||
|
||||
NpyArray()
|
||||
: shape(0)
|
||||
, word_size(0)
|
||||
, fortran_order(0)
|
||||
, num_vals(0)
|
||||
{}
|
||||
|
||||
template <typename T>
|
||||
T* data()
|
||||
{
|
||||
return reinterpret_cast<T*>(&(*data_holder)[0]);
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
const T* data() const
|
||||
{
|
||||
return reinterpret_cast<T*>(&(*data_holder)[0]);
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
std::vector<T> as_vec() const
|
||||
{
|
||||
const T* p = data<T>();
|
||||
return std::vector<T>(p, p + num_vals);
|
||||
}
|
||||
|
||||
size_t num_bytes() const { return data_holder->size(); }
|
||||
|
||||
std::shared_ptr<std::vector<char>> data_holder;
|
||||
std::vector<size_t> shape;
|
||||
size_t word_size;
|
||||
bool fortran_order;
|
||||
size_t num_vals;
|
||||
std::string typeName;
|
||||
};
|
||||
|
||||
using npz_t = std::map<std::string, NpyArray>;
|
||||
|
||||
char BigEndianTest(int size);
|
||||
char map_type(const std::type_info& t);
|
||||
template <typename T>
|
||||
std::vector<char> create_npy_header(const std::vector<size_t>& shape);
|
||||
void parse_npy_header(FILE* fp, size_t& word_size, std::vector<size_t>& shape, bool& fortran_order,
|
||||
std::string& typeName);
|
||||
void parse_npy_header(unsigned char* buffer, size_t& word_size, std::vector<size_t>& shape, bool& fortran_order,
|
||||
std::string& typeName);
|
||||
void parse_zip_footer(FILE* fp, uint16_t& nrecs, size_t& global_header_size, size_t& global_header_offset);
|
||||
npz_t npz_load(std::string fname);
|
||||
NpyArray npz_load(std::string fname, std::string varname);
|
||||
NpyArray npy_load(std::string fname);
|
||||
|
||||
template <typename T>
|
||||
std::vector<char>& operator+=(std::vector<char>& lhs, const T rhs)
|
||||
{
|
||||
// write in little endian
|
||||
for (size_t byte = 0; byte < sizeof(T); byte++) {
|
||||
char val = *((char*)&rhs + byte);
|
||||
lhs.push_back(val);
|
||||
}
|
||||
return lhs;
|
||||
}
|
||||
|
||||
template <>
|
||||
std::vector<char>& operator+=(std::vector<char>& lhs, const std::string rhs);
|
||||
template <>
|
||||
std::vector<char>& operator+=(std::vector<char>& lhs, const char* rhs);
|
||||
|
||||
template <typename T>
|
||||
int npy_save(std::string fname, const T* data, const std::vector<size_t> shape, std::string mode = "w")
|
||||
{
|
||||
std::ofstream ofs(fname, std::ios::out);
|
||||
if (!ofs.is_open()) {
|
||||
return -1;
|
||||
}
|
||||
ofs.close();
|
||||
FILE* fp = NULL;
|
||||
std::vector<size_t> true_data_shape; // if appending, the shape of existing + new data
|
||||
|
||||
if (mode == "a")
|
||||
fp = fopen(fname.c_str(), "r+b");
|
||||
|
||||
if (fp) {
|
||||
// file exists. we need to append to it. read the header, modify the array size
|
||||
size_t word_size;
|
||||
bool fortran_order;
|
||||
std::string typeName;
|
||||
parse_npy_header(fp, word_size, true_data_shape, fortran_order, typeName);
|
||||
assert(!fortran_order);
|
||||
|
||||
if (word_size != sizeof(T)) {
|
||||
std::cout << "libnpy error: " << fname << " has word size " << word_size << " but npy_save appending data sized "
|
||||
<< sizeof(T) << "\n";
|
||||
assert(word_size == sizeof(T));
|
||||
}
|
||||
if (true_data_shape.size() != shape.size()) {
|
||||
std::cout << "libnpy error: npy_save attempting to append misdimensioned data to " << fname << "\n";
|
||||
assert(true_data_shape.size() != shape.size());
|
||||
}
|
||||
|
||||
for (size_t i = 1; i < shape.size(); i++) {
|
||||
if (shape[i] != true_data_shape[i]) {
|
||||
std::cout << "libnpy error: npy_save attempting to append misshaped data to " << fname << "\n";
|
||||
assert(shape[i] == true_data_shape[i]);
|
||||
}
|
||||
}
|
||||
true_data_shape[0] += shape[0];
|
||||
} else {
|
||||
fp = fopen(fname.c_str(), "wb");
|
||||
true_data_shape = shape;
|
||||
}
|
||||
|
||||
std::vector<char> header = create_npy_header<T>(true_data_shape);
|
||||
size_t nels = std::accumulate(shape.begin(), shape.end(), 1, std::multiplies<size_t>());
|
||||
|
||||
fseek(fp, 0, SEEK_SET);
|
||||
fwrite(&header[0], sizeof(char), header.size(), fp);
|
||||
fseek(fp, 0, SEEK_END);
|
||||
fwrite(data, sizeof(T), nels, fp);
|
||||
fclose(fp);
|
||||
return 0;
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
void npz_save(std::string zipname, std::string fname, const T* data, const std::vector<size_t>& shape,
|
||||
std::string mode = "w")
|
||||
{
|
||||
// first, append a .npy to the fname
|
||||
fname += ".npy";
|
||||
|
||||
// now, on with the show
|
||||
FILE* fp = NULL;
|
||||
uint16_t nrecs = 0;
|
||||
size_t global_header_offset = 0;
|
||||
std::vector<char> global_header;
|
||||
|
||||
if (mode == "a")
|
||||
fp = fopen(zipname.c_str(), "r+b");
|
||||
|
||||
if (fp) {
|
||||
// zip file exists. we need to add a new npy file to it.
|
||||
// first read the footer. this gives us the offset and size of the global header
|
||||
// then read and store the global header.
|
||||
// below, we will write the the new data at the start of the global header then append the global header and footer
|
||||
// below it
|
||||
size_t global_header_size;
|
||||
parse_zip_footer(fp, nrecs, global_header_size, global_header_offset);
|
||||
fseek(fp, global_header_offset, SEEK_SET);
|
||||
global_header.resize(global_header_size);
|
||||
size_t res = fread(&global_header[0], sizeof(char), global_header_size, fp);
|
||||
if (res != global_header_size) {
|
||||
throw std::runtime_error("npz_save: header read error while adding to existing zip");
|
||||
}
|
||||
fseek(fp, global_header_offset, SEEK_SET);
|
||||
} else {
|
||||
fp = fopen(zipname.c_str(), "wb");
|
||||
}
|
||||
|
||||
std::vector<char> npy_header = create_npy_header<T>(shape);
|
||||
|
||||
size_t nels = std::accumulate(shape.begin(), shape.end(), 1, std::multiplies<size_t>());
|
||||
size_t nbytes = nels * sizeof(T) + npy_header.size();
|
||||
|
||||
#if 0
|
||||
// get the CRC of the data to be added
|
||||
uint32_t crc = crc32(0L, (uint8_t*)&npy_header[0], npy_header.size());
|
||||
crc = crc32(crc, (uint8_t*)data, nels * sizeof(T));
|
||||
#else
|
||||
uint32_t crc = 0;
|
||||
#endif
|
||||
|
||||
// build the local header
|
||||
std::vector<char> local_header;
|
||||
local_header += "PK"; // first part of sig
|
||||
local_header += (uint16_t)0x0403; // second part of sig
|
||||
local_header += (uint16_t)20; // min version to extract
|
||||
local_header += (uint16_t)0; // general purpose bit flag
|
||||
local_header += (uint16_t)0; // compression method
|
||||
local_header += (uint16_t)0; // file last mod time
|
||||
local_header += (uint16_t)0; // file last mod date
|
||||
local_header += (uint32_t)crc; // crc
|
||||
local_header += (uint32_t)nbytes; // compressed size
|
||||
local_header += (uint32_t)nbytes; // uncompressed size
|
||||
local_header += (uint16_t)fname.size(); // fname length
|
||||
local_header += (uint16_t)0; // extra field length
|
||||
local_header += fname;
|
||||
|
||||
// build global header
|
||||
global_header += "PK"; // first part of sig
|
||||
global_header += (uint16_t)0x0201; // second part of sig
|
||||
global_header += (uint16_t)20; // version made by
|
||||
global_header.insert(global_header.end(), local_header.begin() + 4, local_header.begin() + 30);
|
||||
global_header += (uint16_t)0; // file comment length
|
||||
global_header += (uint16_t)0; // disk number where file starts
|
||||
global_header += (uint16_t)0; // internal file attributes
|
||||
global_header += (uint32_t)0; // external file attributes
|
||||
global_header += (uint32_t)
|
||||
global_header_offset; // relative offset of local file header, since it begins where the global header used to begin
|
||||
global_header += fname;
|
||||
|
||||
// build footer
|
||||
std::vector<char> footer;
|
||||
footer += "PK"; // first part of sig
|
||||
footer += (uint16_t)0x0605; // second part of sig
|
||||
footer += (uint16_t)0; // number of this disk
|
||||
footer += (uint16_t)0; // disk where footer starts
|
||||
footer += (uint16_t)(nrecs + 1); // number of records on this disk
|
||||
footer += (uint16_t)(nrecs + 1); // total number of records
|
||||
footer += (uint32_t)global_header.size(); // nbytes of global headers
|
||||
footer +=
|
||||
(uint32_t)(global_header_offset + nbytes + local_header.size()); // offset of start of global headers, since global
|
||||
// header now starts after newly written array
|
||||
footer += (uint16_t)0; // zip file comment length
|
||||
|
||||
// write everything
|
||||
fwrite(&local_header[0], sizeof(char), local_header.size(), fp);
|
||||
fwrite(&npy_header[0], sizeof(char), npy_header.size(), fp);
|
||||
fwrite(data, sizeof(T), nels, fp);
|
||||
fwrite(&global_header[0], sizeof(char), global_header.size(), fp);
|
||||
fwrite(&footer[0], sizeof(char), footer.size(), fp);
|
||||
fclose(fp);
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
void npy_save(std::string fname, const std::vector<T> data, std::string mode = "w")
|
||||
{
|
||||
std::vector<size_t> shape;
|
||||
shape.push_back(data.size());
|
||||
npy_save(fname, &data[0], shape, mode);
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
void npz_save(std::string zipname, std::string fname, const std::vector<T> data, std::string mode = "w")
|
||||
{
|
||||
std::vector<size_t> shape;
|
||||
shape.push_back(data.size());
|
||||
npz_save(zipname, fname, &data[0], shape, mode);
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
std::vector<char> create_npy_header(const std::vector<size_t>& shape)
|
||||
{
|
||||
const char* tpye_name = typeid(T).name();
|
||||
std::vector<char> dict;
|
||||
dict += "{'descr': '";
|
||||
dict += BigEndianTest(sizeof(T));
|
||||
if (std::string(tpye_name) == "N4rknn7float16E") {
|
||||
dict += "f";
|
||||
} else {
|
||||
dict += map_type(typeid(T));
|
||||
}
|
||||
dict += std::to_string(sizeof(T));
|
||||
dict += "', 'fortran_order': False, 'shape': (";
|
||||
dict += std::to_string(shape[0]);
|
||||
for (size_t i = 1; i < shape.size(); i++) {
|
||||
dict += ", ";
|
||||
dict += std::to_string(shape[i]);
|
||||
}
|
||||
if (shape.size() == 1)
|
||||
dict += ",";
|
||||
dict += "), }";
|
||||
// pad with spaces so that preamble+dict is modulo 16 bytes. preamble is 10 bytes. dict needs to end with \n
|
||||
int remainder = 16 - (10 + dict.size()) % 16;
|
||||
dict.insert(dict.end(), remainder, ' ');
|
||||
dict.back() = '\n';
|
||||
|
||||
std::vector<char> header;
|
||||
header += (char)0x93;
|
||||
header += "NUMPY";
|
||||
header += (char)0x01; // major version of numpy format
|
||||
header += (char)0x00; // minor version of numpy format
|
||||
header += (uint16_t)dict.size();
|
||||
header.insert(header.end(), dict.begin(), dict.end());
|
||||
|
||||
return header;
|
||||
}
|
||||
|
||||
} // namespace cnpy
|
||||
|
||||
#endif
|
||||
BIN
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/Linux/aarch64/librockchip_mpp.so
vendored
Normal file
BIN
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/Linux/aarch64/librockchip_mpp.so
vendored
Normal file
Binary file not shown.
BIN
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/Linux/aarch64/librockchip_mpp.so.0
vendored
Normal file
BIN
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/Linux/aarch64/librockchip_mpp.so.0
vendored
Normal file
Binary file not shown.
BIN
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/Linux/aarch64/librockchip_mpp.so.1
vendored
Normal file
BIN
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/Linux/aarch64/librockchip_mpp.so.1
vendored
Normal file
Binary file not shown.
327
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_buffer.h
vendored
Normal file
327
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_buffer.h
vendored
Normal file
|
|
@ -0,0 +1,327 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_BUFFER_H__
|
||||
#define __MPP_BUFFER_H__
|
||||
|
||||
#include "rk_type.h"
|
||||
#include "mpp_err.h"
|
||||
|
||||
/*
|
||||
* MppBuffer module has several functions:
|
||||
*
|
||||
* 1. buffer get / put / reference management / external commit / get info.
|
||||
* this part is the basic user interface for MppBuffer.
|
||||
*
|
||||
* function:
|
||||
*
|
||||
* mpp_buffer_get
|
||||
* mpp_buffer_put
|
||||
* mpp_buffer_inc_ref
|
||||
* mpp_buffer_commit
|
||||
* mpp_buffer_info_get
|
||||
*
|
||||
* 2. user buffer working flow control abstraction.
|
||||
* buffer should attach to certain group, and buffer mode control the buffer usage flow.
|
||||
* this part is also a part of user interface.
|
||||
*
|
||||
* function:
|
||||
*
|
||||
* mpp_buffer_group_get
|
||||
* mpp_buffer_group_normal_get
|
||||
* mpp_buffer_group_limit_get
|
||||
* mpp_buffer_group_put
|
||||
* mpp_buffer_group_limit_config
|
||||
*
|
||||
* 3. buffer allocator management
|
||||
* this part is for allocator on different os, it does not have user interface
|
||||
* it will support normal buffer, Android ion buffer, Linux v4l2 vb2 buffer
|
||||
* user can only use MppBufferType to choose.
|
||||
*
|
||||
*/
|
||||
|
||||
/*
|
||||
* mpp buffer group support two work flow mode:
|
||||
*
|
||||
* normal flow: all buffer are generated by MPP
|
||||
* under this mode, buffer pool is maintained internally
|
||||
*
|
||||
* typical call flow:
|
||||
*
|
||||
* mpp_buffer_group_get() return A
|
||||
* mpp_buffer_get(A) return a ref +1 -> used
|
||||
* mpp_buffer_inc_ref(a) ref +1
|
||||
* mpp_buffer_put(a) ref -1
|
||||
* mpp_buffer_put(a) ref -1 -> unused
|
||||
* mpp_buffer_group_put(A)
|
||||
*
|
||||
* commit flow: all buffer are commited out of MPP
|
||||
* under this mode, buffers is commit by external api.
|
||||
* normally MPP only use it but not generate it.
|
||||
*
|
||||
* typical call flow:
|
||||
*
|
||||
* ==== external allocator ====
|
||||
* mpp_buffer_group_get() return A
|
||||
* mpp_buffer_commit(A, x)
|
||||
* mpp_buffer_commit(A, y)
|
||||
*
|
||||
* ======= internal user ======
|
||||
* mpp_buffer_get(A) return a
|
||||
* mpp_buffer_get(A) return b
|
||||
* mpp_buffer_put(a)
|
||||
* mpp_buffer_put(b)
|
||||
*
|
||||
* ==== external allocator ====
|
||||
* mpp_buffer_group_put(A)
|
||||
*
|
||||
* NOTE: commit interface required group handle to record group information
|
||||
*/
|
||||
|
||||
/*
|
||||
* mpp buffer group has two buffer limit mode: normal and limit
|
||||
*
|
||||
* normal mode: allows any buffer size and always general new buffer is no unused buffer
|
||||
* is available.
|
||||
* This mode normally use with normal flow and is used for table / stream buffer
|
||||
*
|
||||
* limit mode : restrict the buffer's size and count in the buffer group. if try to calloc
|
||||
* buffer with different size or extra count it will fail.
|
||||
* This mode normally use with commit flow and is used for frame buffer
|
||||
*/
|
||||
|
||||
/*
|
||||
* NOTE: normal mode is recommanded to work with normal flow, working with limit mode is not.
|
||||
* limit mode is recommanded to work with commit flow, working with normal mode is not.
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_BUFFER_INTERNAL,
|
||||
MPP_BUFFER_EXTERNAL,
|
||||
MPP_BUFFER_MODE_BUTT,
|
||||
} MppBufferMode;
|
||||
|
||||
/*
|
||||
* the mpp buffer has serval types:
|
||||
*
|
||||
* normal : normal malloc buffer for unit test or hardware simulation
|
||||
* ion : use ion device under Android/Linux, MppBuffer will encapsulte ion file handle
|
||||
* ext_dma : the DMABUF(DMA buffers) come from the application
|
||||
* drm : use the drm device interface for memory management
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_BUFFER_TYPE_NORMAL,
|
||||
MPP_BUFFER_TYPE_ION,
|
||||
MPP_BUFFER_TYPE_EXT_DMA,
|
||||
MPP_BUFFER_TYPE_DRM,
|
||||
MPP_BUFFER_TYPE_DMA_HEAP,
|
||||
MPP_BUFFER_TYPE_BUTT,
|
||||
} MppBufferType;
|
||||
|
||||
#define MPP_BUFFER_TYPE_MASK 0x0000FFFF
|
||||
|
||||
/*
|
||||
* MPP_BUFFER_FLAGS cooperate with MppBufferType
|
||||
* 16 high bits of MppBufferType are used in flags
|
||||
*
|
||||
* eg:
|
||||
* DRM CMA buffer : MPP_BUFFER_TYPE_DRM | MPP_BUFFER_FLAGS_CONTIG
|
||||
* = 0x00010003
|
||||
* DRM SECURE buffer: MPP_BUFFER_TYPE_DRM | MPP_BUFFER_FLAGS_SECURE
|
||||
* = 0x00080003
|
||||
*
|
||||
* The dma buffer source can also be set by format: flags | type.
|
||||
* dma buffer source flags:
|
||||
* MPP_BUFFER_FLAGS_CONTIG means cma
|
||||
* MPP_BUFFER_FLAGS_CACHABLE means cachable
|
||||
* MPP_BUFFER_FLAGS_DMA32 means dma32
|
||||
*
|
||||
* flags originate from drm_rockchip_gem_mem_type
|
||||
*/
|
||||
#define MPP_BUFFER_FLAGS_MASK 0x003f0000 //ROCKCHIP_BO_MASK << 16
|
||||
#define MPP_BUFFER_FLAGS_CONTIG 0x00010000 //ROCKCHIP_BO_CONTIG << 16
|
||||
#define MPP_BUFFER_FLAGS_CACHABLE 0x00020000 //ROCKCHIP_BO_CACHABLE << 16
|
||||
#define MPP_BUFFER_FLAGS_WC 0x00040000 //ROCKCHIP_BO_WC << 16
|
||||
#define MPP_BUFFER_FLAGS_SECURE 0x00080000 //ROCKCHIP_BO_SECURE << 16
|
||||
#define MPP_BUFFER_FLAGS_ALLOC_KMAP 0x00100000 //ROCKCHIP_BO_ALLOC_KMAP << 16
|
||||
#define MPP_BUFFER_FLAGS_DMA32 0x00200000 //ROCKCHIP_BO_DMA32 << 16
|
||||
|
||||
/*
|
||||
* MppBufferInfo variable's meaning is different in different MppBufferType
|
||||
*
|
||||
* Common
|
||||
* index - the buffer index used to track buffer in buffer pool
|
||||
* size - the buffer size
|
||||
*
|
||||
* MPP_BUFFER_TYPE_NORMAL
|
||||
*
|
||||
* ptr - virtual address of normal malloced buffer
|
||||
* fd - unused and set to -1, the allocator would return its
|
||||
* internal buffer counter number
|
||||
*
|
||||
* MPP_BUFFER_TYPE_ION
|
||||
*
|
||||
* ptr - virtual address of ion buffer in user space
|
||||
* hnd - ion handle in user space
|
||||
* fd - ion buffer file handle for map / unmap
|
||||
*
|
||||
*/
|
||||
typedef struct MppBufferInfo_t {
|
||||
MppBufferType type;
|
||||
size_t size;
|
||||
void *ptr;
|
||||
void *hnd;
|
||||
int fd;
|
||||
int index;
|
||||
} MppBufferInfo;
|
||||
|
||||
#define BUFFER_GROUP_SIZE_DEFAULT (SZ_1M*80)
|
||||
|
||||
/*
|
||||
* mpp_buffer_import_with_tag(MppBufferGroup group, MppBufferInfo *info, MppBuffer *buffer)
|
||||
*
|
||||
* 1. group - specified the MppBuffer to be attached to.
|
||||
* group can be NULL then this buffer will attached to default legecy group
|
||||
* Default to NULL on mpp_buffer_import case
|
||||
*
|
||||
* 2. info - input information for the output MppBuffer
|
||||
* info can NOT be NULL. It must contain at least one of ptr/fd.
|
||||
*
|
||||
* 3. buffer - generated MppBuffer from MppBufferInfo.
|
||||
* buffer can be NULL then the buffer is commit to group with unused status.
|
||||
* Otherwise generated buffer will be directly got and ref_count increased.
|
||||
* Default to NULL on mpp_buffer_commit case
|
||||
*
|
||||
* mpp_buffer_commit usage:
|
||||
*
|
||||
* Add a external buffer info to group. This buffer will be on unused status.
|
||||
* Typical usage is on Android. MediaPlayer gralloc Graphic buffer then commit these buffer
|
||||
* to decoder's buffer group. Then decoder will recycle these buffer and return buffer reference
|
||||
* to MediaPlayer for display.
|
||||
*
|
||||
* mpp_buffer_import usage:
|
||||
*
|
||||
* Transfer a external buffer info to MppBuffer but it is not expected to attached to certain
|
||||
* buffer group. So the group is set to NULL. Then this buffer can be used for MppFrame/MppPacket.
|
||||
* Typical usage is for image processing. Image processing normally will be a oneshot operation
|
||||
* It does not need complicated group management. But in other hand mpp still need to know the
|
||||
* imported buffer is leak or not and trace its usage inside mpp process. So we attach this kind
|
||||
* of buffer to default misc buffer group for management.
|
||||
*/
|
||||
#define mpp_buffer_commit(group, info) \
|
||||
mpp_buffer_import_with_tag(group, info, NULL, MODULE_TAG, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_import(buffer, info) \
|
||||
mpp_buffer_import_with_tag(NULL, info, buffer, MODULE_TAG, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_get(group, buffer, size) \
|
||||
mpp_buffer_get_with_tag(group, buffer, size, MODULE_TAG, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_put(buffer) \
|
||||
mpp_buffer_put_with_caller(buffer, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_inc_ref(buffer) \
|
||||
mpp_buffer_inc_ref_with_caller(buffer, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_info_get(buffer, info) \
|
||||
mpp_buffer_info_get_with_caller(buffer, info, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_read(buffer, offset, data, size) \
|
||||
mpp_buffer_read_with_caller(buffer, offset, data, size, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_write(buffer, offset, data, size) \
|
||||
mpp_buffer_write_with_caller(buffer, offset, data, size, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_get_ptr(buffer) \
|
||||
mpp_buffer_get_ptr_with_caller(buffer, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_get_fd(buffer) \
|
||||
mpp_buffer_get_fd_with_caller(buffer, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_get_size(buffer) \
|
||||
mpp_buffer_get_size_with_caller(buffer, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_get_index(buffer) \
|
||||
mpp_buffer_get_index_with_caller(buffer, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_set_index(buffer, index) \
|
||||
mpp_buffer_set_index_with_caller(buffer, index, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_get_offset(buffer) \
|
||||
mpp_buffer_get_offset_with_caller(buffer, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_set_offset(buffer, offset) \
|
||||
mpp_buffer_set_offset_with_caller(buffer, offset, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_group_get_internal(group, type, ...) \
|
||||
mpp_buffer_group_get(group, type, MPP_BUFFER_INTERNAL, MODULE_TAG, __FUNCTION__)
|
||||
|
||||
#define mpp_buffer_group_get_external(group, type, ...) \
|
||||
mpp_buffer_group_get(group, type, MPP_BUFFER_EXTERNAL, MODULE_TAG, __FUNCTION__)
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
/*
|
||||
* MppBuffer interface
|
||||
* these interface will change value of group and buffer so before calling functions
|
||||
* parameter need to be checked.
|
||||
*
|
||||
* IMPORTANT:
|
||||
* mpp_buffer_import_with_tag - compounded interface for commit and import
|
||||
*
|
||||
*/
|
||||
MPP_RET mpp_buffer_import_with_tag(MppBufferGroup group, MppBufferInfo *info, MppBuffer *buffer,
|
||||
const char *tag, const char *caller);
|
||||
MPP_RET mpp_buffer_get_with_tag(MppBufferGroup group, MppBuffer *buffer, size_t size,
|
||||
const char *tag, const char *caller);
|
||||
MPP_RET mpp_buffer_put_with_caller(MppBuffer buffer, const char *caller);
|
||||
MPP_RET mpp_buffer_inc_ref_with_caller(MppBuffer buffer, const char *caller);
|
||||
|
||||
MPP_RET mpp_buffer_info_get_with_caller(MppBuffer buffer, MppBufferInfo *info, const char *caller);
|
||||
MPP_RET mpp_buffer_read_with_caller(MppBuffer buffer, size_t offset, void *data, size_t size, const char *caller);
|
||||
MPP_RET mpp_buffer_write_with_caller(MppBuffer buffer, size_t offset, void *data, size_t size, const char *caller);
|
||||
void *mpp_buffer_get_ptr_with_caller(MppBuffer buffer, const char *caller);
|
||||
int mpp_buffer_get_fd_with_caller(MppBuffer buffer, const char *caller);
|
||||
size_t mpp_buffer_get_size_with_caller(MppBuffer buffer, const char *caller);
|
||||
int mpp_buffer_get_index_with_caller(MppBuffer buffer, const char *caller);
|
||||
MPP_RET mpp_buffer_set_index_with_caller(MppBuffer buffer, int index, const char *caller);
|
||||
size_t mpp_buffer_get_offset_with_caller(MppBuffer buffer, const char *caller);
|
||||
MPP_RET mpp_buffer_set_offset_with_caller(MppBuffer buffer, size_t offset, const char *caller);
|
||||
|
||||
MPP_RET mpp_buffer_group_get(MppBufferGroup *group, MppBufferType type, MppBufferMode mode,
|
||||
const char *tag, const char *caller);
|
||||
MPP_RET mpp_buffer_group_put(MppBufferGroup group);
|
||||
MPP_RET mpp_buffer_group_clear(MppBufferGroup group);
|
||||
RK_S32 mpp_buffer_group_unused(MppBufferGroup group);
|
||||
size_t mpp_buffer_group_usage(MppBufferGroup group);
|
||||
MppBufferMode mpp_buffer_group_mode(MppBufferGroup group);
|
||||
MppBufferType mpp_buffer_group_type(MppBufferGroup group);
|
||||
|
||||
/*
|
||||
* size : 0 - no limit, other - max buffer size
|
||||
* count : 0 - no limit, other - max buffer count
|
||||
*/
|
||||
MPP_RET mpp_buffer_group_limit_config(MppBufferGroup group, size_t size, RK_S32 count);
|
||||
|
||||
RK_U32 mpp_buffer_total_now();
|
||||
RK_U32 mpp_buffer_total_max();
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__MPP_BUFFER_H__*/
|
||||
62
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_compat.h
vendored
Normal file
62
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_compat.h
vendored
Normal file
|
|
@ -0,0 +1,62 @@
|
|||
/*
|
||||
* Copyright 2021 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_COMPAT_H__
|
||||
#define __MPP_COMPAT_H__
|
||||
|
||||
#include "rk_type.h"
|
||||
#include "mpp_err.h"
|
||||
|
||||
typedef enum MppCompatId_e {
|
||||
MPP_COMPAT_INC_FBC_BUF_SIZE,
|
||||
MPP_COMPAT_ENC_ASYNC_INPUT,
|
||||
MPP_COMPAT_DEC_FBC_HDR_256_ODD,
|
||||
MPP_COMPAT_BUTT,
|
||||
} MppCompatId;
|
||||
|
||||
typedef enum MppCompatType_e {
|
||||
MPP_COMPAT_BOOL,
|
||||
MPP_COMPAT_S32,
|
||||
MPP_COMPAT_TYPE_BUTT,
|
||||
} MppCompatType;
|
||||
|
||||
typedef struct MppCompat_t MppCompat;
|
||||
|
||||
/* external user can only update value_ext to notify mpp to change its behavior */
|
||||
struct MppCompat_t {
|
||||
const MppCompatId feature_id;
|
||||
const MppCompatType feature_type;
|
||||
const RK_S32 value_mpp;
|
||||
RK_S32 value_usr;
|
||||
const char *name;
|
||||
MppCompat * const next;
|
||||
};
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
MppCompat *mpp_compat_query(void);
|
||||
MppCompat *mpp_compat_query_by_id(MppCompatId id);
|
||||
MPP_RET mpp_compat_update(MppCompat *compat, RK_S32 value);
|
||||
|
||||
void mpp_compat_show(void);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__MPP_COMPAT_H__*/
|
||||
54
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_err.h
vendored
Normal file
54
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_err.h
vendored
Normal file
|
|
@ -0,0 +1,54 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_ERR_H__
|
||||
#define __MPP_ERR_H__
|
||||
|
||||
#define RK_OK 0
|
||||
#define RK_SUCCESS 0
|
||||
|
||||
typedef enum {
|
||||
MPP_SUCCESS = RK_SUCCESS,
|
||||
MPP_OK = RK_OK,
|
||||
|
||||
MPP_NOK = -1,
|
||||
MPP_ERR_UNKNOW = -2,
|
||||
MPP_ERR_NULL_PTR = -3,
|
||||
MPP_ERR_MALLOC = -4,
|
||||
MPP_ERR_OPEN_FILE = -5,
|
||||
MPP_ERR_VALUE = -6,
|
||||
MPP_ERR_READ_BIT = -7,
|
||||
MPP_ERR_TIMEOUT = -8,
|
||||
MPP_ERR_PERM = -9,
|
||||
|
||||
MPP_ERR_BASE = -1000,
|
||||
|
||||
/* The error in stream processing */
|
||||
MPP_ERR_LIST_STREAM = MPP_ERR_BASE - 1,
|
||||
MPP_ERR_INIT = MPP_ERR_BASE - 2,
|
||||
MPP_ERR_VPU_CODEC_INIT = MPP_ERR_BASE - 3,
|
||||
MPP_ERR_STREAM = MPP_ERR_BASE - 4,
|
||||
MPP_ERR_FATAL_THREAD = MPP_ERR_BASE - 5,
|
||||
MPP_ERR_NOMEM = MPP_ERR_BASE - 6,
|
||||
MPP_ERR_PROTOL = MPP_ERR_BASE - 7,
|
||||
MPP_FAIL_SPLIT_FRAME = MPP_ERR_BASE - 8,
|
||||
MPP_ERR_VPUHW = MPP_ERR_BASE - 9,
|
||||
MPP_EOS_STREAM_REACHED = MPP_ERR_BASE - 11,
|
||||
MPP_ERR_BUFFER_FULL = MPP_ERR_BASE - 12,
|
||||
MPP_ERR_DISPLAY_FULL = MPP_ERR_BASE - 13,
|
||||
} MPP_RET;
|
||||
|
||||
#endif /*__MPP_ERR_H__*/
|
||||
432
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_frame.h
vendored
Normal file
432
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_frame.h
vendored
Normal file
|
|
@ -0,0 +1,432 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_FRAME_H__
|
||||
#define __MPP_FRAME_H__
|
||||
|
||||
#include "mpp_buffer.h"
|
||||
#include "mpp_meta.h"
|
||||
|
||||
/*
|
||||
* bit definition for mode flag in MppFrame
|
||||
*/
|
||||
/* progressive frame */
|
||||
#define MPP_FRAME_FLAG_FRAME (0x00000000)
|
||||
/* top field only */
|
||||
#define MPP_FRAME_FLAG_TOP_FIELD (0x00000001)
|
||||
/* bottom field only */
|
||||
#define MPP_FRAME_FLAG_BOT_FIELD (0x00000002)
|
||||
/* paired field */
|
||||
#define MPP_FRAME_FLAG_PAIRED_FIELD (MPP_FRAME_FLAG_TOP_FIELD|MPP_FRAME_FLAG_BOT_FIELD)
|
||||
/* paired field with field order of top first */
|
||||
#define MPP_FRAME_FLAG_TOP_FIRST (0x00000004)
|
||||
/* paired field with field order of bottom first */
|
||||
#define MPP_FRAME_FLAG_BOT_FIRST (0x00000008)
|
||||
/* paired field with unknown field order (MBAFF) */
|
||||
#define MPP_FRAME_FLAG_DEINTERLACED (MPP_FRAME_FLAG_TOP_FIRST|MPP_FRAME_FLAG_BOT_FIRST)
|
||||
#define MPP_FRAME_FLAG_FIELD_ORDER_MASK (0x0000000C)
|
||||
// for multiview stream
|
||||
#define MPP_FRAME_FLAG_VIEW_ID_MASK (0x000000f0)
|
||||
|
||||
#define MPP_FRAME_FLAG_IEP_DEI_MASK (0x00000f00)
|
||||
#define MPP_FRAME_FLAG_IEP_DEI_I2O1 (0x00000100)
|
||||
#define MPP_FRAME_FLAG_IEP_DEI_I4O2 (0x00000200)
|
||||
#define MPP_FRAME_FLAG_IEP_DEI_I4O1 (0x00000300)
|
||||
|
||||
/*
|
||||
* MPEG vs JPEG YUV range.
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_FRAME_RANGE_UNSPECIFIED = 0,
|
||||
MPP_FRAME_RANGE_MPEG = 1, ///< the normal 219*2^(n-8) "MPEG" YUV ranges
|
||||
MPP_FRAME_RANGE_JPEG = 2, ///< the normal 2^n-1 "JPEG" YUV ranges
|
||||
MPP_FRAME_RANGE_NB, ///< Not part of ABI
|
||||
} MppFrameColorRange;
|
||||
|
||||
typedef enum {
|
||||
MPP_FRAME_VIDEO_FMT_COMPONEMT = 0,
|
||||
MPP_FRAME_VIDEO_FMT_PAL = 1,
|
||||
MPP_FRAME_VIDEO_FMT_NTSC = 2,
|
||||
MPP_FRAME_VIDEO_FMT_SECAM = 3,
|
||||
MPP_FRAME_VIDEO_FMT_MAC = 4,
|
||||
MPP_FRAME_VIDEO_FMT_UNSPECIFIED = 5,
|
||||
MPP_FRAME_VIDEO_FMT_RESERVED0 = 6,
|
||||
MPP_FRAME_VIDEO_FMT_RESERVED1 = 7,
|
||||
} MppFrameVideoFormat;
|
||||
|
||||
/*
|
||||
* Chromaticity coordinates of the source primaries.
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_FRAME_PRI_RESERVED0 = 0,
|
||||
MPP_FRAME_PRI_BT709 = 1, ///< also ITU-R BT1361 / IEC 61966-2-4 / SMPTE RP177 Annex B
|
||||
MPP_FRAME_PRI_UNSPECIFIED = 2,
|
||||
MPP_FRAME_PRI_RESERVED = 3,
|
||||
MPP_FRAME_PRI_BT470M = 4, ///< also FCC Title 47 Code of Federal Regulations 73.682 (a)(20)
|
||||
|
||||
MPP_FRAME_PRI_BT470BG = 5, ///< also ITU-R BT601-6 625 / ITU-R BT1358 625 / ITU-R BT1700 625 PAL & SECAM
|
||||
MPP_FRAME_PRI_SMPTE170M = 6, ///< also ITU-R BT601-6 525 / ITU-R BT1358 525 / ITU-R BT1700 NTSC/SMPTE ST 170 (2004)
|
||||
MPP_FRAME_PRI_SMPTE240M = 7, ///< functionally identical to above/SMPTE ST 240
|
||||
MPP_FRAME_PRI_FILM = 8, ///< colour filters using Illuminant C
|
||||
MPP_FRAME_PRI_BT2020 = 9, ///< ITU-R BT2020 / ITU-R BT.2100-2
|
||||
MPP_FRAME_PRI_SMPTEST428_1 = 10, ///< SMPTE ST 428-1 (CIE 1931 XYZ)
|
||||
MPP_FRAME_PRI_SMPTE431 = 11, ///< SMPTE ST 431-2 (2011) / DCI P3
|
||||
MPP_FRAME_PRI_SMPTE432 = 12, ///< SMPTE ST 432-1 (2010) / P3 D65 / Display P3
|
||||
MPP_FRAME_PRI_JEDEC_P22 = 22, ///< JEDEC P22 phosphors
|
||||
MPP_FRAME_PRI_NB, ///< Not part of ABI
|
||||
} MppFrameColorPrimaries;
|
||||
|
||||
/*
|
||||
* Color Transfer Characteristic.
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_FRAME_TRC_RESERVED0 = 0,
|
||||
MPP_FRAME_TRC_BT709 = 1, ///< also ITU-R BT1361
|
||||
MPP_FRAME_TRC_UNSPECIFIED = 2,
|
||||
MPP_FRAME_TRC_RESERVED = 3,
|
||||
MPP_FRAME_TRC_GAMMA22 = 4, ///< also ITU-R BT470M / ITU-R BT1700 625 PAL & SECAM
|
||||
MPP_FRAME_TRC_GAMMA28 = 5, ///< also ITU-R BT470BG
|
||||
MPP_FRAME_TRC_SMPTE170M = 6, ///< also ITU-R BT601-6 525 or 625 / ITU-R BT1358 525 or 625 / ITU-R BT1700 NTSC
|
||||
MPP_FRAME_TRC_SMPTE240M = 7,
|
||||
MPP_FRAME_TRC_LINEAR = 8, ///< "Linear transfer characteristics"
|
||||
MPP_FRAME_TRC_LOG = 9, ///< "Logarithmic transfer characteristic (100:1 range)"
|
||||
MPP_FRAME_TRC_LOG_SQRT = 10, ///< "Logarithmic transfer characteristic (100 * Sqrt(10) : 1 range)"
|
||||
MPP_FRAME_TRC_IEC61966_2_4 = 11, ///< IEC 61966-2-4
|
||||
MPP_FRAME_TRC_BT1361_ECG = 12, ///< ITU-R BT1361 Extended Colour Gamut
|
||||
MPP_FRAME_TRC_IEC61966_2_1 = 13, ///< IEC 61966-2-1 (sRGB or sYCC)
|
||||
MPP_FRAME_TRC_BT2020_10 = 14, ///< ITU-R BT2020 for 10 bit system
|
||||
MPP_FRAME_TRC_BT2020_12 = 15, ///< ITU-R BT2020 for 12 bit system
|
||||
MPP_FRAME_TRC_SMPTEST2084 = 16, ///< SMPTE ST 2084 for 10-, 12-, 14- and 16-bit systems
|
||||
MPP_FRAME_TRC_SMPTEST428_1 = 17, ///< SMPTE ST 428-1
|
||||
MPP_FRAME_TRC_ARIB_STD_B67 = 18, ///< ARIB STD-B67, known as "Hybrid log-gamma"
|
||||
MPP_FRAME_TRC_NB, ///< Not part of ABI
|
||||
} MppFrameColorTransferCharacteristic;
|
||||
|
||||
/*
|
||||
* YUV colorspace type.
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_FRAME_SPC_RGB = 0, ///< order of coefficients is actually GBR, also IEC 61966-2-1 (sRGB)
|
||||
MPP_FRAME_SPC_BT709 = 1, ///< also ITU-R BT1361 / IEC 61966-2-4 xvYCC709 / SMPTE RP177 Annex B
|
||||
MPP_FRAME_SPC_UNSPECIFIED = 2,
|
||||
MPP_FRAME_SPC_RESERVED = 3,
|
||||
MPP_FRAME_SPC_FCC = 4, ///< FCC Title 47 Code of Federal Regulations 73.682 (a)(20)
|
||||
MPP_FRAME_SPC_BT470BG = 5, ///< also ITU-R BT601-6 625 / ITU-R BT1358 625 / ITU-R BT1700 625 PAL & SECAM / IEC 61966-2-4 xvYCC601
|
||||
MPP_FRAME_SPC_SMPTE170M = 6, ///< also ITU-R BT601-6 525 / ITU-R BT1358 525 / ITU-R BT1700 NTSC / functionally identical to above
|
||||
MPP_FRAME_SPC_SMPTE240M = 7,
|
||||
MPP_FRAME_SPC_YCOCG = 8, ///< Used by Dirac / VC-2 and H.264 FRext, see ITU-T SG16
|
||||
MPP_FRAME_SPC_BT2020_NCL = 9, ///< ITU-R BT2020 non-constant luminance system
|
||||
MPP_FRAME_SPC_BT2020_CL = 10, ///< ITU-R BT2020 constant luminance system
|
||||
MPP_FRAME_SPC_SMPTE2085 = 11, ///< SMPTE 2085, Y'D'zD'x
|
||||
MPP_FRAME_SPC_CHROMA_DERIVED_NCL = 12, ///< Chromaticity-derived non-constant luminance system
|
||||
MPP_FRAME_SPC_CHROMA_DERIVED_CL = 13, ///< Chromaticity-derived constant luminance system
|
||||
MPP_FRAME_SPC_ICTCP = 14, ///< ITU-R BT.2100-0, ICtCp
|
||||
MPP_FRAME_SPC_NB, ///< Not part of ABI
|
||||
} MppFrameColorSpace;
|
||||
|
||||
/*
|
||||
* Location of chroma samples.
|
||||
*
|
||||
* Illustration showing the location of the first (top left) chroma sample of the
|
||||
* image, the left shows only luma, the right
|
||||
* shows the location of the chroma sample, the 2 could be imagined to overlay
|
||||
* each other but are drawn separately due to limitations of ASCII
|
||||
*
|
||||
* 1st 2nd 1st 2nd horizontal luma sample positions
|
||||
* v v v v
|
||||
* ______ ______
|
||||
*1st luma line > |X X ... |3 4 X ... X are luma samples,
|
||||
* | |1 2 1-6 are possible chroma positions
|
||||
*2nd luma line > |X X ... |5 6 X ... 0 is undefined/unknown position
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_CHROMA_LOC_UNSPECIFIED = 0,
|
||||
MPP_CHROMA_LOC_LEFT = 1, ///< mpeg2/4 4:2:0, h264 default for 4:2:0
|
||||
MPP_CHROMA_LOC_CENTER = 2, ///< mpeg1 4:2:0, jpeg 4:2:0, h263 4:2:0
|
||||
MPP_CHROMA_LOC_TOPLEFT = 3, ///< ITU-R 601, SMPTE 274M 296M S314M(DV 4:1:1), mpeg2 4:2:2
|
||||
MPP_CHROMA_LOC_TOP = 4,
|
||||
MPP_CHROMA_LOC_BOTTOMLEFT = 5,
|
||||
MPP_CHROMA_LOC_BOTTOM = 6,
|
||||
MPP_CHROMA_LOC_NB, ///< Not part of ABI
|
||||
} MppFrameChromaLocation;
|
||||
|
||||
#define MPP_FRAME_FMT_MASK (0x000fffff)
|
||||
|
||||
#define MPP_FRAME_FMT_COLOR_MASK (0x000f0000)
|
||||
#define MPP_FRAME_FMT_YUV (0x00000000)
|
||||
#define MPP_FRAME_FMT_RGB (0x00010000)
|
||||
|
||||
#define MPP_FRAME_FBC_MASK (0x00f00000)
|
||||
#define MPP_FRAME_FBC_NONE (0x00000000)
|
||||
|
||||
#define MPP_FRAME_HDR_MASK (0x0f000000)
|
||||
#define MPP_FRAME_HDR_NONE (0x00000000)
|
||||
|
||||
#define MPP_FRAME_HDR (0x01000000)
|
||||
|
||||
/*
|
||||
* AFBC_V1 is for ISP output.
|
||||
* It has default payload offset to be calculated * from width and height:
|
||||
* Payload offset = MPP_ALIGN(MPP_ALIGN(width, 16) * MPP_ALIGN(height, 16) / 16, SZ_4K)
|
||||
*/
|
||||
#define MPP_FRAME_FBC_AFBC_V1 (0x00100000)
|
||||
/*
|
||||
* AFBC_V2 is for video decoder output.
|
||||
* It stores payload offset in first 32-bit in header address
|
||||
* Payload offset is always set to zero.
|
||||
*/
|
||||
#define MPP_FRAME_FBC_AFBC_V2 (0x00200000)
|
||||
|
||||
#define MPP_FRAME_FMT_LE_MASK (0x01000000)
|
||||
|
||||
#define MPP_FRAME_FMT_IS_YUV(fmt) (((fmt & MPP_FRAME_FMT_COLOR_MASK) == MPP_FRAME_FMT_YUV) && \
|
||||
((fmt & MPP_FRAME_FMT_MASK) < MPP_FMT_YUV_BUTT))
|
||||
#define MPP_FRAME_FMT_IS_YUV_10BIT(fmt) ((fmt & MPP_FRAME_FMT_MASK) == MPP_FMT_YUV420SP_10BIT || \
|
||||
(fmt & MPP_FRAME_FMT_MASK) == MPP_FMT_YUV422SP_10BIT)
|
||||
#define MPP_FRAME_FMT_IS_RGB(fmt) (((fmt & MPP_FRAME_FMT_COLOR_MASK) == MPP_FRAME_FMT_RGB) && \
|
||||
((fmt & MPP_FRAME_FMT_MASK) < MPP_FMT_RGB_BUTT))
|
||||
|
||||
/*
|
||||
* For MPP_FRAME_FBC_AFBC_V1 the 16byte aligned stride is used.
|
||||
*/
|
||||
#define MPP_FRAME_FMT_IS_FBC(fmt) (fmt & MPP_FRAME_FBC_MASK)
|
||||
|
||||
#define MPP_FRAME_FMT_IS_HDR(fmt) (fmt & MPP_FRAME_HDR_MASK)
|
||||
|
||||
#define MPP_FRAME_FMT_IS_LE(fmt) ((fmt & MPP_FRAME_FMT_LE_MASK) == MPP_FRAME_FMT_LE_MASK)
|
||||
#define MPP_FRAME_FMT_IS_BE(fmt) ((fmt & MPP_FRAME_FMT_LE_MASK) == 0)
|
||||
|
||||
/* mpp color format index definition */
|
||||
typedef enum {
|
||||
MPP_FMT_YUV420SP = (MPP_FRAME_FMT_YUV + 0), /* YYYY... UV... (NV12) */
|
||||
/*
|
||||
* A rockchip specific pixel format, without gap between pixel aganist
|
||||
* the P010_10LE/P010_10BE
|
||||
*/
|
||||
MPP_FMT_YUV420SP_10BIT = (MPP_FRAME_FMT_YUV + 1),
|
||||
MPP_FMT_YUV422SP = (MPP_FRAME_FMT_YUV + 2), /* YYYY... UVUV... (NV16) */
|
||||
MPP_FMT_YUV422SP_10BIT = (MPP_FRAME_FMT_YUV + 3), ///< Not part of ABI
|
||||
MPP_FMT_YUV420P = (MPP_FRAME_FMT_YUV + 4), /* YYYY... U...V... (I420) */
|
||||
MPP_FMT_YUV420SP_VU = (MPP_FRAME_FMT_YUV + 5), /* YYYY... VUVUVU... (NV21) */
|
||||
MPP_FMT_YUV422P = (MPP_FRAME_FMT_YUV + 6), /* YYYY... UU...VV...(422P) */
|
||||
MPP_FMT_YUV422SP_VU = (MPP_FRAME_FMT_YUV + 7), /* YYYY... VUVUVU... (NV61) */
|
||||
MPP_FMT_YUV422_YUYV = (MPP_FRAME_FMT_YUV + 8), /* YUYVYUYV... (YUY2) */
|
||||
MPP_FMT_YUV422_YVYU = (MPP_FRAME_FMT_YUV + 9), /* YVYUYVYU... (YVY2) */
|
||||
MPP_FMT_YUV422_UYVY = (MPP_FRAME_FMT_YUV + 10), /* UYVYUYVY... (UYVY) */
|
||||
MPP_FMT_YUV422_VYUY = (MPP_FRAME_FMT_YUV + 11), /* VYUYVYUY... (VYUY) */
|
||||
MPP_FMT_YUV400 = (MPP_FRAME_FMT_YUV + 12), /* YYYY... */
|
||||
MPP_FMT_YUV440SP = (MPP_FRAME_FMT_YUV + 13), /* YYYY... UVUV... */
|
||||
MPP_FMT_YUV411SP = (MPP_FRAME_FMT_YUV + 14), /* YYYY... UV... */
|
||||
MPP_FMT_YUV444SP = (MPP_FRAME_FMT_YUV + 15), /* YYYY... UVUVUVUV... */
|
||||
MPP_FMT_YUV444P = (MPP_FRAME_FMT_YUV + 16), /* YYYY... UUUU... VVVV... */
|
||||
MPP_FMT_YUV_BUTT,
|
||||
|
||||
MPP_FMT_RGB565 = (MPP_FRAME_FMT_RGB + 0), /* 16-bit RGB */
|
||||
MPP_FMT_BGR565 = (MPP_FRAME_FMT_RGB + 1), /* 16-bit RGB */
|
||||
MPP_FMT_RGB555 = (MPP_FRAME_FMT_RGB + 2), /* 15-bit RGB */
|
||||
MPP_FMT_BGR555 = (MPP_FRAME_FMT_RGB + 3), /* 15-bit RGB */
|
||||
MPP_FMT_RGB444 = (MPP_FRAME_FMT_RGB + 4), /* 12-bit RGB */
|
||||
MPP_FMT_BGR444 = (MPP_FRAME_FMT_RGB + 5), /* 12-bit RGB */
|
||||
MPP_FMT_RGB888 = (MPP_FRAME_FMT_RGB + 6), /* 24-bit RGB */
|
||||
MPP_FMT_BGR888 = (MPP_FRAME_FMT_RGB + 7), /* 24-bit RGB */
|
||||
MPP_FMT_RGB101010 = (MPP_FRAME_FMT_RGB + 8), /* 30-bit RGB */
|
||||
MPP_FMT_BGR101010 = (MPP_FRAME_FMT_RGB + 9), /* 30-bit RGB */
|
||||
MPP_FMT_ARGB8888 = (MPP_FRAME_FMT_RGB + 10), /* 32-bit RGB */
|
||||
MPP_FMT_ABGR8888 = (MPP_FRAME_FMT_RGB + 11), /* 32-bit RGB */
|
||||
MPP_FMT_BGRA8888 = (MPP_FRAME_FMT_RGB + 12), /* 32-bit RGB */
|
||||
MPP_FMT_RGBA8888 = (MPP_FRAME_FMT_RGB + 13), /* 32-bit RGB */
|
||||
MPP_FMT_RGB_BUTT,
|
||||
|
||||
MPP_FMT_BUTT,
|
||||
} MppFrameFormat;
|
||||
|
||||
/**
|
||||
* Rational number (pair of numerator and denominator).
|
||||
*/
|
||||
typedef struct MppFrameRational {
|
||||
RK_S32 num; ///< Numerator
|
||||
RK_S32 den; ///< Denominator
|
||||
} MppFrameRational;
|
||||
|
||||
typedef struct MppFrameMasteringDisplayMetadata {
|
||||
RK_U16 display_primaries[3][2];
|
||||
RK_U16 white_point[2];
|
||||
RK_U32 max_luminance;
|
||||
RK_U32 min_luminance;
|
||||
} MppFrameMasteringDisplayMetadata;
|
||||
|
||||
typedef struct MppFrameContentLightMetadata {
|
||||
RK_U16 MaxCLL;
|
||||
RK_U16 MaxFALL;
|
||||
} MppFrameContentLightMetadata;
|
||||
|
||||
typedef struct MppFrameHdrDynamicMeta {
|
||||
RK_U32 hdr_fmt;
|
||||
RK_U32 size;
|
||||
RK_U8 data[];
|
||||
} MppFrameHdrDynamicMeta;
|
||||
|
||||
typedef enum MppFrameError {
|
||||
/* General error not specified */
|
||||
MPP_FRAME_ERR_UNKNOW = 0x0001,
|
||||
|
||||
/* Critical error for decoder not support error */
|
||||
MPP_FRAME_ERR_UNSUPPORT = 0x0002,
|
||||
|
||||
/*
|
||||
* Fatal error for decoder can not parse a valid frame for hardware.
|
||||
* the pixel data is all invalid.
|
||||
*/
|
||||
MPP_FRAME_ERR_DEC_INVALID = 0x0010,
|
||||
|
||||
/*
|
||||
* Normal error for decoder found hardware error on decoding.
|
||||
*/
|
||||
MPP_FRAME_ERR_DEC_HW_ERR = 0x0100,
|
||||
|
||||
/*
|
||||
* Normal error for decoder found missing reference frame on decoding.
|
||||
*/
|
||||
MPP_FRAME_ERR_DEC_MISS_REF = 0x0200,
|
||||
} MppFrameError;
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
/*
|
||||
* MppFrame interface
|
||||
*/
|
||||
MPP_RET mpp_frame_init(MppFrame *frame);
|
||||
MPP_RET mpp_frame_deinit(MppFrame *frame);
|
||||
|
||||
/*
|
||||
* normal parameter
|
||||
*
|
||||
* offset_x
|
||||
* <-------->
|
||||
*
|
||||
* <---------------+ hor_stride +--------------->
|
||||
*
|
||||
* +------------------------------------------------------+ ^ ^
|
||||
* | | | |
|
||||
* | | | | offset_y
|
||||
* | | | |
|
||||
* | +--------------------------------+ ^ | | v
|
||||
* | | | | | |
|
||||
* | | | + | +
|
||||
* | | | |
|
||||
* | | valid data area | height | ver_stride
|
||||
* | | | |
|
||||
* | | | + | +
|
||||
* | | | | | |
|
||||
* | +--------------------------------+ v | |
|
||||
* | | |
|
||||
* | <----------+ width +---------> | |
|
||||
* | | |
|
||||
* +------------------------------------------------------+ v
|
||||
*
|
||||
*/
|
||||
RK_U32 mpp_frame_get_width(const MppFrame frame);
|
||||
void mpp_frame_set_width(MppFrame frame, RK_U32 width);
|
||||
RK_U32 mpp_frame_get_height(const MppFrame frame);
|
||||
void mpp_frame_set_height(MppFrame frame, RK_U32 height);
|
||||
RK_U32 mpp_frame_get_hor_stride(const MppFrame frame);
|
||||
void mpp_frame_set_hor_stride(MppFrame frame, RK_U32 hor_stride);
|
||||
RK_U32 mpp_frame_get_ver_stride(const MppFrame frame);
|
||||
void mpp_frame_set_ver_stride(MppFrame frame, RK_U32 ver_stride);
|
||||
void mpp_frame_set_hor_stride_pixel(MppFrame frame, RK_U32 hor_stride_pixel);
|
||||
RK_U32 mpp_frame_get_hor_stride_pixel(const MppFrame frame);
|
||||
void mpp_frame_set_fbc_hdr_stride(MppFrame frame, RK_U32 fbc_hdr_stride);
|
||||
RK_U32 mpp_frame_get_fbc_hdr_stride(const MppFrame frame);
|
||||
|
||||
RK_U32 mpp_frame_get_offset_x(const MppFrame frame);
|
||||
void mpp_frame_set_offset_x(MppFrame frame, RK_U32 offset_x);
|
||||
RK_U32 mpp_frame_get_offset_y(const MppFrame frame);
|
||||
void mpp_frame_set_offset_y(MppFrame frame, RK_U32 offset_y);
|
||||
RK_U32 mpp_frame_get_mode(const MppFrame frame);
|
||||
void mpp_frame_set_mode(MppFrame frame, RK_U32 mode);
|
||||
RK_U32 mpp_frame_get_discard(const MppFrame frame);
|
||||
void mpp_frame_set_discard(MppFrame frame, RK_U32 discard);
|
||||
RK_U32 mpp_frame_get_viewid(const MppFrame frame);
|
||||
void mpp_frame_set_viewid(MppFrame frame, RK_U32 viewid);
|
||||
RK_U32 mpp_frame_get_poc(const MppFrame frame);
|
||||
void mpp_frame_set_poc(MppFrame frame, RK_U32 poc);
|
||||
RK_S64 mpp_frame_get_pts(const MppFrame frame);
|
||||
void mpp_frame_set_pts(MppFrame frame, RK_S64 pts);
|
||||
RK_S64 mpp_frame_get_dts(const MppFrame frame);
|
||||
void mpp_frame_set_dts(MppFrame frame, RK_S64 dts);
|
||||
RK_U32 mpp_frame_get_errinfo(const MppFrame frame);
|
||||
void mpp_frame_set_errinfo(MppFrame frame, RK_U32 errinfo);
|
||||
size_t mpp_frame_get_buf_size(const MppFrame frame);
|
||||
void mpp_frame_set_buf_size(MppFrame frame, size_t buf_size);
|
||||
void mpp_frame_set_thumbnail_en(MppFrame frame, RK_U32 thumbnail_en);
|
||||
RK_U32 mpp_frame_get_thumbnail_en(const MppFrame frame);
|
||||
|
||||
/*
|
||||
* flow control parmeter
|
||||
*/
|
||||
RK_U32 mpp_frame_get_eos(const MppFrame frame);
|
||||
void mpp_frame_set_eos(MppFrame frame, RK_U32 eos);
|
||||
RK_U32 mpp_frame_get_info_change(const MppFrame frame);
|
||||
void mpp_frame_set_info_change(MppFrame frame, RK_U32 info_change);
|
||||
|
||||
/*
|
||||
* buffer parameter
|
||||
*/
|
||||
MppBuffer mpp_frame_get_buffer(const MppFrame frame);
|
||||
void mpp_frame_set_buffer(MppFrame frame, MppBuffer buffer);
|
||||
|
||||
/*
|
||||
* meta data parameter
|
||||
*/
|
||||
RK_S32 mpp_frame_has_meta(const MppFrame frame);
|
||||
MppMeta mpp_frame_get_meta(const MppFrame frame);
|
||||
void mpp_frame_set_meta(MppFrame frame, MppMeta meta);
|
||||
|
||||
/*
|
||||
* color related parameter
|
||||
*/
|
||||
MppFrameColorRange mpp_frame_get_color_range(const MppFrame frame);
|
||||
void mpp_frame_set_color_range(MppFrame frame, MppFrameColorRange color_range);
|
||||
MppFrameColorPrimaries mpp_frame_get_color_primaries(const MppFrame frame);
|
||||
void mpp_frame_set_color_primaries(MppFrame frame, MppFrameColorPrimaries color_primaries);
|
||||
MppFrameColorTransferCharacteristic mpp_frame_get_color_trc(const MppFrame frame);
|
||||
void mpp_frame_set_color_trc(MppFrame frame, MppFrameColorTransferCharacteristic color_trc);
|
||||
MppFrameColorSpace mpp_frame_get_colorspace(const MppFrame frame);
|
||||
void mpp_frame_set_colorspace(MppFrame frame, MppFrameColorSpace colorspace);
|
||||
MppFrameChromaLocation mpp_frame_get_chroma_location(const MppFrame frame);
|
||||
void mpp_frame_set_chroma_location(MppFrame frame, MppFrameChromaLocation chroma_location);
|
||||
MppFrameFormat mpp_frame_get_fmt(MppFrame frame);
|
||||
void mpp_frame_set_fmt(MppFrame frame, MppFrameFormat fmt);
|
||||
MppFrameRational mpp_frame_get_sar(const MppFrame frame);
|
||||
void mpp_frame_set_sar(MppFrame frame, MppFrameRational sar);
|
||||
MppFrameMasteringDisplayMetadata mpp_frame_get_mastering_display(const MppFrame frame);
|
||||
void mpp_frame_set_mastering_display(MppFrame frame, MppFrameMasteringDisplayMetadata mastering_display);
|
||||
MppFrameContentLightMetadata mpp_frame_get_content_light(const MppFrame frame);
|
||||
void mpp_frame_set_content_light(MppFrame frame, MppFrameContentLightMetadata content_light);
|
||||
MppFrameHdrDynamicMeta* mpp_frame_get_hdr_dynamic_meta(const MppFrame frame);
|
||||
void mpp_frame_set_hdr_dynamic_meta(MppFrame frame, MppFrameHdrDynamicMeta *vivi_data);
|
||||
|
||||
/*
|
||||
* HDR parameter
|
||||
*/
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__MPP_FRAME_H__*/
|
||||
99
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_log.h
vendored
Normal file
99
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_log.h
vendored
Normal file
|
|
@ -0,0 +1,99 @@
|
|||
/*
|
||||
* Copyright 2022 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_LOG_H__
|
||||
#define __MPP_LOG_H__
|
||||
|
||||
#include "rk_type.h"
|
||||
#include "mpp_log_def.h"
|
||||
|
||||
/*
|
||||
* _c function will add condition check
|
||||
* _f function will add function name to the log
|
||||
* _cf function will add both function name and condition check
|
||||
*/
|
||||
|
||||
/*
|
||||
* mpp runtime log system usage:
|
||||
* mpp_logf is for fatal logging. For use when aborting
|
||||
* mpp_loge is for error logging. For use with unrecoverable failures.
|
||||
* mpp_logw is for warning logging. For use with recoverable failures.
|
||||
* mpp_logi is for informational logging.
|
||||
* mpp_logd is for debug logging.
|
||||
* mpp_logv is for verbose logging
|
||||
*/
|
||||
|
||||
#define mpp_logf(fmt, ...) _mpp_log_l(MPP_LOG_FATAL, MODULE_TAG, fmt, NULL, ## __VA_ARGS__)
|
||||
#define mpp_loge(fmt, ...) _mpp_log_l(MPP_LOG_ERROR, MODULE_TAG, fmt, NULL, ## __VA_ARGS__)
|
||||
#define mpp_logw(fmt, ...) _mpp_log_l(MPP_LOG_WARN, MODULE_TAG, fmt, NULL, ## __VA_ARGS__)
|
||||
#define mpp_logi(fmt, ...) _mpp_log_l(MPP_LOG_INFO, MODULE_TAG, fmt, NULL, ## __VA_ARGS__)
|
||||
#define mpp_logd(fmt, ...) _mpp_log_l(MPP_LOG_DEBUG, MODULE_TAG, fmt, NULL, ## __VA_ARGS__)
|
||||
#define mpp_logv(fmt, ...) _mpp_log_l(MPP_LOG_VERBOSE, MODULE_TAG, fmt, NULL, ## __VA_ARGS__)
|
||||
|
||||
#define mpp_logf_f(fmt, ...) _mpp_log_l(MPP_LOG_FATAL, MODULE_TAG, fmt, __FUNCTION__, ## __VA_ARGS__)
|
||||
#define mpp_loge_f(fmt, ...) _mpp_log_l(MPP_LOG_ERROR, MODULE_TAG, fmt, __FUNCTION__, ## __VA_ARGS__)
|
||||
#define mpp_logw_f(fmt, ...) _mpp_log_l(MPP_LOG_WARN, MODULE_TAG, fmt, __FUNCTION__, ## __VA_ARGS__)
|
||||
#define mpp_logi_f(fmt, ...) _mpp_log_l(MPP_LOG_INFO, MODULE_TAG, fmt, __FUNCTION__, ## __VA_ARGS__)
|
||||
#define mpp_logd_f(fmt, ...) _mpp_log_l(MPP_LOG_DEBUG, MODULE_TAG, fmt, __FUNCTION__, ## __VA_ARGS__)
|
||||
#define mpp_logv_f(fmt, ...) _mpp_log_l(MPP_LOG_VERBOSE, MODULE_TAG, fmt, __FUNCTION__, ## __VA_ARGS__)
|
||||
|
||||
#define mpp_logf_c(cond, fmt, ...) do { if (cond) mpp_logf(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_loge_c(cond, fmt, ...) do { if (cond) mpp_loge(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_logw_c(cond, fmt, ...) do { if (cond) mpp_logw(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_logi_c(cond, fmt, ...) do { if (cond) mpp_logi(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_logd_c(cond, fmt, ...) do { if (cond) mpp_logd(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_logv_c(cond, fmt, ...) do { if (cond) mpp_logv(fmt, ## __VA_ARGS__); } while (0)
|
||||
|
||||
#define mpp_logf_cf(cond, fmt, ...) do { if (cond) mpp_logf_f(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_loge_cf(cond, fmt, ...) do { if (cond) mpp_loge_f(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_logw_cf(cond, fmt, ...) do { if (cond) mpp_logw_f(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_logi_cf(cond, fmt, ...) do { if (cond) mpp_logi_f(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_logd_cf(cond, fmt, ...) do { if (cond) mpp_logd_f(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_logv_cf(cond, fmt, ...) do { if (cond) mpp_logv_f(fmt, ## __VA_ARGS__); } while (0)
|
||||
|
||||
/*
|
||||
* mpp runtime log system usage:
|
||||
* mpp_err is for error status message, it will print for sure.
|
||||
* mpp_log is for important message like open/close/reset/flush, it will print too.
|
||||
*/
|
||||
|
||||
#define mpp_log(fmt, ...) mpp_logi(fmt, ## __VA_ARGS__)
|
||||
#define mpp_err(fmt, ...) mpp_loge(fmt, ## __VA_ARGS__)
|
||||
|
||||
#define mpp_log_f(fmt, ...) mpp_logi_f(fmt, ## __VA_ARGS__)
|
||||
#define mpp_err_f(fmt, ...) mpp_loge_f(fmt, ## __VA_ARGS__)
|
||||
|
||||
#define mpp_log_c(cond, fmt, ...) do { if (cond) mpp_log(fmt, ## __VA_ARGS__); } while (0)
|
||||
#define mpp_log_cf(cond, fmt, ...) do { if (cond) mpp_log_f(fmt, ## __VA_ARGS__); } while (0)
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
void _mpp_log_l(int level, const char *tag, const char *fmt, const char *func, ...);
|
||||
|
||||
void mpp_set_log_level(int level);
|
||||
int mpp_get_log_level(void);
|
||||
|
||||
/* deprecated function */
|
||||
void _mpp_log(const char *tag, const char *fmt, const char *func, ...);
|
||||
void _mpp_err(const char *tag, const char *fmt, const char *func, ...);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__MPP_LOG_H__*/
|
||||
37
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_log_def.h
vendored
Normal file
37
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_log_def.h
vendored
Normal file
|
|
@ -0,0 +1,37 @@
|
|||
/*
|
||||
* Copyright 2022 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_LOG_DEF_H__
|
||||
#define __MPP_LOG_DEF_H__
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
#define MPP_LOG_UNKNOWN 0 /* internal use only */
|
||||
#define MPP_LOG_FATAL 1 /* fatal error on aborting */
|
||||
#define MPP_LOG_ERROR 2 /* error log on unrecoverable failures */
|
||||
#define MPP_LOG_WARN 3 /* warning log on recoverable failures */
|
||||
#define MPP_LOG_INFO 4 /* Informational log */
|
||||
#define MPP_LOG_DEBUG 5 /* Debug log */
|
||||
#define MPP_LOG_VERBOSE 6 /* Verbose log */
|
||||
#define MPP_LOG_SILENT 7 /* internal use only */
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__MPP_LOG_DEF_H__*/
|
||||
179
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_meta.h
vendored
Normal file
179
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_meta.h
vendored
Normal file
|
|
@ -0,0 +1,179 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_META_H__
|
||||
#define __MPP_META_H__
|
||||
|
||||
#include <stdint.h>
|
||||
#include "rk_type.h"
|
||||
|
||||
#define FOURCC_META(a, b, c, d) ((RK_U32)(a) << 24 | \
|
||||
((RK_U32)(b) << 16) | \
|
||||
((RK_U32)(c) << 8) | \
|
||||
((RK_U32)(d) << 0))
|
||||
|
||||
/*
|
||||
* Mpp Metadata definition
|
||||
*
|
||||
* Metadata is for information transmision in mpp.
|
||||
* Mpp task will contain two meta data:
|
||||
*
|
||||
* 1. Data flow metadata
|
||||
* This metadata contains information of input / output data flow. For example
|
||||
* A. decoder input side task the input packet must be defined and output frame
|
||||
* may not be defined. Then decoder will try malloc or use committed buffer to
|
||||
* complete decoding.
|
||||
* B. decoder output side task
|
||||
*
|
||||
*
|
||||
* 2. Flow control metadata
|
||||
*
|
||||
*/
|
||||
typedef enum MppMetaDataType_e {
|
||||
/*
|
||||
* mpp meta data of data flow
|
||||
* reference counter will be used for these meta data type
|
||||
*/
|
||||
TYPE_FRAME = FOURCC_META('m', 'f', 'r', 'm'),
|
||||
TYPE_PACKET = FOURCC_META('m', 'p', 'k', 't'),
|
||||
TYPE_BUFFER = FOURCC_META('m', 'b', 'u', 'f'),
|
||||
|
||||
/* mpp meta data of normal data type */
|
||||
TYPE_S32 = FOURCC_META('s', '3', '2', ' '),
|
||||
TYPE_S64 = FOURCC_META('s', '6', '4', ' '),
|
||||
TYPE_PTR = FOURCC_META('p', 't', 'r', ' '),
|
||||
} MppMetaType;
|
||||
|
||||
typedef enum MppMetaKey_e {
|
||||
/* data flow key */
|
||||
KEY_INPUT_FRAME = FOURCC_META('i', 'f', 'r', 'm'),
|
||||
KEY_INPUT_PACKET = FOURCC_META('i', 'p', 'k', 't'),
|
||||
KEY_OUTPUT_FRAME = FOURCC_META('o', 'f', 'r', 'm'),
|
||||
KEY_OUTPUT_PACKET = FOURCC_META('o', 'p', 'k', 't'),
|
||||
/* output motion information for motion detection */
|
||||
KEY_MOTION_INFO = FOURCC_META('m', 'v', 'i', 'f'),
|
||||
KEY_HDR_INFO = FOURCC_META('h', 'd', 'r', ' '),
|
||||
KEY_HDR_META_OFFSET = FOURCC_META('h', 'd', 'r', 'o'),
|
||||
KEY_HDR_META_SIZE = FOURCC_META('h', 'd', 'r', 'l'),
|
||||
|
||||
/* flow control key */
|
||||
KEY_INPUT_BLOCK = FOURCC_META('i', 'b', 'l', 'k'),
|
||||
KEY_OUTPUT_BLOCK = FOURCC_META('o', 'b', 'l', 'k'),
|
||||
KEY_INPUT_IDR_REQ = FOURCC_META('i', 'i', 'd', 'r'), /* input idr frame request flag */
|
||||
KEY_OUTPUT_INTRA = FOURCC_META('o', 'i', 'd', 'r'), /* output intra frame indicator */
|
||||
|
||||
/* mpp_frame / mpp_packet meta data info key */
|
||||
KEY_TEMPORAL_ID = FOURCC_META('t', 'l', 'i', 'd'),
|
||||
KEY_LONG_REF_IDX = FOURCC_META('l', 't', 'i', 'd'),
|
||||
KEY_ENC_AVERAGE_QP = FOURCC_META('a', 'v', 'g', 'q'),
|
||||
KEY_ROI_DATA = FOURCC_META('r', 'o', 'i', ' '),
|
||||
KEY_OSD_DATA = FOURCC_META('o', 's', 'd', ' '),
|
||||
KEY_OSD_DATA2 = FOURCC_META('o', 's', 'd', '2'),
|
||||
KEY_USER_DATA = FOURCC_META('u', 's', 'r', 'd'),
|
||||
KEY_USER_DATAS = FOURCC_META('u', 'r', 'd', 's'),
|
||||
|
||||
/*
|
||||
* For vepu580 roi buffer config mode
|
||||
* The encoder roi structure is so complex that we should provide a buffer
|
||||
* tunnel for externl user to config encoder hardware by direct sending
|
||||
* roi data buffer.
|
||||
* This way can reduce the config parsing and roi buffer data generating
|
||||
* overhead in mpp.
|
||||
*/
|
||||
KEY_ROI_DATA2 = FOURCC_META('r', 'o', 'i', '2'),
|
||||
|
||||
/*
|
||||
* qpmap for rv1109/1126 encoder qpmap config
|
||||
* Input data is a MppBuffer which contains an array of 16bit Vepu541RoiCfg.
|
||||
* And each 16bit represents a 16x16 block qp info.
|
||||
*
|
||||
* H.264 - 16x16 block qp is arranged in raster order:
|
||||
* each value is a 16bit data
|
||||
* 00 01 02 03 04 05 06 07 -> 00 01 02 03 04 05 06 07
|
||||
* 10 11 12 13 14 15 16 17 10 11 12 13 14 15 16 17
|
||||
* 20 21 22 23 24 25 26 27 20 21 22 23 24 25 26 27
|
||||
* 30 31 32 33 34 35 36 37 30 31 32 33 34 35 36 37
|
||||
*
|
||||
* H.265 - 16x16 block qp is reorder to 64x64/32x32 ctu order then 64x64 / 32x32 ctu raster order
|
||||
* 64x64 ctu
|
||||
* 00 01 02 03 04 05 06 07 -> 00 01 02 03 10 11 12 13 20 21 22 23 30 31 32 33 04 05 06 07 14 15 16 17 24 25 26 27 34 35 36 37
|
||||
* 10 11 12 13 14 15 16 17
|
||||
* 20 21 22 23 24 25 26 27
|
||||
* 30 31 32 33 34 35 36 37
|
||||
* 32x32 ctu
|
||||
* 00 01 02 03 04 05 06 07 -> 00 01 10 11 02 03 12 13 04 05 14 15 06 07 16 17
|
||||
* 10 11 12 13 14 15 16 17 20 21 30 31 22 23 32 33 24 25 34 35 26 27 36 37
|
||||
* 20 21 22 23 24 25 26 27
|
||||
* 30 31 32 33 34 35 36 37
|
||||
*/
|
||||
KEY_QPMAP0 = FOURCC_META('e', 'q', 'm', '0'),
|
||||
|
||||
/* input motion list for smart p rate control */
|
||||
KEY_MV_LIST = FOURCC_META('m', 'v', 'l', 't'),
|
||||
|
||||
/* frame long-term reference frame operation */
|
||||
KEY_ENC_MARK_LTR = FOURCC_META('m', 'l', 't', 'r'),
|
||||
KEY_ENC_USE_LTR = FOURCC_META('u', 'l', 't', 'r'),
|
||||
|
||||
/* MLVEC specified encoder feature */
|
||||
KEY_ENC_FRAME_QP = FOURCC_META('f', 'r', 'm', 'q'),
|
||||
KEY_ENC_BASE_LAYER_PID = FOURCC_META('b', 'p', 'i', 'd'),
|
||||
|
||||
/* Thumbnail info for decoder output frame */
|
||||
KEY_DEC_TBN_EN = FOURCC_META('t', 'b', 'e', 'n'),
|
||||
KEY_DEC_TBN_Y_OFFSET = FOURCC_META('t', 'b', 'y', 'o'),
|
||||
KEY_DEC_TBN_UV_OFFSET = FOURCC_META('t', 'b', 'c', 'o'),
|
||||
} MppMetaKey;
|
||||
|
||||
#define mpp_meta_get(meta) mpp_meta_get_with_tag(meta, MODULE_TAG, __FUNCTION__)
|
||||
|
||||
#include "mpp_frame.h"
|
||||
#include "mpp_packet.h"
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
MPP_RET mpp_meta_get_with_tag(MppMeta *meta, const char *tag, const char *caller);
|
||||
MPP_RET mpp_meta_put(MppMeta meta);
|
||||
RK_S32 mpp_meta_size(MppMeta meta);
|
||||
|
||||
MPP_RET mpp_meta_set_s32(MppMeta meta, MppMetaKey key, RK_S32 val);
|
||||
MPP_RET mpp_meta_set_s64(MppMeta meta, MppMetaKey key, RK_S64 val);
|
||||
MPP_RET mpp_meta_set_ptr(MppMeta meta, MppMetaKey key, void *val);
|
||||
MPP_RET mpp_meta_get_s32(MppMeta meta, MppMetaKey key, RK_S32 *val);
|
||||
MPP_RET mpp_meta_get_s64(MppMeta meta, MppMetaKey key, RK_S64 *val);
|
||||
MPP_RET mpp_meta_get_ptr(MppMeta meta, MppMetaKey key, void **val);
|
||||
|
||||
MPP_RET mpp_meta_set_frame (MppMeta meta, MppMetaKey key, MppFrame frame);
|
||||
MPP_RET mpp_meta_set_packet(MppMeta meta, MppMetaKey key, MppPacket packet);
|
||||
MPP_RET mpp_meta_set_buffer(MppMeta meta, MppMetaKey key, MppBuffer buffer);
|
||||
MPP_RET mpp_meta_get_frame (MppMeta meta, MppMetaKey key, MppFrame *frame);
|
||||
MPP_RET mpp_meta_get_packet(MppMeta meta, MppMetaKey key, MppPacket *packet);
|
||||
MPP_RET mpp_meta_get_buffer(MppMeta meta, MppMetaKey key, MppBuffer *buffer);
|
||||
|
||||
MPP_RET mpp_meta_get_s32_d(MppMeta meta, MppMetaKey key, RK_S32 *val, RK_S32 def);
|
||||
MPP_RET mpp_meta_get_s64_d(MppMeta meta, MppMetaKey key, RK_S64 *val, RK_S64 def);
|
||||
MPP_RET mpp_meta_get_ptr_d(MppMeta meta, MppMetaKey key, void **val, void *def);
|
||||
MPP_RET mpp_meta_get_frame_d(MppMeta meta, MppMetaKey key, MppFrame *frame, MppFrame def);
|
||||
MPP_RET mpp_meta_get_packet_d(MppMeta meta, MppMetaKey key, MppPacket *packet, MppPacket def);
|
||||
MPP_RET mpp_meta_get_buffer_d(MppMeta meta, MppMetaKey key, MppBuffer *buffer, MppBuffer def);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__MPP_META_H__*/
|
||||
117
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_packet.h
vendored
Normal file
117
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_packet.h
vendored
Normal file
|
|
@ -0,0 +1,117 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_PACKET_H__
|
||||
#define __MPP_PACKET_H__
|
||||
|
||||
#include "mpp_meta.h"
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
/*
|
||||
* MppPacket interface
|
||||
*
|
||||
* mpp_packet_init = mpp_packet_new + mpp_packet_set_data + mpp_packet_set_size
|
||||
* mpp_packet_copy_init = mpp_packet_init + memcpy
|
||||
*/
|
||||
MPP_RET mpp_packet_new(MppPacket *packet);
|
||||
MPP_RET mpp_packet_init(MppPacket *packet, void *data, size_t size);
|
||||
MPP_RET mpp_packet_init_with_buffer(MppPacket *packet, MppBuffer buffer);
|
||||
MPP_RET mpp_packet_copy_init(MppPacket *packet, const MppPacket src);
|
||||
MPP_RET mpp_packet_deinit(MppPacket *packet);
|
||||
|
||||
/*
|
||||
* data : ( R/W ) start address of the whole packet memory
|
||||
* size : ( R/W ) total size of the whole packet memory
|
||||
* pos : ( R/W ) current access position of the whole packet memory, used for buffer read/write
|
||||
* length : ( R/W ) the rest length from current position to end of buffer
|
||||
* NOTE: normally length is updated only by set_pos,
|
||||
* so set length must be used carefully for special usage
|
||||
*/
|
||||
void mpp_packet_set_data(MppPacket packet, void *data);
|
||||
void mpp_packet_set_size(MppPacket packet, size_t size);
|
||||
void mpp_packet_set_pos(MppPacket packet, void *pos);
|
||||
void mpp_packet_set_length(MppPacket packet, size_t size);
|
||||
|
||||
void* mpp_packet_get_data(const MppPacket packet);
|
||||
void* mpp_packet_get_pos(const MppPacket packet);
|
||||
size_t mpp_packet_get_size(const MppPacket packet);
|
||||
size_t mpp_packet_get_length(const MppPacket packet);
|
||||
|
||||
|
||||
void mpp_packet_set_pts(MppPacket packet, RK_S64 pts);
|
||||
RK_S64 mpp_packet_get_pts(const MppPacket packet);
|
||||
void mpp_packet_set_dts(MppPacket packet, RK_S64 dts);
|
||||
RK_S64 mpp_packet_get_dts(const MppPacket packet);
|
||||
|
||||
void mpp_packet_set_flag(MppPacket packet, RK_U32 flag);
|
||||
RK_U32 mpp_packet_get_flag(const MppPacket packet);
|
||||
|
||||
MPP_RET mpp_packet_set_eos(MppPacket packet);
|
||||
MPP_RET mpp_packet_clr_eos(MppPacket packet);
|
||||
RK_U32 mpp_packet_get_eos(MppPacket packet);
|
||||
MPP_RET mpp_packet_set_extra_data(MppPacket packet);
|
||||
|
||||
void mpp_packet_set_buffer(MppPacket packet, MppBuffer buffer);
|
||||
MppBuffer mpp_packet_get_buffer(const MppPacket packet);
|
||||
|
||||
/*
|
||||
* data access interface
|
||||
*/
|
||||
MPP_RET mpp_packet_read(MppPacket packet, size_t offset, void *data, size_t size);
|
||||
MPP_RET mpp_packet_write(MppPacket packet, size_t offset, void *data, size_t size);
|
||||
|
||||
/*
|
||||
* meta data access interface
|
||||
*/
|
||||
RK_S32 mpp_packet_has_meta(const MppPacket packet);
|
||||
MppMeta mpp_packet_get_meta(const MppPacket packet);
|
||||
|
||||
/*
|
||||
* multi packet sequence interface for slice/split encoding/decoding
|
||||
* partition - the packet is a part of a while image
|
||||
* soi - Start Of Image
|
||||
* eoi - End Of Image
|
||||
*/
|
||||
RK_U32 mpp_packet_is_partition(const MppPacket packet);
|
||||
RK_U32 mpp_packet_is_soi(const MppPacket packet);
|
||||
RK_U32 mpp_packet_is_eoi(const MppPacket packet);
|
||||
|
||||
/*
|
||||
* packet segement pack info for
|
||||
* segment number - number of segment
|
||||
* segment info - base address of segment info
|
||||
*/
|
||||
typedef struct MppPktSeg_t MppPktSeg;
|
||||
|
||||
struct MppPktSeg_t {
|
||||
RK_S32 index;
|
||||
RK_S32 type;
|
||||
RK_U32 offset;
|
||||
RK_U32 len;
|
||||
const MppPktSeg *next;
|
||||
};
|
||||
|
||||
RK_U32 mpp_packet_get_segment_nb(const MppPacket packet);
|
||||
const MppPktSeg *mpp_packet_get_segment_info(const MppPacket packet);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__MPP_PACKET_H__*/
|
||||
257
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_rc_api.h
vendored
Normal file
257
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_rc_api.h
vendored
Normal file
|
|
@ -0,0 +1,257 @@
|
|||
/*
|
||||
* Copyright 2016 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_RC_API_H__
|
||||
#define __MPP_RC_API_H__
|
||||
|
||||
#include "mpp_err.h"
|
||||
#include "rk_venc_rc.h"
|
||||
#include "mpp_rc_defs.h"
|
||||
|
||||
/*
|
||||
* Mpp rate control has three parts:
|
||||
*
|
||||
* 1. MPI user config module
|
||||
* MppEncRcCfg structure is provided to user for overall rate control config
|
||||
* Mpp will receive MppEncRcCfg from user, check parameter and set it to
|
||||
* encoder.
|
||||
*
|
||||
* 2. Encoder rate control module
|
||||
* Encoder will implement the rate control strategy required by users
|
||||
* including CBR, VBR, AVBR and so on.
|
||||
* This module only implement the target bit calculation behavior and
|
||||
* quality restriction. And the quality level will be controlled by hal.
|
||||
*
|
||||
* 3. Hal rate control module
|
||||
* Hal will implement the rate control on hardware. Hal will calculate the
|
||||
* QP parameter for hardware according to the frame level target bit
|
||||
* specified by the encoder. And the report the real bitrate and quality to
|
||||
* encoder.
|
||||
*
|
||||
* The header defines the communication interfaces and structures used between
|
||||
* MPI, encoder and hal.
|
||||
*/
|
||||
|
||||
typedef enum RcMode_e {
|
||||
RC_VBR,
|
||||
RC_CBR,
|
||||
RC_FIXQP,
|
||||
RC_AVBR,
|
||||
RC_CVBR,
|
||||
RC_QVBR,
|
||||
RC_LEARNING,
|
||||
RC_MODE_BUTT,
|
||||
} RcMode;
|
||||
|
||||
typedef enum GopMode_e {
|
||||
NORMAL_P,
|
||||
SMART_P,
|
||||
} GopMode;
|
||||
|
||||
/*
|
||||
* frame rate parameters have great effect on rate control
|
||||
*
|
||||
* fps_in_flex
|
||||
* 0 - fix input frame rate
|
||||
* 1 - variable input frame rate
|
||||
*
|
||||
* fps_in_num
|
||||
* input frame rate numerator, if 0 then default 30
|
||||
*
|
||||
* fps_in_denorm
|
||||
* input frame rate denorminator, if 0 then default 1
|
||||
*
|
||||
* fps_out_flex
|
||||
* 0 - fix output frame rate
|
||||
* 1 - variable output frame rate
|
||||
*
|
||||
* fps_out_num
|
||||
* output frame rate numerator, if 0 then default 30
|
||||
*
|
||||
* fps_out_denorm
|
||||
* output frame rate denorminator, if 0 then default 1
|
||||
*/
|
||||
typedef struct RcFpsCfg_t {
|
||||
RK_S32 fps_in_flex;
|
||||
RK_S32 fps_in_num;
|
||||
RK_S32 fps_in_denorm;
|
||||
RK_S32 fps_out_flex;
|
||||
RK_S32 fps_out_num;
|
||||
RK_S32 fps_out_denorm;
|
||||
} RcFpsCfg;
|
||||
|
||||
typedef struct RcSuperframeCfg_t {
|
||||
MppEncRcSuperFrameMode super_mode;
|
||||
RK_U32 super_i_thd;
|
||||
RK_U32 super_p_thd;
|
||||
MppEncRcPriority rc_priority;
|
||||
} RcSuperframeCfg;
|
||||
|
||||
typedef struct RcDebreathCfg_t {
|
||||
RK_U32 enable;
|
||||
RK_U32 strength;
|
||||
} RcDebreathCfg;
|
||||
|
||||
typedef struct RcHierQPCfg_t {
|
||||
RK_S32 hier_qp_en;
|
||||
RK_S32 hier_qp_delta[4];
|
||||
RK_S32 hier_frame_num[4];
|
||||
} RcHierQPCfg;
|
||||
|
||||
/*
|
||||
* Control parameter from external config
|
||||
*
|
||||
* It will be updated on rc/prep/gopref config changed.
|
||||
*/
|
||||
typedef struct RcCfg_s {
|
||||
/* encode image size */
|
||||
RK_S32 width;
|
||||
RK_S32 height;
|
||||
|
||||
/* Use rc_mode to find different api */
|
||||
RcMode mode;
|
||||
|
||||
RcFpsCfg fps;
|
||||
|
||||
GopMode gop_mode;
|
||||
/* I frame gop len */
|
||||
RK_S32 igop;
|
||||
/* visual gop len */
|
||||
RK_S32 vgop;
|
||||
|
||||
/* bitrate parameter */
|
||||
RK_S32 bps_min;
|
||||
RK_S32 bps_target;
|
||||
RK_S32 bps_max;
|
||||
RK_S32 stats_time;
|
||||
|
||||
/* max I frame bit ratio to P frame bit */
|
||||
RK_S32 max_i_bit_prop;
|
||||
RK_S32 min_i_bit_prop;
|
||||
RK_S32 init_ip_ratio;
|
||||
/* layer bitrate proportion */
|
||||
RK_S32 layer_bit_prop[4];
|
||||
|
||||
/* quality parameter */
|
||||
RK_S32 init_quality;
|
||||
RK_S32 max_quality;
|
||||
RK_S32 min_quality;
|
||||
RK_S32 max_i_quality;
|
||||
RK_S32 min_i_quality;
|
||||
RK_S32 i_quality_delta;
|
||||
RK_S32 vi_quality_delta;
|
||||
/* layer quality proportion */
|
||||
RK_S32 layer_quality_delta[4];
|
||||
|
||||
/* reencode parameter */
|
||||
RK_S32 max_reencode_times;
|
||||
|
||||
/* still / motion desision parameter */
|
||||
RK_S32 min_still_prop;
|
||||
RK_S32 max_still_quality;
|
||||
|
||||
/*
|
||||
* vbr parameter
|
||||
*
|
||||
* vbr_hi_prop - high proportion bitrate for reduce quality
|
||||
* vbr_lo_prop - low proportion bitrate for increase quality
|
||||
*/
|
||||
RK_S32 vbr_hi_prop;
|
||||
RK_S32 vbr_lo_prop;
|
||||
|
||||
MppEncRcDropFrmMode drop_mode;
|
||||
RK_U32 drop_thd;
|
||||
RK_U32 drop_gap;
|
||||
|
||||
RcSuperframeCfg super_cfg;
|
||||
RcDebreathCfg debreath_cfg;
|
||||
RcHierQPCfg hier_qp_cfg;
|
||||
RK_U32 refresh_len;
|
||||
} RcCfg;
|
||||
|
||||
/*
|
||||
* Different rate control strategy will be implemented by different API config
|
||||
*/
|
||||
typedef struct RcImplApi_t {
|
||||
char *name;
|
||||
MppCodingType type;
|
||||
RK_U32 ctx_size;
|
||||
|
||||
MPP_RET (*init)(void *ctx, RcCfg *cfg);
|
||||
MPP_RET (*deinit)(void *ctx);
|
||||
|
||||
MPP_RET (*check_drop)(void *ctx, EncRcTask *task);
|
||||
MPP_RET (*check_reenc)(void *ctx, EncRcTask *task);
|
||||
|
||||
/*
|
||||
* frm_start - frame level rate control frm_start.
|
||||
* The EncRcTaskInfo will be output to hal for hardware to implement.
|
||||
* frm_end - frame level rate control frm_end.
|
||||
* The EncRcTaskInfo is returned for real quality and bitrate.
|
||||
*/
|
||||
MPP_RET (*frm_start)(void *ctx, EncRcTask *task);
|
||||
MPP_RET (*frm_end)(void *ctx, EncRcTask *task);
|
||||
|
||||
/*
|
||||
* hal_start - hardware level rate control start.
|
||||
* The EncRcTaskInfo will be output to hal for hardware to implement.
|
||||
* hal_end - hardware level rate control end.
|
||||
* The EncRcTaskInfo is returned for real quality and bitrate.
|
||||
*/
|
||||
MPP_RET (*hal_start)(void *ctx, EncRcTask *task);
|
||||
MPP_RET (*hal_end)(void *ctx, EncRcTask *task);
|
||||
} RcImplApi;
|
||||
|
||||
/*
|
||||
* structures for RC API register and query
|
||||
*/
|
||||
typedef struct RcApiBrief_t {
|
||||
const char *name;
|
||||
MppCodingType type;
|
||||
} RcApiBrief;
|
||||
|
||||
typedef struct RcApiQueryAll_t {
|
||||
/* input param for query */
|
||||
RcApiBrief *brief;
|
||||
RK_S32 max_count;
|
||||
|
||||
/* output query count */
|
||||
RK_S32 count;
|
||||
} RcApiQueryAll;
|
||||
|
||||
typedef struct RcApiQueryType_t {
|
||||
/* input param for query */
|
||||
RcApiBrief *brief;
|
||||
RK_S32 max_count;
|
||||
MppCodingType type;
|
||||
|
||||
/* output query count */
|
||||
RK_S32 count;
|
||||
} RcApiQueryType;
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
MPP_RET rc_api_add(const RcImplApi *api);
|
||||
MPP_RET rc_brief_get_all(RcApiQueryAll *query);
|
||||
MPP_RET rc_brief_get_by_type(RcApiQueryType *query);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /* __MPP_RC_API_H__ */
|
||||
212
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_rc_defs.h
vendored
Normal file
212
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_rc_defs.h
vendored
Normal file
|
|
@ -0,0 +1,212 @@
|
|||
/*
|
||||
* Copyright 2016 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_RC_DEFS_H__
|
||||
#define __MPP_RC_DEFS_H__
|
||||
|
||||
#include "rk_venc_ref.h"
|
||||
|
||||
#define MAX_CPB_REFS (8)
|
||||
|
||||
typedef enum EncFrmType_e {
|
||||
INTER_P_FRAME = 0,
|
||||
INTER_B_FRAME = 1,
|
||||
INTRA_FRAME = 2,
|
||||
INTER_VI_FRAME = 3,
|
||||
INTRA_RFH_FRAME = 4,
|
||||
} EncFrmType;
|
||||
|
||||
/*
|
||||
* EncFrmStatus controls record the encoding frame status and also control
|
||||
* work flow of encoder. It is the communicat channel between encoder implement
|
||||
* module, rate control module and hardware module.
|
||||
*
|
||||
* bit 0 ~ 31 frame status
|
||||
* 0 ~ 15 current frame status
|
||||
* 16 ~ 31 reference frame status
|
||||
* bit 32 ~ 63 encoding flow control
|
||||
*/
|
||||
typedef union EncFrmStatus_u {
|
||||
struct {
|
||||
/*
|
||||
* bit 0 ~ 31 frame status
|
||||
*/
|
||||
/* status flag */
|
||||
RK_U32 valid : 1;
|
||||
/*
|
||||
* 0 - write the reconstructed frame pixel to memory
|
||||
* 1 - do not write the reconstructed frame pixel to memory
|
||||
*/
|
||||
RK_U32 non_recn : 1;
|
||||
|
||||
/*
|
||||
* 0 - normal frame and normal dpb management
|
||||
* 1 - save recon frame as first pass extra frame. Used in two pass mode
|
||||
*/
|
||||
RK_U32 save_pass1 : 1;
|
||||
|
||||
/*
|
||||
* 0 - use normal input source frame as input
|
||||
* 1 - use the previously stored first pass recon frame as input frame
|
||||
*/
|
||||
RK_U32 use_pass1 : 1;
|
||||
|
||||
/* reference status flag */
|
||||
/*
|
||||
* 0 - inter frame
|
||||
* 1 - intra frame
|
||||
*/
|
||||
RK_U32 is_intra : 1;
|
||||
|
||||
/*
|
||||
* Valid when is_intra is true
|
||||
* 0 - normal intra frame
|
||||
* 1 - IDR frame
|
||||
*/
|
||||
RK_U32 is_idr : 1;
|
||||
|
||||
/*
|
||||
* 0 - mark as reference frame
|
||||
* 1 - mark as non-refernce frame
|
||||
*/
|
||||
RK_U32 is_non_ref : 1;
|
||||
|
||||
/*
|
||||
* Valid when is_non_ref is false
|
||||
* 0 - mark as short-term reference frame
|
||||
* 1 - mark as long-term refernce frame
|
||||
*/
|
||||
RK_U32 is_lt_ref : 1;
|
||||
|
||||
/* bit 8 - 15 */
|
||||
RK_U32 lt_idx : 4;
|
||||
RK_U32 temporal_id : 4;
|
||||
|
||||
/* distance between current frame and reference frame */
|
||||
MppEncRefMode ref_mode : 6;
|
||||
RK_S32 ref_arg : 8;
|
||||
RK_S32 ref_dist : 2;
|
||||
|
||||
/*
|
||||
* bit 32 ~ 63 encoder flow control flags
|
||||
*/
|
||||
/*
|
||||
* 0 - normal frame encoding
|
||||
* 1 - current frame will be dropped
|
||||
*/
|
||||
RK_U32 drop : 1;
|
||||
|
||||
/*
|
||||
* 0 - rate control module does not change frame type parameter
|
||||
* 1 - rate control module changes frame type parameter reencode is needed
|
||||
* to reprocess the dpb process. Also this means dpb module will follow
|
||||
* the frame status parameter provided by rate control module.
|
||||
*/
|
||||
RK_U32 re_dpb_proc : 1;
|
||||
|
||||
/*
|
||||
* 0 - current frame encoding is in normal flow
|
||||
* 1 - current frame encoding is in reencode flow
|
||||
*/
|
||||
RK_U32 reencode : 1;
|
||||
|
||||
/*
|
||||
* When true current frame size is super large then the frame should be reencoded.
|
||||
*/
|
||||
RK_U32 super_frame : 1;
|
||||
|
||||
/*
|
||||
* When true currnet frame is force to encoded as software skip frame
|
||||
*/
|
||||
RK_U32 force_pskip : 1;
|
||||
|
||||
/*
|
||||
* Current frame is intra refresh frame
|
||||
*/
|
||||
RK_U32 is_i_refresh : 1;
|
||||
/*
|
||||
* Current frame needs add recovery point prefix
|
||||
*/
|
||||
RK_U32 is_i_recovery : 1;
|
||||
RK_U32 reserved1 : 1;
|
||||
|
||||
/* reencode times */
|
||||
RK_U32 reencode_times : 8;
|
||||
|
||||
/* sequential index for each frame */
|
||||
RK_U32 seq_idx : 16;
|
||||
};
|
||||
RK_U64 val;
|
||||
} EncFrmStatus;
|
||||
|
||||
typedef struct EncCpbStatus_t {
|
||||
RK_S32 seq_idx;
|
||||
|
||||
EncFrmStatus curr;
|
||||
EncFrmStatus refr;
|
||||
|
||||
/* initial cpb status for current frame encoding */
|
||||
EncFrmStatus init[MAX_CPB_REFS];
|
||||
/* final cpb status after current frame encoding */
|
||||
EncFrmStatus final[MAX_CPB_REFS];
|
||||
} EncCpbStatus;
|
||||
|
||||
#define ENC_RC_FORCE_QP (0x00000001)
|
||||
|
||||
typedef struct EncRcForceCfg_t {
|
||||
RK_U32 force_flag;
|
||||
RK_S32 force_qp;
|
||||
RK_U32 reserve[6];
|
||||
} EncRcForceCfg;
|
||||
|
||||
/*
|
||||
* communication channel between rc / hal / hardware
|
||||
*
|
||||
* rc -> hal bit_target / bit_max / bit_min
|
||||
* hal -> hw quality_target / quality_max / quality_min
|
||||
* hw -> rc / hal bit_real / quality_real / madi / madp
|
||||
*/
|
||||
typedef struct EncRcCommonInfo_t {
|
||||
EncFrmType frame_type;
|
||||
|
||||
/* rc to hal */
|
||||
RK_S32 bit_target;
|
||||
RK_S32 bit_max;
|
||||
RK_S32 bit_min;
|
||||
|
||||
RK_S32 quality_target;
|
||||
RK_S32 quality_max;
|
||||
RK_S32 quality_min;
|
||||
|
||||
/* rc from hardware */
|
||||
RK_S32 bit_real;
|
||||
RK_S32 quality_real;
|
||||
RK_S32 madi;
|
||||
RK_S32 madp;
|
||||
|
||||
RK_U32 iblk4_prop; // scale 256
|
||||
RK_S32 reserve[15];
|
||||
} EncRcTaskInfo;
|
||||
|
||||
typedef struct EncRcTask_s {
|
||||
EncCpbStatus cpb;
|
||||
EncFrmStatus frm;
|
||||
EncRcTaskInfo info;
|
||||
EncRcForceCfg force;
|
||||
MppFrame frame;
|
||||
} EncRcTask;
|
||||
|
||||
#endif /* __MPP_RC_DEFS_H__ */
|
||||
237
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_task.h
vendored
Normal file
237
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/mpp_task.h
vendored
Normal file
|
|
@ -0,0 +1,237 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __MPP_TASK_H__
|
||||
#define __MPP_TASK_H__
|
||||
|
||||
#include "mpp_meta.h"
|
||||
|
||||
/*
|
||||
* Advanced task flow
|
||||
* Advanced task flow introduces three concepts: port, task and item
|
||||
*
|
||||
* Port is from OpenMAX
|
||||
* Port has two type: input port and output port which are all for data transaction.
|
||||
* Port work like a queue. task will be dequeue from or enqueue to one port.
|
||||
* On input side user will dequeue task from input port, setup task and enqueue task
|
||||
* back to input port.
|
||||
* On output side user will dequeue task from output port, get the information from
|
||||
* and then enqueue task back to output port.
|
||||
*
|
||||
* Task indicates one transaction on the port.
|
||||
* Task has two working mode: async mode and sync mode
|
||||
* If mpp is work in sync mode on task enqueue function return the task has been done
|
||||
* If mpp is work in async mode on task enqueue function return the task is just put
|
||||
* on the task queue for process.
|
||||
* Task can carry different items. Task just like a container of items
|
||||
*
|
||||
* Item indicates MppPacket or MppFrame which is contained in one task
|
||||
*/
|
||||
|
||||
/*
|
||||
* One mpp task queue has two ports: input and output
|
||||
*
|
||||
* The whole picture is:
|
||||
* Top layer mpp has two ports: mpp_input_port and mpp_output_port
|
||||
* But internally these two ports belongs to two task queue.
|
||||
* The mpp_input_port is the mpp_input_task_queue's input port.
|
||||
* The mpp_output_port is the mpp_output_task_queue's output port.
|
||||
*
|
||||
* Each port uses its task queue to communication
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_PORT_INPUT,
|
||||
MPP_PORT_OUTPUT,
|
||||
MPP_PORT_BUTT,
|
||||
} MppPortType;
|
||||
|
||||
/*
|
||||
* Advance task work flow mode:
|
||||
******************************************************************************
|
||||
* 1. async mode (default_val)
|
||||
*
|
||||
* mpp_init(type, coding, MPP_WORK_ASYNC)
|
||||
*
|
||||
* input thread
|
||||
* a - poll(input)
|
||||
* b - dequeue(input, *task)
|
||||
* c - task_set_item(packet/frame)
|
||||
* d - enqueue(input, task) // when enqueue return the task is not done yet
|
||||
*
|
||||
* output thread
|
||||
* a - poll(output)
|
||||
* b - dequeue(output, *task)
|
||||
* c - task_get_item(frame/packet)
|
||||
* d - enqueue(output, task)
|
||||
******************************************************************************
|
||||
* 2. sync mode
|
||||
*
|
||||
* mpp_init(type, coding, MPP_WORK_SYNC)
|
||||
*
|
||||
* a - poll(input)
|
||||
* b - dequeue(input, *task)
|
||||
* c - task_set_item(packet/frame)
|
||||
* d - enqueue(task) // when enqueue return the task is finished
|
||||
******************************************************************************
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_TASK_ASYNC,
|
||||
MPP_TASK_SYNC,
|
||||
MPP_TASK_WORK_MODE_BUTT,
|
||||
} MppTaskWorkMode;
|
||||
|
||||
/*
|
||||
* Mpp port poll type
|
||||
*
|
||||
* MPP_POLL_BLOCK - for block poll
|
||||
* MPP_POLL_NON_BLOCK - for non-block poll
|
||||
* small than MPP_POLL_MAX - for poll with timeout in ms
|
||||
* small than MPP_POLL_BUTT or larger than MPP_POLL_MAX is invalid value
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_POLL_BUTT = -2,
|
||||
MPP_POLL_BLOCK = -1,
|
||||
MPP_POLL_NON_BLOCK = 0,
|
||||
MPP_POLL_MAX = 8000,
|
||||
} MppPollType;
|
||||
|
||||
/*
|
||||
* Mpp timeout define
|
||||
* MPP_TIMEOUT_BLOCK - for block poll
|
||||
* MPP_TIMEOUT_NON_BLOCK - for non-block poll
|
||||
* small than MPP_TIMEOUT_MAX - for poll with timeout in ms
|
||||
* small than MPP_TIMEOUT_BUTT or larger than MPP_TIMEOUT_MAX is invalid value
|
||||
*/
|
||||
#define MPP_TIMEOUT_BUTT (-2L)
|
||||
#define MPP_TIMEOUT_BLOCK (-1L)
|
||||
#define MPP_TIMEOUT_NON_BLOCK (0L)
|
||||
#define MPP_TIMEOUT_MAX (8000L)
|
||||
|
||||
/*
|
||||
* MppTask is descriptor of a task which send to mpp for process
|
||||
* mpp can support different type of work mode, for example:
|
||||
*
|
||||
* decoder:
|
||||
*
|
||||
* 1. typical decoder mode:
|
||||
* input - MppPacket (normal cpu buffer, need cpu copy)
|
||||
* output - MppFrame (ion/drm buffer in external/internal mode)
|
||||
* 2. secure decoder mode:
|
||||
* input - MppPacket (externel ion/drm buffer, cpu can not access)
|
||||
* output - MppFrame (ion/drm buffer in external/internal mode, cpu can not access)
|
||||
*
|
||||
* interface usage:
|
||||
*
|
||||
* typical flow
|
||||
* input side:
|
||||
* task_dequeue(ctx, PORT_INPUT, &task);
|
||||
* task_put_item(task, MODE_INPUT, packet)
|
||||
* task_enqueue(ctx, PORT_INPUT, task);
|
||||
* output side:
|
||||
* task_dequeue(ctx, PORT_OUTPUT, &task);
|
||||
* task_get_item(task, MODE_OUTPUT, &frame)
|
||||
* task_enqueue(ctx, PORT_OUTPUT, task);
|
||||
*
|
||||
* secure flow
|
||||
* input side:
|
||||
* task_dequeue(ctx, PORT_INPUT, &task);
|
||||
* task_put_item(task, MODE_INPUT, packet)
|
||||
* task_put_item(task, MODE_OUTPUT, frame) // buffer will be specified here
|
||||
* task_enqueue(ctx, PORT_INPUT, task);
|
||||
* output side:
|
||||
* task_dequeue(ctx, PORT_OUTPUT, &task);
|
||||
* task_get_item(task, MODE_OUTPUT, &frame)
|
||||
* task_enqueue(ctx, PORT_OUTPUT, task);
|
||||
*
|
||||
* encoder:
|
||||
*
|
||||
* 1. typical encoder mode:
|
||||
* input - MppFrame (ion/drm buffer in external mode)
|
||||
* output - MppPacket (normal cpu buffer, need cpu copy)
|
||||
* 2. user input encoder mode:
|
||||
* input - MppFrame (normal cpu buffer, need to build hardware table for this buffer)
|
||||
* output - MppPacket (normal cpu buffer, need cpu copy)
|
||||
* 3. secure encoder mode:
|
||||
* input - MppFrame (ion/drm buffer in external mode, cpu can not access)
|
||||
* output - MppPacket (externel ion/drm buffer, cpu can not access)
|
||||
*
|
||||
* typical / user input flow
|
||||
* input side:
|
||||
* task_dequeue(ctx, PORT_INPUT, &task);
|
||||
* task_put_item(task, MODE_INPUT, frame)
|
||||
* task_enqueue(ctx, PORT_INPUT, task);
|
||||
* output side:
|
||||
* task_dequeue(ctx, PORT_OUTPUT, &task);
|
||||
* task_get_item(task, MODE_OUTPUT, &packet)
|
||||
* task_enqueue(ctx, PORT_OUTPUT, task);
|
||||
*
|
||||
* secure flow
|
||||
* input side:
|
||||
* task_dequeue(ctx, PORT_INPUT, &task);
|
||||
* task_put_item(task, MODE_OUTPUT, packet) // buffer will be specified here
|
||||
* task_put_item(task, MODE_INPUT, frame)
|
||||
* task_enqueue(ctx, PORT_INPUT, task);
|
||||
* output side:
|
||||
* task_dequeue(ctx, PORT_OUTPUT, &task);
|
||||
* task_get_item(task, MODE_OUTPUT, &packet)
|
||||
* task_get_item(task, MODE_OUTPUT, &frame)
|
||||
* task_enqueue(ctx, PORT_OUTPUT, task);
|
||||
*
|
||||
* NOTE: this flow can specify the output frame. User will setup both intput frame and output packet
|
||||
* buffer at the input side. Then at output side when user gets a finished task user can get the output
|
||||
* packet and corresponding released input frame.
|
||||
*
|
||||
* image processing
|
||||
*
|
||||
* 1. typical image process mode:
|
||||
* input - MppFrame (ion/drm buffer in external mode)
|
||||
* output - MppFrame (ion/drm buffer in external mode)
|
||||
*
|
||||
* typical / user input flow
|
||||
* input side:
|
||||
* task_dequeue(ctx, PORT_INPUT, &task);
|
||||
* task_put_item(task, MODE_INPUT, frame)
|
||||
* task_enqueue(ctx, PORT_INPUT, task);
|
||||
* output side:
|
||||
* task_dequeue(ctx, PORT_OUTPUT, &task);
|
||||
* task_get_item(task, MODE_OUTPUT, &frame)
|
||||
* task_enqueue(ctx, PORT_OUTPUT, task);
|
||||
*/
|
||||
/* NOTE: use index rather then handle to descripbe task */
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
MPP_RET mpp_task_meta_set_s32(MppTask task, MppMetaKey key, RK_S32 val);
|
||||
MPP_RET mpp_task_meta_set_s64(MppTask task, MppMetaKey key, RK_S64 val);
|
||||
MPP_RET mpp_task_meta_set_ptr(MppTask task, MppMetaKey key, void *val);
|
||||
MPP_RET mpp_task_meta_set_frame (MppTask task, MppMetaKey key, MppFrame frame);
|
||||
MPP_RET mpp_task_meta_set_packet(MppTask task, MppMetaKey key, MppPacket packet);
|
||||
MPP_RET mpp_task_meta_set_buffer(MppTask task, MppMetaKey key, MppBuffer buffer);
|
||||
|
||||
MPP_RET mpp_task_meta_get_s32(MppTask task, MppMetaKey key, RK_S32 *val, RK_S32 default_val);
|
||||
MPP_RET mpp_task_meta_get_s64(MppTask task, MppMetaKey key, RK_S64 *val, RK_S64 default_val);
|
||||
MPP_RET mpp_task_meta_get_ptr(MppTask task, MppMetaKey key, void **val, void *default_val);
|
||||
MPP_RET mpp_task_meta_get_frame (MppTask task, MppMetaKey key, MppFrame *frame);
|
||||
MPP_RET mpp_task_meta_get_packet(MppTask task, MppMetaKey key, MppPacket *packet);
|
||||
MPP_RET mpp_task_meta_get_buffer(MppTask task, MppMetaKey key, MppBuffer *buffer);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__MPP_QUEUE_H__*/
|
||||
118
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_hdr_meta_com.h
vendored
Normal file
118
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_hdr_meta_com.h
vendored
Normal file
|
|
@ -0,0 +1,118 @@
|
|||
/*
|
||||
* Copyright 2022 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __RK_HDR_META_COM_H__
|
||||
#define __RK_HDR_META_COM_H__
|
||||
|
||||
#include "rk_type.h"
|
||||
|
||||
typedef enum HdrCodecType_e {
|
||||
HDR_AVS2 = 0,
|
||||
HDR_HEVC = 1,
|
||||
HDR_H264 = 2,
|
||||
HDR_AV1 = 3,
|
||||
HDR_CODEC_BUT,
|
||||
} HdrCodecType;
|
||||
|
||||
typedef enum HdrFormat_e {
|
||||
HDR_NONE = 0,
|
||||
HDR10 = 1,
|
||||
HLG = 2,
|
||||
// RESERVED3 = 3, //reserved for more future static hdr format
|
||||
// RESERVED4 = 4, //reserved for more future static hdr format
|
||||
HDRVIVID = 5,
|
||||
// RESERVED6 = 6, //reserved for hdr vivid
|
||||
// RESERVED7 = 7, //reserved for hdr vivid
|
||||
HDR10PLUS = 8,
|
||||
// RESERVED9 = 9, //reserved for hdr10+
|
||||
// RESERVED10 = 10,//reserved for hdr10+
|
||||
DOLBY = 11,
|
||||
// RESERVED12 = 12, //reserved for other dynamic hdr format
|
||||
// RESERVED13 = 13, //reserved for other dynamic hdr format
|
||||
HDR_FORMAT_MAX,
|
||||
} HdrFormat;
|
||||
|
||||
typedef enum HdrPayloadFormat_e {
|
||||
STATIC = 0,
|
||||
DYNAMIC = 1,
|
||||
HDR_PAYLOAD_FORMAT_MAX,
|
||||
} HdrPayloadFormat;
|
||||
|
||||
typedef struct HdrStaticMeta_t {
|
||||
RK_U32 color_space;
|
||||
RK_U32 color_primaries;
|
||||
RK_U32 color_trc;
|
||||
RK_U32 red_x;
|
||||
RK_U32 red_y;
|
||||
RK_U32 green_x;
|
||||
RK_U32 green_y;
|
||||
RK_U32 blue_x;
|
||||
RK_U32 blue_y;
|
||||
RK_U32 white_point_x;
|
||||
RK_U32 white_point_y;
|
||||
RK_U32 min_luminance;
|
||||
RK_U32 max_luminance;
|
||||
RK_U32 max_cll;
|
||||
RK_U32 max_fall;
|
||||
RK_U32 reserved[4];
|
||||
} HdrStaticMeta;
|
||||
|
||||
/*
|
||||
* HDR metadata format from codec
|
||||
*
|
||||
* +----------+
|
||||
* | header1 |
|
||||
* +----------+
|
||||
* | |
|
||||
* | payload |
|
||||
* | |
|
||||
* +----------+
|
||||
* | header2 |
|
||||
* +----------+
|
||||
* | |
|
||||
* | payload |
|
||||
* | |
|
||||
* +----------+
|
||||
* | header3 |
|
||||
* +----------+
|
||||
* | |
|
||||
* | payload |
|
||||
* | |
|
||||
* +----------+
|
||||
*/
|
||||
typedef struct RkMetaHdrHeader_t {
|
||||
/* For transmission */
|
||||
RK_U16 magic; /* magic word for checking overwrite error */
|
||||
RK_U16 size; /* total header+payload length including header */
|
||||
RK_U16 message_total; /* total message count in current transmission */
|
||||
RK_U16 message_index; /* current message index in the transmission */
|
||||
|
||||
/* For payload identification */
|
||||
RK_U16 version; /* payload structure version */
|
||||
RK_U16 hdr_format; /* HDR protocol: HDR10, HLG, Dolby, HDRVivid ... */
|
||||
RK_U16 hdr_payload_type; /* HDR data type: static data, dynamic data ... */
|
||||
RK_U16 video_format; /* video format: H.264, H.265, AVS2 ... */
|
||||
|
||||
/* For extenstion usage */
|
||||
RK_U32 reserve[4];
|
||||
|
||||
/* payload data aligned to 32bits */
|
||||
RK_U32 payload[];
|
||||
} RkMetaHdrHeader;
|
||||
|
||||
void fill_hdr_meta_to_frame(MppFrame frame, HdrCodecType codec_type);
|
||||
|
||||
#endif
|
||||
274
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_mpi.h
vendored
Normal file
274
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_mpi.h
vendored
Normal file
|
|
@ -0,0 +1,274 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __RK_MPI_H__
|
||||
#define __RK_MPI_H__
|
||||
|
||||
/**
|
||||
* @addtogroup rk_mpi
|
||||
* @brief Rockchip Media Process Interface
|
||||
* @details Media Process Platform(MPP) provides application programming
|
||||
* interface for the application layer, by which applications can
|
||||
* call hardware encode and decode. Current MPP fully supports
|
||||
* chipset RK3288/RK3228/RK3229/RK3399/RK3328/RV1108. Old chipset
|
||||
* like RK29xx/RK30xx/RK31XX/RK3368 is partly supported due to lack
|
||||
* of some hardware register generation module.
|
||||
*/
|
||||
|
||||
#include "rk_mpi_cmd.h"
|
||||
#include "mpp_task.h"
|
||||
|
||||
/**
|
||||
* @ingroup rk_mpi
|
||||
* @brief MPP main work function set
|
||||
* @details all api function are seperated into two sets: data io api set
|
||||
* and control api set
|
||||
*
|
||||
* (1). the data api set is for data input/output flow including:
|
||||
*
|
||||
* (1.1) simple data api set:
|
||||
*
|
||||
* decode : both send video stream packet to decoder and get video frame from
|
||||
* decoder at the same time.
|
||||
*
|
||||
* encode : both send video frame to encoder and get encoded video stream from
|
||||
* encoder at the same time.
|
||||
*
|
||||
* decode_put_packet: send video stream packet to decoder only, async interface
|
||||
*
|
||||
* decode_get_frame : get video frame from decoder only, async interface
|
||||
*
|
||||
* encode_put_frame : send video frame to encoder only, async interface
|
||||
*
|
||||
* encode_get_packet: get encoded video packet from encoder only, async interface
|
||||
*
|
||||
* (1.2) advanced task api set:
|
||||
*
|
||||
* poll : poll port for dequeue
|
||||
*
|
||||
* dequeue : pop a task from mpp task queue
|
||||
*
|
||||
* enqueue : push a task to mpp task queue
|
||||
*
|
||||
* (2). the control api set is for mpp context control including:
|
||||
*
|
||||
* control : similiar to ioctl in kernel driver, setup or get mpp internal parameter
|
||||
*
|
||||
* reset : clear all data in mpp context, discard all packet and frame,
|
||||
* reset all components to initialized status
|
||||
*/
|
||||
typedef struct MppApi_t {
|
||||
/**
|
||||
* @brief size of struct MppApi
|
||||
*/
|
||||
RK_U32 size;
|
||||
/**
|
||||
* @brief mpp api version, generated by Git
|
||||
*/
|
||||
RK_U32 version;
|
||||
|
||||
// simple data flow interface
|
||||
/**
|
||||
* @brief both send video stream packet to decoder and get video frame from
|
||||
* decoder at the same time
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @param[in] packet The input video stream, its usage can refer mpp_packet.h.
|
||||
* @param[out] frame The output picture, its usage can refer mpp_frame.h.
|
||||
* @return 0 and positive for success, negative for failure. The return
|
||||
* value is an error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*decode)(MppCtx ctx, MppPacket packet, MppFrame *frame);
|
||||
/**
|
||||
* @brief send video stream packet to decoder only, async interface
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @param[in] packet The input video stream, its usage can refer mpp_packet.h.
|
||||
* @return 0 and positive for success, negative for failure. The return
|
||||
* value is an error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*decode_put_packet)(MppCtx ctx, MppPacket packet);
|
||||
/**
|
||||
* @brief get video frame from decoder only, async interface
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @param[out] frame The output picture, its usage can refer mpp_frame.h.
|
||||
* @return 0 and positive for success, negative for failure. The return
|
||||
* value is an error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*decode_get_frame)(MppCtx ctx, MppFrame *frame);
|
||||
/**
|
||||
* @brief both send video frame to encoder and get encoded video stream from
|
||||
* encoder at the same time
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @param[in] frame The input video data, its usage can refer mpp_frame.h.
|
||||
* @param[out] packet The output compressed data, its usage can refer mpp_packet.h.
|
||||
* @return 0 and positive for success, negative for failure. The return
|
||||
* value is an error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*encode)(MppCtx ctx, MppFrame frame, MppPacket *packet);
|
||||
/**
|
||||
* @brief send video frame to encoder only, async interface
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @param[in] frame The input video data, its usage can refer mpp_frame.h.
|
||||
* @return 0 and positive for success, negative for failure. The return
|
||||
* value is an error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*encode_put_frame)(MppCtx ctx, MppFrame frame);
|
||||
/**
|
||||
* @brief get encoded video packet from encoder only, async interface
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @param[out] packet The output compressed data, its usage can refer mpp_packet.h.
|
||||
* @return 0 and positive for success, negative for failure. The return
|
||||
* value is an error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*encode_get_packet)(MppCtx ctx, MppPacket *packet);
|
||||
|
||||
/**
|
||||
* @brief ISP interface, will be supported in the future.
|
||||
*/
|
||||
MPP_RET (*isp)(MppCtx ctx, MppFrame dst, MppFrame src);
|
||||
/**
|
||||
* @brief ISP interface, will be supported in the future.
|
||||
*/
|
||||
MPP_RET (*isp_put_frame)(MppCtx ctx, MppFrame frame);
|
||||
/**
|
||||
* @brief ISP interface, will be supported in the future.
|
||||
*/
|
||||
MPP_RET (*isp_get_frame)(MppCtx ctx, MppFrame *frame);
|
||||
|
||||
// advance data flow interface
|
||||
/**
|
||||
* @brief poll port for dequeue
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @param[in] type input port or output port which are both for data transaction
|
||||
* @param[in] timeout mpp poll type, its usage can refer mpp_task.h.
|
||||
* @return 0 and positive for success, negative for failure. The return
|
||||
* value is an error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*poll)(MppCtx ctx, MppPortType type, MppPollType timeout);
|
||||
/**
|
||||
* @brief dequeue MppTask, pop a task from mpp task queue
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @param[in] type input port or output port which are both for data transaction
|
||||
* @param[out] task MppTask popped from mpp task queue, its usage can refer mpp_task.h.
|
||||
* @return 0 and positive for success, negative for failure. The return
|
||||
* value is an error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*dequeue)(MppCtx ctx, MppPortType type, MppTask *task);
|
||||
/**
|
||||
* @brief enqueue MppTask, push a task to mpp task queue
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @param[in] type input port or output port which are both for data transaction
|
||||
* @param[in] task MppTask which is sent to mpp for process, its usage can refer mpp_task.h.
|
||||
* @return 0 and positive for success, negative for failure. The return
|
||||
* value is an error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*enqueue)(MppCtx ctx, MppPortType type, MppTask task);
|
||||
|
||||
// control interface
|
||||
/**
|
||||
* @brief discard all packet and frame, reset all component,
|
||||
* for both decoder and encoder
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @return 0 for success, others for failure. The return value is an
|
||||
* error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*reset)(MppCtx ctx);
|
||||
/**
|
||||
* @brief control function for mpp property setting
|
||||
* @param[in] ctx The context of mpp, created by mpp_create() and initiated
|
||||
* by mpp_init().
|
||||
* @param[in] cmd The mpi command, its definition can refer rk_mpi_cmd.h.
|
||||
* @param[in,out] param The mpi command parameter
|
||||
* @return 0 for success, others for failure. The return value is an
|
||||
* error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET (*control)(MppCtx ctx, MpiCmd cmd, MppParam param);
|
||||
|
||||
/**
|
||||
* @brief The reserved segment, may be used in the future
|
||||
*/
|
||||
RK_U32 reserv[16];
|
||||
} MppApi;
|
||||
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
/**
|
||||
* @ingroup rk_mpi
|
||||
* @brief Create empty context structure and mpi function pointers.
|
||||
* Use functions in MppApi to access mpp services.
|
||||
* @param[in,out] ctx pointer of the mpp context, refer to MpiImpl_t.
|
||||
* @param[in,out] mpi pointer of mpi function, refer to MppApi.
|
||||
* @return 0 for success, others for failure. The return value is an
|
||||
* error code. For details, please refer mpp_err.h.
|
||||
* @note This interface creates base flow context, all function calls
|
||||
* are based on it.
|
||||
*/
|
||||
MPP_RET mpp_create(MppCtx *ctx, MppApi **mpi);
|
||||
/**
|
||||
* @ingroup rk_mpi
|
||||
* @brief Call after mpp_create to setup mpp type and video format.
|
||||
* This function will call internal context init function.
|
||||
* @param[in] ctx The context of mpp, created by mpp_create().
|
||||
* @param[in] type specify decoder or encoder, refer to MppCtxType.
|
||||
* @param[in] coding specify video compression coding, refer to MppCodingType.
|
||||
* @return 0 for success, others for failure. The return value is an
|
||||
* error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET mpp_init(MppCtx ctx, MppCtxType type, MppCodingType coding);
|
||||
/**
|
||||
* @ingroup rk_mpi
|
||||
* @brief Destroy mpp context and free both context and mpi structure,
|
||||
* it matches with mpp_init().
|
||||
* @param[in] ctx The context of mpp, created by mpp_create().
|
||||
* @return 0 for success, others for failure. The return value is an
|
||||
* error code. For details, please refer mpp_err.h.
|
||||
*/
|
||||
MPP_RET mpp_destroy(MppCtx ctx);
|
||||
/**
|
||||
* @ingroup rk_mpi
|
||||
* @brief judge given format is supported or not by MPP.
|
||||
* @param[in] type specify decoder or encoder, refer to MppCtxType.
|
||||
* @param[in] coding specify video compression coding, refer to MppCodingType.
|
||||
* @return 0 for support, -1 for unsupported.
|
||||
*/
|
||||
MPP_RET mpp_check_support_format(MppCtxType type, MppCodingType coding);
|
||||
/**
|
||||
* @ingroup rk_mpi
|
||||
* @brief List all formats supported by MPP
|
||||
* @param NULL no need to input parameter
|
||||
* @return No return value. This function just prints format information supported
|
||||
* by MPP on standard output.
|
||||
*/
|
||||
void mpp_show_support_format(void);
|
||||
void mpp_show_color_format(void);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__RK_MPI_H__*/
|
||||
209
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_mpi_cmd.h
vendored
Normal file
209
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_mpi_cmd.h
vendored
Normal file
|
|
@ -0,0 +1,209 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __RK_MPI_CMD_H__
|
||||
#define __RK_MPI_CMD_H__
|
||||
|
||||
/*
|
||||
* Command id bit usage is defined as follows:
|
||||
* bit 20 - 23 - module id
|
||||
* bit 16 - 19 - contex id
|
||||
* bit 0 - 15 - command id
|
||||
*/
|
||||
#define CMD_MODULE_ID_MASK (0x00F00000)
|
||||
#define CMD_MODULE_OSAL (0x00100000)
|
||||
#define CMD_MODULE_MPP (0x00200000)
|
||||
#define CMD_MODULE_CODEC (0x00300000)
|
||||
#define CMD_MODULE_HAL (0x00400000)
|
||||
|
||||
#define CMD_CTX_ID_MASK (0x000F0000)
|
||||
#define CMD_CTX_ID_DEC (0x00010000)
|
||||
#define CMD_CTX_ID_ENC (0x00020000)
|
||||
#define CMD_CTX_ID_ISP (0x00030000)
|
||||
|
||||
/* separate encoder / decoder control command to different segment */
|
||||
#define CMD_CFG_ID_MASK (0x0000FF00)
|
||||
|
||||
/* mpp status control command */
|
||||
#define CMD_STATE_OPS (0x00000100)
|
||||
|
||||
/* decoder control command */
|
||||
#define CMD_DEC_CFG_ALL (0x00000000)
|
||||
#define CMD_DEC_QUERY (0x00000100)
|
||||
#define CMD_DEC_CFG (0x00000200)
|
||||
|
||||
/* encoder control command */
|
||||
#define CMD_ENC_CFG_ALL (0x00000000)
|
||||
#define CMD_ENC_QUERY (0x00000100)
|
||||
#define CMD_ENC_CFG_RC_API (0x00000200)
|
||||
|
||||
#define CMD_ENC_CFG_MISC (0x00008000)
|
||||
#define CMD_ENC_CFG_SPLIT (0x00008100)
|
||||
#define CMD_ENC_CFG_REF (0x00008200)
|
||||
#define CMD_ENC_CFG_ROI (0x00008300)
|
||||
#define CMD_ENC_CFG_OSD (0x00008400)
|
||||
|
||||
typedef enum {
|
||||
MPP_OSAL_CMD_BASE = CMD_MODULE_OSAL,
|
||||
MPP_OSAL_CMD_END,
|
||||
|
||||
MPP_CMD_BASE = CMD_MODULE_MPP,
|
||||
MPP_ENABLE_DEINTERLACE,
|
||||
MPP_SET_INPUT_BLOCK, /* deprecated */
|
||||
MPP_SET_INTPUT_BLOCK_TIMEOUT, /* deprecated */
|
||||
MPP_SET_OUTPUT_BLOCK, /* deprecated */
|
||||
MPP_SET_OUTPUT_BLOCK_TIMEOUT, /* deprecated */
|
||||
/*
|
||||
* timeout setup, refer to MPP_TIMEOUT_XXX
|
||||
* zero - non block
|
||||
* negative - block with no timeout
|
||||
* positive - timeout in milisecond
|
||||
*/
|
||||
MPP_SET_INPUT_TIMEOUT, /* parameter type RK_S64 */
|
||||
MPP_SET_OUTPUT_TIMEOUT, /* parameter type RK_S64 */
|
||||
MPP_SET_DISABLE_THREAD, /* MPP no thread mode and use external thread to decode */
|
||||
|
||||
MPP_STATE_CMD_BASE = CMD_MODULE_MPP | CMD_STATE_OPS,
|
||||
MPP_START,
|
||||
MPP_STOP,
|
||||
MPP_PAUSE,
|
||||
MPP_RESUME,
|
||||
|
||||
MPP_CMD_END,
|
||||
|
||||
MPP_CODEC_CMD_BASE = CMD_MODULE_CODEC,
|
||||
MPP_CODEC_GET_FRAME_INFO,
|
||||
MPP_CODEC_CMD_END,
|
||||
|
||||
MPP_DEC_CMD_BASE = CMD_MODULE_CODEC | CMD_CTX_ID_DEC,
|
||||
MPP_DEC_SET_FRAME_INFO, /* vpu api legacy control for buffer slot dimension init */
|
||||
MPP_DEC_SET_EXT_BUF_GROUP, /* IMPORTANT: set external buffer group to mpp decoder */
|
||||
MPP_DEC_SET_INFO_CHANGE_READY,
|
||||
MPP_DEC_SET_PRESENT_TIME_ORDER, /* use input time order for output */
|
||||
MPP_DEC_SET_PARSER_SPLIT_MODE, /* Need to setup before init */
|
||||
MPP_DEC_SET_PARSER_FAST_MODE, /* Need to setup before init */
|
||||
MPP_DEC_GET_STREAM_COUNT,
|
||||
MPP_DEC_GET_VPUMEM_USED_COUNT,
|
||||
MPP_DEC_SET_VC1_EXTRA_DATA,
|
||||
MPP_DEC_SET_OUTPUT_FORMAT,
|
||||
MPP_DEC_SET_DISABLE_ERROR, /* When set it will disable sw/hw error (H.264 / H.265) */
|
||||
MPP_DEC_SET_IMMEDIATE_OUT,
|
||||
MPP_DEC_SET_ENABLE_DEINTERLACE, /* MPP enable deinterlace by default. Vpuapi can disable it */
|
||||
MPP_DEC_SET_ENABLE_FAST_PLAY, /* enable idr output immediately */
|
||||
MPP_DEC_SET_DISABLE_THREAD, /* MPP no thread mode and use external thread to decode */
|
||||
MPP_DEC_SET_MAX_USE_BUFFER_SIZE,
|
||||
MPP_DEC_SET_ENABLE_MVC, /* enable MVC decoding*/
|
||||
|
||||
MPP_DEC_CMD_QUERY = CMD_MODULE_CODEC | CMD_CTX_ID_DEC | CMD_DEC_QUERY,
|
||||
/* query decoder runtime information for decode stage */
|
||||
MPP_DEC_QUERY, /* set and get MppDecQueryCfg structure */
|
||||
|
||||
CMD_DEC_CMD_CFG = CMD_MODULE_CODEC | CMD_CTX_ID_DEC | CMD_DEC_CFG,
|
||||
MPP_DEC_SET_CFG, /* set MppDecCfg structure */
|
||||
MPP_DEC_GET_CFG, /* get MppDecCfg structure */
|
||||
|
||||
MPP_DEC_CMD_END,
|
||||
|
||||
MPP_ENC_CMD_BASE = CMD_MODULE_CODEC | CMD_CTX_ID_ENC,
|
||||
/* basic encoder setup control */
|
||||
MPP_ENC_SET_CFG, /* set MppEncCfg structure */
|
||||
MPP_ENC_GET_CFG, /* get MppEncCfg structure */
|
||||
MPP_ENC_SET_PREP_CFG, /* deprecated set MppEncPrepCfg structure, use MPP_ENC_SET_CFG instead */
|
||||
MPP_ENC_GET_PREP_CFG, /* deprecated get MppEncPrepCfg structure, use MPP_ENC_GET_CFG instead */
|
||||
MPP_ENC_SET_RC_CFG, /* deprecated set MppEncRcCfg structure, use MPP_ENC_SET_CFG instead */
|
||||
MPP_ENC_GET_RC_CFG, /* deprecated get MppEncRcCfg structure, use MPP_ENC_GET_CFG instead */
|
||||
MPP_ENC_SET_CODEC_CFG, /* deprecated set MppEncCodecCfg structure, use MPP_ENC_SET_CFG instead */
|
||||
MPP_ENC_GET_CODEC_CFG, /* deprecated get MppEncCodecCfg structure, use MPP_ENC_GET_CFG instead */
|
||||
/* runtime encoder setup control */
|
||||
MPP_ENC_SET_IDR_FRAME, /* next frame will be encoded as intra frame */
|
||||
MPP_ENC_SET_OSD_LEGACY_0, /* deprecated */
|
||||
MPP_ENC_SET_OSD_LEGACY_1, /* deprecated */
|
||||
MPP_ENC_SET_OSD_LEGACY_2, /* deprecated */
|
||||
MPP_ENC_GET_HDR_SYNC, /* get vps / sps / pps which has better sync behavior parameter is MppPacket */
|
||||
MPP_ENC_GET_EXTRA_INFO, /* deprecated */
|
||||
MPP_ENC_SET_SEI_CFG, /* SEI: Supplement Enhancemant Information, parameter is MppSeiMode */
|
||||
MPP_ENC_GET_SEI_DATA, /* SEI: Supplement Enhancemant Information, parameter is MppPacket */
|
||||
MPP_ENC_PRE_ALLOC_BUFF, /* deprecated */
|
||||
MPP_ENC_SET_QP_RANGE, /* used for adjusting qp range, the parameter can be 1 or 2 */
|
||||
MPP_ENC_SET_ROI_CFG, /* set MppEncROICfg structure */
|
||||
MPP_ENC_SET_CTU_QP, /* for H265 Encoder,set CTU's size and QP */
|
||||
|
||||
MPP_ENC_CMD_QUERY = CMD_MODULE_CODEC | CMD_CTX_ID_ENC | CMD_ENC_QUERY,
|
||||
/* query encoder runtime information for encode stage */
|
||||
MPP_ENC_QUERY, /* set and get MppEncQueryCfg structure */
|
||||
|
||||
/* User define rate control stategy API control */
|
||||
MPP_ENC_CFG_RC_API = CMD_MODULE_CODEC | CMD_CTX_ID_ENC | CMD_ENC_CFG_RC_API,
|
||||
/*
|
||||
* Get RcApiQueryAll structure
|
||||
* Get all available rate control stategy string and count
|
||||
*/
|
||||
MPP_ENC_GET_RC_API_ALL = MPP_ENC_CFG_RC_API + 1,
|
||||
/*
|
||||
* Get RcApiQueryType structure
|
||||
* Get available rate control stategy string with certain type
|
||||
*/
|
||||
MPP_ENC_GET_RC_API_BY_TYPE = MPP_ENC_CFG_RC_API + 2,
|
||||
/*
|
||||
* Set RcImplApi structure
|
||||
* Add new or update rate control stategy function pointers
|
||||
*/
|
||||
MPP_ENC_SET_RC_API_CFG = MPP_ENC_CFG_RC_API + 3,
|
||||
/*
|
||||
* Get RcApiBrief structure
|
||||
* Get current used rate control stategy brief information (type and name)
|
||||
*/
|
||||
MPP_ENC_GET_RC_API_CURRENT = MPP_ENC_CFG_RC_API + 4,
|
||||
/*
|
||||
* Set RcApiBrief structure
|
||||
* Set current used rate control stategy brief information (type and name)
|
||||
*/
|
||||
MPP_ENC_SET_RC_API_CURRENT = MPP_ENC_CFG_RC_API + 5,
|
||||
|
||||
MPP_ENC_CFG_MISC = CMD_MODULE_CODEC | CMD_CTX_ID_ENC | CMD_ENC_CFG_MISC,
|
||||
MPP_ENC_SET_HEADER_MODE, /* set MppEncHeaderMode */
|
||||
MPP_ENC_GET_HEADER_MODE, /* get MppEncHeaderMode */
|
||||
|
||||
MPP_ENC_CFG_SPLIT = CMD_MODULE_CODEC | CMD_CTX_ID_ENC | CMD_ENC_CFG_SPLIT,
|
||||
MPP_ENC_SET_SPLIT, /* set MppEncSliceSplit structure */
|
||||
MPP_ENC_GET_SPLIT, /* get MppEncSliceSplit structure */
|
||||
|
||||
MPP_ENC_CFG_REF = CMD_MODULE_CODEC | CMD_CTX_ID_ENC | CMD_ENC_CFG_REF,
|
||||
MPP_ENC_SET_REF_CFG, /* set MppEncRefCfg structure */
|
||||
|
||||
MPP_ENC_CFG_OSD = CMD_MODULE_CODEC | CMD_CTX_ID_ENC | CMD_ENC_CFG_OSD,
|
||||
MPP_ENC_SET_OSD_PLT_CFG, /* set OSD palette, parameter should be pointer to MppEncOSDPltCfg */
|
||||
MPP_ENC_GET_OSD_PLT_CFG, /* get OSD palette, parameter should be pointer to MppEncOSDPltCfg */
|
||||
MPP_ENC_SET_OSD_DATA_CFG, /* set OSD data with at most 8 regions, parameter should be pointer to MppEncOSDData */
|
||||
|
||||
MPP_ENC_CMD_END,
|
||||
|
||||
MPP_ISP_CMD_BASE = CMD_MODULE_CODEC | CMD_CTX_ID_ISP,
|
||||
MPP_ISP_CMD_END,
|
||||
|
||||
MPP_HAL_CMD_BASE = CMD_MODULE_HAL,
|
||||
MPP_HAL_CMD_END,
|
||||
|
||||
MPI_CMD_BUTT,
|
||||
} MpiCmd;
|
||||
|
||||
#include "rk_vdec_cmd.h"
|
||||
#include "rk_vdec_cfg.h"
|
||||
#include "rk_venc_cmd.h"
|
||||
#include "rk_venc_cfg.h"
|
||||
#include "rk_venc_ref.h"
|
||||
|
||||
#endif /*__RK_MPI_CMD_H__*/
|
||||
142
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_type.h
vendored
Normal file
142
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_type.h
vendored
Normal file
|
|
@ -0,0 +1,142 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __RK_TYPE_H__
|
||||
#define __RK_TYPE_H__
|
||||
|
||||
#include <stddef.h>
|
||||
|
||||
#if defined(_WIN32) && !defined(__MINGW32CE__)
|
||||
|
||||
typedef unsigned char RK_U8;
|
||||
typedef unsigned short RK_U16;
|
||||
typedef unsigned int RK_U32;
|
||||
typedef unsigned long RK_ULONG;
|
||||
typedef unsigned __int64 RK_U64;
|
||||
|
||||
typedef signed char RK_S8;
|
||||
typedef signed short RK_S16;
|
||||
typedef signed int RK_S32;
|
||||
typedef signed long RK_LONG;
|
||||
typedef signed __int64 RK_S64;
|
||||
|
||||
#else
|
||||
|
||||
typedef unsigned char RK_U8;
|
||||
typedef unsigned short RK_U16;
|
||||
typedef unsigned int RK_U32;
|
||||
typedef unsigned long RK_ULONG;
|
||||
typedef unsigned long long int RK_U64;
|
||||
|
||||
|
||||
typedef signed char RK_S8;
|
||||
typedef signed short RK_S16;
|
||||
typedef signed int RK_S32;
|
||||
typedef signed long RK_LONG;
|
||||
typedef signed long long int RK_S64;
|
||||
|
||||
#endif
|
||||
|
||||
#ifndef MODULE_TAG
|
||||
#define MODULE_TAG NULL
|
||||
#endif
|
||||
|
||||
/**
|
||||
* @ingroup rk_mpi
|
||||
* @brief The type of mpp context
|
||||
* @details This type is used when calling mpp_init(), which including decoder,
|
||||
* encoder and Image Signal Process(ISP). So far decoder and encoder
|
||||
* are supported perfectly, and ISP will be supported in the future.
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_CTX_DEC, /**< decoder */
|
||||
MPP_CTX_ENC, /**< encoder */
|
||||
MPP_CTX_ISP, /**< isp */
|
||||
MPP_CTX_BUTT, /**< undefined */
|
||||
} MppCtxType;
|
||||
|
||||
/**
|
||||
* @ingroup rk_mpi
|
||||
* @brief Enumeration used to define the possible video compression codings.
|
||||
* sync with the omx_video.h
|
||||
*
|
||||
* @note This essentially refers to file extensions. If the coding is
|
||||
* being used to specify the ENCODE type, then additional work
|
||||
* must be done to configure the exact flavor of the compression
|
||||
* to be used. For decode cases where the user application can
|
||||
* not differentiate between MPEG-4 and H.264 bit streams, it is
|
||||
* up to the codec to handle this.
|
||||
*/
|
||||
typedef enum {
|
||||
MPP_VIDEO_CodingUnused, /**< Value when coding is N/A */
|
||||
MPP_VIDEO_CodingAutoDetect, /**< Autodetection of coding type */
|
||||
MPP_VIDEO_CodingMPEG2, /**< AKA: H.262 */
|
||||
MPP_VIDEO_CodingH263, /**< H.263 */
|
||||
MPP_VIDEO_CodingMPEG4, /**< MPEG-4 */
|
||||
MPP_VIDEO_CodingWMV, /**< Windows Media Video (WMV1,WMV2,WMV3)*/
|
||||
MPP_VIDEO_CodingRV, /**< all versions of Real Video */
|
||||
MPP_VIDEO_CodingAVC, /**< H.264/AVC */
|
||||
MPP_VIDEO_CodingMJPEG, /**< Motion JPEG */
|
||||
MPP_VIDEO_CodingVP8, /**< VP8 */
|
||||
MPP_VIDEO_CodingVP9, /**< VP9 */
|
||||
MPP_VIDEO_CodingVC1 = 0x01000000, /**< Windows Media Video (WMV1,WMV2,WMV3)*/
|
||||
MPP_VIDEO_CodingFLV1, /**< Sorenson H.263 */
|
||||
MPP_VIDEO_CodingDIVX3, /**< DIVX3 */
|
||||
MPP_VIDEO_CodingVP6,
|
||||
MPP_VIDEO_CodingHEVC, /**< H.265/HEVC */
|
||||
MPP_VIDEO_CodingAVSPLUS, /**< AVS+ */
|
||||
MPP_VIDEO_CodingAVS, /**< AVS profile=0x20 */
|
||||
MPP_VIDEO_CodingAVS2, /**< AVS2 */
|
||||
MPP_VIDEO_CodingAV1, /**< av1 */
|
||||
MPP_VIDEO_CodingKhronosExtensions = 0x6F000000, /**< Reserved region for introducing Khronos Standard Extensions */
|
||||
MPP_VIDEO_CodingVendorStartUnused = 0x7F000000, /**< Reserved region for introducing Vendor Extensions */
|
||||
MPP_VIDEO_CodingMax = 0x7FFFFFFF
|
||||
} MppCodingType;
|
||||
|
||||
/*
|
||||
* All external interface object list here.
|
||||
* The interface object is defined as void * for expandability
|
||||
* The cross include between these objects will introduce extra
|
||||
* compiling difficulty. So we move them together in this header.
|
||||
*
|
||||
* Object interface header list:
|
||||
*
|
||||
* MppCtx - rk_mpi.h
|
||||
* MppParam - rk_mpi.h
|
||||
*
|
||||
* MppFrame - mpp_frame.h
|
||||
* MppPacket - mpp_packet.h
|
||||
*
|
||||
* MppBuffer - mpp_buffer.h
|
||||
* MppBufferGroup - mpp_buffer.h
|
||||
*
|
||||
* MppTask - mpp_task.h
|
||||
* MppMeta - mpp_meta.h
|
||||
*/
|
||||
|
||||
typedef void* MppCtx;
|
||||
typedef void* MppParam;
|
||||
|
||||
typedef void* MppFrame;
|
||||
typedef void* MppPacket;
|
||||
|
||||
typedef void* MppBuffer;
|
||||
typedef void* MppBufferGroup;
|
||||
|
||||
typedef void* MppTask;
|
||||
typedef void* MppMeta;
|
||||
|
||||
#endif /*__RK_TYPE_H__*/
|
||||
50
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_vdec_cfg.h
vendored
Normal file
50
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_vdec_cfg.h
vendored
Normal file
|
|
@ -0,0 +1,50 @@
|
|||
/*
|
||||
* Copyright 2020 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __RK_VDEC_CFG_H__
|
||||
#define __RK_VDEC_CFG_H__
|
||||
|
||||
#include "rk_type.h"
|
||||
#include "mpp_err.h"
|
||||
|
||||
typedef void* MppDecCfg;
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
MPP_RET mpp_dec_cfg_init(MppDecCfg *cfg);
|
||||
MPP_RET mpp_dec_cfg_deinit(MppDecCfg cfg);
|
||||
|
||||
MPP_RET mpp_dec_cfg_set_s32(MppDecCfg cfg, const char *name, RK_S32 val);
|
||||
MPP_RET mpp_dec_cfg_set_u32(MppDecCfg cfg, const char *name, RK_U32 val);
|
||||
MPP_RET mpp_dec_cfg_set_s64(MppDecCfg cfg, const char *name, RK_S64 val);
|
||||
MPP_RET mpp_dec_cfg_set_u64(MppDecCfg cfg, const char *name, RK_U64 val);
|
||||
MPP_RET mpp_dec_cfg_set_ptr(MppDecCfg cfg, const char *name, void *val);
|
||||
|
||||
MPP_RET mpp_dec_cfg_get_s32(MppDecCfg cfg, const char *name, RK_S32 *val);
|
||||
MPP_RET mpp_dec_cfg_get_u32(MppDecCfg cfg, const char *name, RK_U32 *val);
|
||||
MPP_RET mpp_dec_cfg_get_s64(MppDecCfg cfg, const char *name, RK_S64 *val);
|
||||
MPP_RET mpp_dec_cfg_get_u64(MppDecCfg cfg, const char *name, RK_U64 *val);
|
||||
MPP_RET mpp_dec_cfg_get_ptr(MppDecCfg cfg, const char *name, void **val);
|
||||
|
||||
void mpp_dec_cfg_show(void);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__RK_VDEC_CFG_H__*/
|
||||
69
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_vdec_cmd.h
vendored
Normal file
69
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_vdec_cmd.h
vendored
Normal file
|
|
@ -0,0 +1,69 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __RK_VDEC_CMD_H__
|
||||
#define __RK_VDEC_CMD_H__
|
||||
|
||||
#include "rk_type.h"
|
||||
#include "mpp_err.h"
|
||||
|
||||
/*
|
||||
* decoder query interface is only for debug usage
|
||||
*/
|
||||
#define MPP_DEC_QUERY_STATUS (0x00000001)
|
||||
#define MPP_DEC_QUERY_WAIT (0x00000002)
|
||||
#define MPP_DEC_QUERY_FPS (0x00000004)
|
||||
#define MPP_DEC_QUERY_BPS (0x00000008)
|
||||
#define MPP_DEC_QUERY_DEC_IN_PKT (0x00000010)
|
||||
#define MPP_DEC_QUERY_DEC_WORK (0x00000020)
|
||||
#define MPP_DEC_QUERY_DEC_OUT_FRM (0x00000040)
|
||||
|
||||
#define MPP_DEC_QUERY_ALL (MPP_DEC_QUERY_STATUS | \
|
||||
MPP_DEC_QUERY_WAIT | \
|
||||
MPP_DEC_QUERY_FPS | \
|
||||
MPP_DEC_QUERY_BPS | \
|
||||
MPP_DEC_QUERY_DEC_IN_PKT | \
|
||||
MPP_DEC_QUERY_DEC_WORK | \
|
||||
MPP_DEC_QUERY_DEC_OUT_FRM)
|
||||
|
||||
typedef struct MppDecQueryCfg_t {
|
||||
/*
|
||||
* 32 bit query flag for query data check
|
||||
* Each bit represent a query data switch.
|
||||
* bit 0 - for querying decoder runtime status
|
||||
* bit 1 - for querying decoder runtime waiting status
|
||||
* bit 2 - for querying decoder realtime decode fps
|
||||
* bit 3 - for querying decoder realtime input bps
|
||||
* bit 4 - for querying decoder input packet count
|
||||
* bit 5 - for querying decoder start hardware times
|
||||
* bit 6 - for querying decoder output frame count
|
||||
*/
|
||||
RK_U32 query_flag;
|
||||
|
||||
/* 64 bit query data output */
|
||||
RK_U32 rt_status;
|
||||
RK_U32 rt_wait;
|
||||
RK_U32 rt_fps;
|
||||
RK_U32 rt_bps;
|
||||
RK_U32 dec_in_pkt_cnt;
|
||||
RK_U32 dec_hw_run_cnt;
|
||||
RK_U32 dec_out_frm_cnt;
|
||||
} MppDecQueryCfg;
|
||||
|
||||
typedef void* MppExtCbCtx;
|
||||
typedef MPP_RET (*MppExtCbFunc)(MppExtCbCtx cb_ctx, MppCtx mpp, RK_S32 cmd, void *arg);
|
||||
|
||||
#endif /*__RK_VDEC_CMD_H__*/
|
||||
52
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_venc_cfg.h
vendored
Normal file
52
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_venc_cfg.h
vendored
Normal file
|
|
@ -0,0 +1,52 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __RK_VENC_CFG_H__
|
||||
#define __RK_VENC_CFG_H__
|
||||
|
||||
#include "rk_type.h"
|
||||
#include "mpp_err.h"
|
||||
|
||||
typedef void* MppEncCfg;
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
MPP_RET mpp_enc_cfg_init(MppEncCfg *cfg);
|
||||
MPP_RET mpp_enc_cfg_deinit(MppEncCfg cfg);
|
||||
|
||||
MPP_RET mpp_enc_cfg_set_s32(MppEncCfg cfg, const char *name, RK_S32 val);
|
||||
MPP_RET mpp_enc_cfg_set_u32(MppEncCfg cfg, const char *name, RK_U32 val);
|
||||
MPP_RET mpp_enc_cfg_set_s64(MppEncCfg cfg, const char *name, RK_S64 val);
|
||||
MPP_RET mpp_enc_cfg_set_u64(MppEncCfg cfg, const char *name, RK_U64 val);
|
||||
MPP_RET mpp_enc_cfg_set_ptr(MppEncCfg cfg, const char *name, void *val);
|
||||
MPP_RET mpp_enc_cfg_set_st(MppEncCfg cfg, const char *name, void *val);
|
||||
|
||||
MPP_RET mpp_enc_cfg_get_s32(MppEncCfg cfg, const char *name, RK_S32 *val);
|
||||
MPP_RET mpp_enc_cfg_get_u32(MppEncCfg cfg, const char *name, RK_U32 *val);
|
||||
MPP_RET mpp_enc_cfg_get_s64(MppEncCfg cfg, const char *name, RK_S64 *val);
|
||||
MPP_RET mpp_enc_cfg_get_u64(MppEncCfg cfg, const char *name, RK_U64 *val);
|
||||
MPP_RET mpp_enc_cfg_get_ptr(MppEncCfg cfg, const char *name, void **val);
|
||||
MPP_RET mpp_enc_cfg_get_st(MppEncCfg cfg, const char *name, void *val);
|
||||
|
||||
void mpp_enc_cfg_show(void);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__RK_VENC_CFG_H__*/
|
||||
1394
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_venc_cmd.h
vendored
Normal file
1394
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_venc_cmd.h
vendored
Normal file
File diff suppressed because it is too large
Load Diff
66
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_venc_rc.h
vendored
Normal file
66
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_venc_rc.h
vendored
Normal file
|
|
@ -0,0 +1,66 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __RK_VENC_RC_H__
|
||||
#define __RK_VENC_RC_H__
|
||||
|
||||
#include "rk_type.h"
|
||||
|
||||
#define MPP_ENC_MIN_BPS (SZ_1K)
|
||||
#define MPP_ENC_MAX_BPS (SZ_1M * 200)
|
||||
|
||||
/* Rate control parameter */
|
||||
typedef enum MppEncRcMode_e {
|
||||
MPP_ENC_RC_MODE_VBR,
|
||||
MPP_ENC_RC_MODE_CBR,
|
||||
MPP_ENC_RC_MODE_FIXQP,
|
||||
MPP_ENC_RC_MODE_AVBR,
|
||||
MPP_ENC_RC_MODE_BUTT
|
||||
} MppEncRcMode;
|
||||
|
||||
typedef enum MppEncRcPriority_e {
|
||||
MPP_ENC_RC_BY_BITRATE_FIRST,
|
||||
MPP_ENC_RC_BY_FRM_SIZE_FIRST,
|
||||
MPP_ENC_RC_PRIORITY_BUTT
|
||||
} MppEncRcPriority;
|
||||
|
||||
typedef enum MppEncRcDropFrmMode_e {
|
||||
MPP_ENC_RC_DROP_FRM_DISABLED,
|
||||
MPP_ENC_RC_DROP_FRM_NORMAL,
|
||||
MPP_ENC_RC_DROP_FRM_PSKIP,
|
||||
MPP_ENC_RC_DROP_FRM_BUTT
|
||||
} MppEncRcDropFrmMode;
|
||||
|
||||
typedef enum MppEncRcSuperFrameMode_t {
|
||||
MPP_ENC_RC_SUPER_FRM_NONE,
|
||||
MPP_ENC_RC_SUPER_FRM_DROP,
|
||||
MPP_ENC_RC_SUPER_FRM_REENC,
|
||||
MPP_ENC_RC_SUPER_FRM_BUTT
|
||||
} MppEncRcSuperFrameMode;
|
||||
|
||||
typedef enum MppEncRcGopMode_e {
|
||||
MPP_ENC_RC_NORMAL_P,
|
||||
MPP_ENC_RC_SMART_P,
|
||||
MPP_ENC_RC_GOP_MODE_BUTT,
|
||||
} MppEncRcGopMode;
|
||||
|
||||
typedef enum MppEncRcIntraRefreshMode_e {
|
||||
MPP_ENC_RC_INTRA_REFRESH_ROW = 0,
|
||||
MPP_ENC_RC_INTRA_REFRESH_COL,
|
||||
MPP_ENC_RC_INTRA_REFRESH_BUTT
|
||||
} MppEncRcRefreshMode;
|
||||
|
||||
#endif /*__RK_VENC_RC_H__*/
|
||||
242
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_venc_ref.h
vendored
Normal file
242
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/rk_venc_ref.h
vendored
Normal file
|
|
@ -0,0 +1,242 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __RK_VENC_REF_H__
|
||||
#define __RK_VENC_REF_H__
|
||||
|
||||
#include "rk_type.h"
|
||||
#include "mpp_err.h"
|
||||
|
||||
/*
|
||||
* MPP reference management system follows the model of H.264/H.265 reference
|
||||
* frame mangement.
|
||||
*
|
||||
* The reference frame is defined into two type: long-term reference frame and
|
||||
* short-refernce frame (lt_ref and st_ref).
|
||||
*
|
||||
* The lt_ref can be only indexed by long-term reference frame index (lt_idx).
|
||||
* The st_ref can be indexed by its temporal id (tid) and previous count.
|
||||
*
|
||||
* MppEncRefMode defined the way for user to reference the required frame.
|
||||
*
|
||||
* Normal reference mode without argument
|
||||
* REF_TO_PREV_REF_FRM - refer to previous reference frame in encode order (No matter Lt or St)
|
||||
* REF_TO_PREV_ST_REF - refer to previous short-term reference frame
|
||||
* REF_TO_PREV_LT_REF - refer to previous long-term reference frame
|
||||
* REF_TO_PREV_INTRA - refer to previous Intra / IDR frame
|
||||
* REF_TO_ST_REF_SETUP - refer to refernce frame defined in StRefSetup
|
||||
*
|
||||
* Normal reference mode with argument
|
||||
* REF_TO_TEMPORAL_LAYER - refer to previous reference frame with temporal id argument
|
||||
* REF_TO_LT_REF_IDX - refer to long-term reference frame with lt_ref_idx argument
|
||||
* REF_TO_ST_PREV_N_REF - refer to short-term reference frame with diff frame_num argument
|
||||
*
|
||||
* Long-term reference only mode
|
||||
* REF_TO_ST_REF_SETUP - use corresponding mode of original short-term reference frame
|
||||
*
|
||||
* Short-term reference only mode
|
||||
* REF_TO_LT_REF_SETUP - indicate that this frame will be overwrited by long-term config
|
||||
*
|
||||
* By combining frames with these modes user can define many kinds of reference hierarchy
|
||||
* structure. But normally user should use simplified preset hierarchy pattern.
|
||||
*
|
||||
* The rules for virtual cpb management is similiar to H.264/H.265
|
||||
* 1. When one frame is marked as long-term reference frame it will be kept in cpb until
|
||||
* it is replaced by other frame with the same lt_idx or IDR frame.
|
||||
* 2. When one frame is marked as short-term reference frame it will be inert into cpb when
|
||||
* there is enough storage space. When the number of total sum of long-term and short-term
|
||||
* reference frame excess the cpb size limit the oldest short-term frame will be removed.
|
||||
* This is call sliding window in H.264.
|
||||
*/
|
||||
|
||||
/* max 4 temporal layer */
|
||||
#define MPP_ENC_MAX_TEMPORAL_LAYER_NUM 4
|
||||
/* max 4 long-term reference frame */
|
||||
#define MPP_ENC_MAX_LT_REF_NUM 16
|
||||
|
||||
/*
|
||||
* Group Of Picture (GOP) config is separated into three parts:
|
||||
*
|
||||
* 1. Intra / IDR frame config
|
||||
* igop - the interval of two intra / IDR frames
|
||||
*
|
||||
* 2. Long-term reference config (MppEncRefLtFrmCfg)
|
||||
*
|
||||
* Setup long-term reference index max lt_idx, loop interval and reference
|
||||
* mode for auto long-term reference frame generation. The encoder will
|
||||
* mark frame to be long-term reference frame with given interval.
|
||||
*
|
||||
* 2.1 lt_idx
|
||||
* The long-term reference frame index is unique identifier for a long-term
|
||||
* reference frame.
|
||||
* The max long-term reference frame index should NOT larger than
|
||||
* max_num_ref_frames in sps.
|
||||
*
|
||||
* 2.2 lt_gap
|
||||
* When lt_gap is zero the long-term reference frame generation is disabled.
|
||||
* When lt_gap is non-zero (usually 2~3 second interval) then the long-term
|
||||
* reference frame will be generated for error recovery or smart hierarchy.
|
||||
*
|
||||
* 2.2 lt_delay
|
||||
* The lt_delay is the delay time for generation of long-term reference frame.
|
||||
* The start point of lt_delay is the IDR/intra frame genertaed by igop.
|
||||
*
|
||||
* 2.4 ref_mode: Long-term refernce frame reference mode
|
||||
* NOTE: temporal id of longterm reference frame is always zero.
|
||||
*
|
||||
* Examples:
|
||||
* Sequence has only one lt_ref 0 and setup one long-term reference frame
|
||||
* every 300 frame.
|
||||
* {
|
||||
* .lt_idx = 0,
|
||||
* .lt_gap = 300,
|
||||
* .lt_delay = 0,
|
||||
* }
|
||||
* result:
|
||||
* frame 0 ...... 299 300 301 ...... 599 600 601
|
||||
* lt_idx 0 xxxxxx x 0 x xxxxxx x 0 x
|
||||
*
|
||||
* Sequence has lt_ref from 0 to 2 and setup a long-term reference frame
|
||||
* every 100 frame.
|
||||
* {
|
||||
* .lt_idx = 0,
|
||||
* .lt_gap = 300,
|
||||
* .lt_delay = 0,
|
||||
* }
|
||||
* {
|
||||
* .lt_idx = 1,
|
||||
* .lt_gap = 300,
|
||||
* .lt_delay = 100,
|
||||
* }
|
||||
* {
|
||||
* .lt_idx = 2,
|
||||
* .lt_gap = 300,
|
||||
* .lt_delay = 200,
|
||||
* }
|
||||
* result:
|
||||
* frame 0 ... 99 100 101 ... 199 200 201 ... 299 300 301
|
||||
* lt_idx 0 xxx x 1 x xxx x 2 x xxx x 0 x
|
||||
*
|
||||
* 3. Short-term reference config (MppEncStRefSetup)
|
||||
*
|
||||
* 3.1 is_non_ref
|
||||
* The is_non_ref indicated the current frame is reference frame or not.
|
||||
*
|
||||
* 3.2 temporal_id
|
||||
* The temporal id of the current frame configure.
|
||||
*
|
||||
* 3.3 ref_mode: short-term refernce frame reference mode
|
||||
*
|
||||
* 3.4 repeat
|
||||
* The repeat time of the short-term reference frame configure.
|
||||
* The overall frame count with the same config is repeat + 1.
|
||||
*
|
||||
* Examples:
|
||||
*
|
||||
*/
|
||||
|
||||
#define REF_MODE_MODE_MASK (0x1F)
|
||||
#define REF_MODE_ARG_MASK (0xFFFF0000)
|
||||
|
||||
typedef enum MppEncRefMode_e {
|
||||
/* max 32 mode in 32-bit */
|
||||
/* for default ref global config */
|
||||
REF_MODE_GLOBAL,
|
||||
REF_TO_PREV_REF_FRM = REF_MODE_GLOBAL,
|
||||
REF_TO_PREV_ST_REF,
|
||||
REF_TO_PREV_LT_REF,
|
||||
REF_TO_PREV_INTRA,
|
||||
|
||||
/* for global config with args */
|
||||
REF_MODE_GLOBAL_WITH_ARG = 0x4,
|
||||
/* with ref arg as temporal layer id */
|
||||
REF_TO_TEMPORAL_LAYER = REF_MODE_GLOBAL_WITH_ARG,
|
||||
/* with ref arg as long-term reference picture index */
|
||||
REF_TO_LT_REF_IDX,
|
||||
/* with ref arg as short-term reference picture difference frame_num */
|
||||
REF_TO_ST_PREV_N_REF,
|
||||
REF_MODE_GLOBAL_BUTT,
|
||||
|
||||
/* for lt-ref */
|
||||
REF_MODE_LT = 0x18,
|
||||
REF_TO_ST_REF_SETUP,
|
||||
REF_MODE_LT_BUTT,
|
||||
|
||||
/* for st-ref */
|
||||
REF_MODE_ST = 0x1C,
|
||||
REF_TO_LT_REF_SETUP,
|
||||
REF_MODE_ST_BUTT,
|
||||
} MppEncRefMode;
|
||||
|
||||
typedef struct MppEncRefLtFrmCfg_t {
|
||||
RK_S32 lt_idx; /* lt_idx of the reference frame */
|
||||
RK_S32 temporal_id; /* temporal_id of the reference frame */
|
||||
MppEncRefMode ref_mode;
|
||||
RK_S32 ref_arg;
|
||||
RK_S32 lt_gap; /* gap between two lt-ref with same lt_idx */
|
||||
RK_S32 lt_delay; /* delay offset to igop start frame */
|
||||
} MppEncRefLtFrmCfg;
|
||||
|
||||
typedef struct MppEncRefStFrmCfg_t {
|
||||
RK_S32 is_non_ref;
|
||||
RK_S32 temporal_id;
|
||||
MppEncRefMode ref_mode;
|
||||
RK_S32 ref_arg;
|
||||
RK_S32 repeat; /* repeat times */
|
||||
} MppEncRefStFrmCfg;
|
||||
|
||||
typedef struct MppEncRefPreset_t {
|
||||
/* input parameter for query */
|
||||
const char *name;
|
||||
RK_S32 max_lt_cnt;
|
||||
RK_S32 max_st_cnt;
|
||||
MppEncRefLtFrmCfg *lt_cfg;
|
||||
MppEncRefStFrmCfg *st_cfg;
|
||||
|
||||
/* output parameter */
|
||||
RK_S32 lt_cnt;
|
||||
RK_S32 st_cnt;
|
||||
} MppEncRefPreset;
|
||||
|
||||
typedef void* MppEncRefCfg;
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
MPP_RET mpp_enc_ref_cfg_init(MppEncRefCfg *ref);
|
||||
MPP_RET mpp_enc_ref_cfg_deinit(MppEncRefCfg *ref);
|
||||
|
||||
MPP_RET mpp_enc_ref_cfg_reset(MppEncRefCfg ref);
|
||||
MPP_RET mpp_enc_ref_cfg_set_cfg_cnt(MppEncRefCfg ref, RK_S32 lt_cnt, RK_S32 st_cnt);
|
||||
MPP_RET mpp_enc_ref_cfg_add_lt_cfg(MppEncRefCfg ref, RK_S32 cnt, MppEncRefLtFrmCfg *frm);
|
||||
MPP_RET mpp_enc_ref_cfg_add_st_cfg(MppEncRefCfg ref, RK_S32 cnt, MppEncRefStFrmCfg *frm);
|
||||
MPP_RET mpp_enc_ref_cfg_check(MppEncRefCfg ref);
|
||||
|
||||
/*
|
||||
* A new reference configure will restart a new gop and clear cpb by default.
|
||||
* The keep cpb function will let encoder keeps the current cpb status and do NOT
|
||||
* reset all the reference frame in cpb.
|
||||
*/
|
||||
MPP_RET mpp_enc_ref_cfg_set_keep_cpb(MppEncRefCfg ref, RK_S32 keep);
|
||||
MPP_RET mpp_enc_ref_cfg_get_preset(MppEncRefPreset *preset);
|
||||
MPP_RET mpp_enc_ref_cfg_show(MppEncRefCfg ref);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__RK_VENC_REF_H__*/
|
||||
123
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/vpu.h
vendored
Normal file
123
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/vpu.h
vendored
Normal file
|
|
@ -0,0 +1,123 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __VPU_H__
|
||||
#define __VPU_H__
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
#include "rk_type.h"
|
||||
|
||||
#define VPU_SUCCESS (0)
|
||||
#define VPU_FAILURE (-1)
|
||||
|
||||
#define VPU_HW_WAIT_OK VPU_SUCCESS
|
||||
#define VPU_HW_WAIT_ERROR VPU_FAILURE
|
||||
#define VPU_HW_WAIT_TIMEOUT 1
|
||||
|
||||
// vpu decoder 60 registers, size 240B
|
||||
#define VPU_REG_NUM_DEC (60)
|
||||
// vpu post processor 41 registers, size 164B
|
||||
#define VPU_REG_NUM_PP (41)
|
||||
// vpu decoder + post processor 101 registers, size 404B
|
||||
#define VPU_REG_NUM_DEC_PP (VPU_REG_NUM_DEC+VPU_REG_NUM_PP)
|
||||
// vpu encoder 96 registers, size 384B
|
||||
#define VPU_REG_NUM_ENC (96)
|
||||
|
||||
typedef enum {
|
||||
VPU_ENC = 0x0,
|
||||
VPU_DEC = 0x1,
|
||||
VPU_PP = 0x2,
|
||||
VPU_DEC_PP = 0x3,
|
||||
VPU_DEC_HEVC = 0x4,
|
||||
VPU_DEC_RKV = 0x5,
|
||||
VPU_ENC_RKV = 0x6,
|
||||
VPU_DEC_AVSPLUS = 0x7,
|
||||
VPU_ENC_VEPU22 = 0x8,
|
||||
VPU_TYPE_BUTT ,
|
||||
} VPU_CLIENT_TYPE;
|
||||
|
||||
/* Hardware decoder configuration description */
|
||||
|
||||
typedef struct VPUHwDecConfig {
|
||||
RK_U32 maxDecPicWidth; /* Maximum video decoding width supported */
|
||||
RK_U32 maxPpOutPicWidth; /* Maximum output width of Post-Processor */
|
||||
RK_U32 h264Support; /* HW supports h.264 */
|
||||
RK_U32 jpegSupport; /* HW supports JPEG */
|
||||
RK_U32 mpeg4Support; /* HW supports MPEG-4 */
|
||||
RK_U32 customMpeg4Support; /* HW supports custom MPEG-4 features */
|
||||
RK_U32 vc1Support; /* HW supports VC-1 Simple */
|
||||
RK_U32 mpeg2Support; /* HW supports MPEG-2 */
|
||||
RK_U32 ppSupport; /* HW supports post-processor */
|
||||
RK_U32 ppConfig; /* HW post-processor functions bitmask */
|
||||
RK_U32 sorensonSparkSupport; /* HW supports Sorenson Spark */
|
||||
RK_U32 refBufSupport; /* HW supports reference picture buffering */
|
||||
RK_U32 vp6Support; /* HW supports VP6 */
|
||||
RK_U32 vp7Support; /* HW supports VP7 */
|
||||
RK_U32 vp8Support; /* HW supports VP8 */
|
||||
RK_U32 avsSupport; /* HW supports AVS */
|
||||
RK_U32 jpegESupport; /* HW supports JPEG extensions */
|
||||
RK_U32 rvSupport; /* HW supports REAL */
|
||||
RK_U32 mvcSupport; /* HW supports H264 MVC extension */
|
||||
} VPUHwDecConfig_t;
|
||||
|
||||
/* Hardware encoder configuration description */
|
||||
|
||||
typedef struct VPUHwEndConfig {
|
||||
RK_U32 maxEncodedWidth; /* Maximum supported width for video encoding (not JPEG) */
|
||||
RK_U32 h264Enabled; /* HW supports H.264 */
|
||||
RK_U32 jpegEnabled; /* HW supports JPEG */
|
||||
RK_U32 mpeg4Enabled; /* HW supports MPEG-4 */
|
||||
RK_U32 vsEnabled; /* HW supports video stabilization */
|
||||
RK_U32 rgbEnabled; /* HW supports RGB input */
|
||||
RK_U32 reg_size; /* HW bus type in use */
|
||||
RK_U32 reserv[2];
|
||||
} VPUHwEncConfig_t;
|
||||
|
||||
typedef enum {
|
||||
// common command
|
||||
VPU_CMD_REGISTER ,
|
||||
VPU_CMD_REGISTER_ACK_OK ,
|
||||
VPU_CMD_REGISTER_ACK_FAIL ,
|
||||
VPU_CMD_UNREGISTER ,
|
||||
|
||||
VPU_SEND_CONFIG ,
|
||||
VPU_SEND_CONFIG_ACK_OK ,
|
||||
VPU_SEND_CONFIG_ACK_FAIL ,
|
||||
|
||||
VPU_GET_HW_INFO ,
|
||||
VPU_GET_HW_INFO_ACK_OK ,
|
||||
VPU_GET_HW_INFO_ACK_FAIL ,
|
||||
|
||||
VPU_CMD_BUTT ,
|
||||
} VPU_CMD_TYPE;
|
||||
|
||||
int VPUClientInit(VPU_CLIENT_TYPE type);
|
||||
RK_S32 VPUClientRelease(int socket);
|
||||
RK_S32 VPUClientSendReg(int socket, RK_U32 *regs, RK_U32 nregs);
|
||||
RK_S32 VPUClientSendReg2(RK_S32 socket, RK_S32 offset, RK_S32 size, void *param);
|
||||
RK_S32 VPUClientWaitResult(int socket, RK_U32 *regs, RK_U32 nregs, VPU_CMD_TYPE *cmd, RK_S32 *len);
|
||||
RK_S32 VPUClientGetHwCfg(int socket, RK_U32 *cfg, RK_U32 cfg_size);
|
||||
RK_S32 VPUClientGetIOMMUStatus();
|
||||
RK_U32 VPUCheckSupportWidth();
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /* __VPU_H__ */
|
||||
504
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/vpu_api.h
vendored
Normal file
504
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/mpp/include/rockchip/vpu_api.h
vendored
Normal file
|
|
@ -0,0 +1,504 @@
|
|||
/*
|
||||
* Copyright 2015 Rockchip Electronics Co. LTD
|
||||
*
|
||||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||||
* you may not use this file except in compliance with the License.
|
||||
* You may obtain a copy of the License at
|
||||
*
|
||||
* http://www.apache.org/licenses/LICENSE-2.0
|
||||
*
|
||||
* Unless required by applicable law or agreed to in writing, software
|
||||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
* See the License for the specific language governing permissions and
|
||||
* limitations under the License.
|
||||
*/
|
||||
|
||||
#ifndef __VPU_API_H__
|
||||
#define __VPU_API_H__
|
||||
|
||||
#include "rk_type.h"
|
||||
#include "mpp_err.h"
|
||||
|
||||
/**
|
||||
* @brief rockchip media process interface
|
||||
*/
|
||||
|
||||
#define VPU_API_NOPTS_VALUE (0x8000000000000000LL)
|
||||
|
||||
/*
|
||||
* bit definition of ColorType in structure VPU_FRAME
|
||||
*/
|
||||
#define VPU_OUTPUT_FORMAT_TYPE_MASK (0x0000ffff)
|
||||
#define VPU_OUTPUT_FORMAT_ARGB8888 (0x00000000)
|
||||
#define VPU_OUTPUT_FORMAT_ABGR8888 (0x00000001)
|
||||
#define VPU_OUTPUT_FORMAT_RGB888 (0x00000002)
|
||||
#define VPU_OUTPUT_FORMAT_RGB565 (0x00000003)
|
||||
#define VPU_OUTPUT_FORMAT_RGB555 (0x00000004)
|
||||
#define VPU_OUTPUT_FORMAT_YUV420_SEMIPLANAR (0x00000005)
|
||||
#define VPU_OUTPUT_FORMAT_YUV420_PLANAR (0x00000006)
|
||||
#define VPU_OUTPUT_FORMAT_YUV422 (0x00000007)
|
||||
#define VPU_OUTPUT_FORMAT_YUV444 (0x00000008)
|
||||
#define VPU_OUTPUT_FORMAT_YCH420 (0x00000009)
|
||||
#define VPU_OUTPUT_FORMAT_BIT_MASK (0x000f0000)
|
||||
#define VPU_OUTPUT_FORMAT_BIT_8 (0x00000000)
|
||||
#define VPU_OUTPUT_FORMAT_BIT_10 (0x00010000)
|
||||
#define VPU_OUTPUT_FORMAT_BIT_12 (0x00020000)
|
||||
#define VPU_OUTPUT_FORMAT_BIT_14 (0x00030000)
|
||||
#define VPU_OUTPUT_FORMAT_BIT_16 (0x00040000)
|
||||
#define VPU_OUTPUT_FORMAT_FBC_MASK (0x00f00000)
|
||||
#define VPU_OUTPUT_FORMAT_FBC_AFBC_V1 (0x00100000)
|
||||
#define VPU_OUTPUT_FORMAT_FBC_AFBC_V2 (0x00200000)
|
||||
#define VPU_OUTPUT_FORMAT_DYNCRANGE_MASK (0x0f000000)
|
||||
#define VPU_OUTPUT_FORMAT_DYNCRANGE_SDR (0x00000000)
|
||||
#define VPU_OUTPUT_FORMAT_DYNCRANGE_HDR10 (0x01000000)
|
||||
#define VPU_OUTPUT_FORMAT_DYNCRANGE_HDR_HLG (0x02000000)
|
||||
#define VPU_OUTPUT_FORMAT_DYNCRANGE_HDR_DOLBY (0x03000000)
|
||||
|
||||
/**
|
||||
* @brief input picture type
|
||||
*/
|
||||
typedef enum {
|
||||
ENC_INPUT_YUV420_PLANAR = 0, /**< YYYY... UUUU... VVVV */
|
||||
ENC_INPUT_YUV420_SEMIPLANAR = 1, /**< YYYY... UVUVUV... */
|
||||
ENC_INPUT_YUV422_INTERLEAVED_YUYV = 2, /**< YUYVYUYV... */
|
||||
ENC_INPUT_YUV422_INTERLEAVED_UYVY = 3, /**< UYVYUYVY... */
|
||||
ENC_INPUT_RGB565 = 4, /**< 16-bit RGB */
|
||||
ENC_INPUT_BGR565 = 5, /**< 16-bit RGB */
|
||||
ENC_INPUT_RGB555 = 6, /**< 15-bit RGB */
|
||||
ENC_INPUT_BGR555 = 7, /**< 15-bit RGB */
|
||||
ENC_INPUT_RGB444 = 8, /**< 12-bit RGB */
|
||||
ENC_INPUT_BGR444 = 9, /**< 12-bit RGB */
|
||||
ENC_INPUT_RGB888 = 10, /**< 24-bit RGB */
|
||||
ENC_INPUT_BGR888 = 11, /**< 24-bit RGB */
|
||||
ENC_INPUT_RGB101010 = 12, /**< 30-bit RGB */
|
||||
ENC_INPUT_BGR101010 = 13 /**< 30-bit RGB */
|
||||
} EncInputPictureType;
|
||||
|
||||
typedef enum VPU_API_CMD {
|
||||
VPU_API_ENC_SETCFG,
|
||||
VPU_API_ENC_GETCFG,
|
||||
VPU_API_ENC_SETFORMAT,
|
||||
VPU_API_ENC_SETIDRFRAME,
|
||||
|
||||
VPU_API_ENABLE_DEINTERLACE,
|
||||
VPU_API_SET_VPUMEM_CONTEXT,
|
||||
VPU_API_USE_PRESENT_TIME_ORDER,
|
||||
VPU_API_SET_DEFAULT_WIDTH_HEIGH,
|
||||
VPU_API_SET_INFO_CHANGE,
|
||||
VPU_API_USE_FAST_MODE,
|
||||
VPU_API_DEC_GET_STREAM_COUNT,
|
||||
VPU_API_GET_VPUMEM_USED_COUNT,
|
||||
VPU_API_GET_FRAME_INFO,
|
||||
VPU_API_SET_OUTPUT_BLOCK,
|
||||
VPU_API_GET_EOS_STATUS,
|
||||
VPU_API_SET_OUTPUT_MODE,
|
||||
|
||||
/* get sps/pps header */
|
||||
VPU_API_GET_EXTRA_INFO = 0x200,
|
||||
|
||||
VPU_API_SET_IMMEDIATE_OUT = 0x1000,
|
||||
VPU_API_SET_PARSER_SPLIT_MODE, /* NOTE: should control before init */
|
||||
VPU_API_DEC_OUT_FRM_STRUCT_TYPE,
|
||||
VPU_API_DEC_EN_THUMBNAIL,
|
||||
VPU_API_DEC_EN_HDR_META,
|
||||
VPU_API_DEC_EN_MVC,
|
||||
VPU_API_DEC_EN_FBC_HDR_256_ODD,
|
||||
|
||||
VPU_API_ENC_VEPU22_START = 0x2000,
|
||||
VPU_API_ENC_SET_VEPU22_CFG,
|
||||
VPU_API_ENC_GET_VEPU22_CFG,
|
||||
VPU_API_ENC_SET_VEPU22_CTU_QP,
|
||||
VPU_API_ENC_SET_VEPU22_ROI,
|
||||
|
||||
VPU_API_ENC_MPP = 0x3000,
|
||||
VPU_API_ENC_MPP_SETCFG,
|
||||
VPU_API_ENC_MPP_GETCFG,
|
||||
|
||||
/* mlvec dynamic configure */
|
||||
VPU_API_ENC_MLVEC_CFG = 0x4000,
|
||||
VPU_API_ENC_SET_MAX_TID,
|
||||
VPU_API_ENC_SET_MARK_LTR,
|
||||
VPU_API_ENC_SET_USE_LTR,
|
||||
VPU_API_ENC_SET_FRAME_QP,
|
||||
VPU_API_ENC_SET_BASE_LAYER_PID,
|
||||
} VPU_API_CMD;
|
||||
|
||||
typedef struct {
|
||||
RK_U32 TimeLow;
|
||||
RK_U32 TimeHigh;
|
||||
} TIME_STAMP;
|
||||
|
||||
typedef struct {
|
||||
RK_U32 CodecType;
|
||||
RK_U32 ImgWidth;
|
||||
RK_U32 ImgHeight;
|
||||
RK_U32 ImgHorStride;
|
||||
RK_U32 ImgVerStride;
|
||||
RK_U32 BufSize;
|
||||
} VPU_GENERIC;
|
||||
|
||||
typedef struct VPUMem {
|
||||
RK_U32 phy_addr;
|
||||
RK_U32 *vir_addr;
|
||||
RK_U32 size;
|
||||
RK_U32 *offset;
|
||||
} VPUMemLinear_t;
|
||||
|
||||
typedef struct tVPU_FRAME {
|
||||
RK_U32 FrameBusAddr[2]; // 0: Y address; 1: UV address;
|
||||
RK_U32 FrameWidth; // buffer horizontal stride
|
||||
RK_U32 FrameHeight; // buffer vertical stride
|
||||
RK_U32 OutputWidth; // deprecated
|
||||
RK_U32 OutputHeight; // deprecated
|
||||
RK_U32 DisplayWidth; // valid width for display
|
||||
RK_U32 DisplayHeight; // valid height for display
|
||||
RK_U32 CodingType;
|
||||
RK_U32 FrameType; // frame; top_field_first; bot_field_first
|
||||
RK_U32 ColorType;
|
||||
RK_U32 DecodeFrmNum;
|
||||
TIME_STAMP ShowTime;
|
||||
RK_U32 ErrorInfo; // error information
|
||||
RK_U32 employ_cnt;
|
||||
VPUMemLinear_t vpumem;
|
||||
struct tVPU_FRAME *next_frame;
|
||||
union {
|
||||
struct {
|
||||
RK_U32 Res0[2];
|
||||
struct {
|
||||
RK_U32 ColorPrimaries : 8;
|
||||
RK_U32 ColorTransfer : 8;
|
||||
RK_U32 ColorCoeffs : 8;
|
||||
RK_U32 ColorRange : 1;
|
||||
RK_U32 Res1 : 7;
|
||||
};
|
||||
|
||||
RK_U32 Res2;
|
||||
};
|
||||
|
||||
RK_U32 Res[4];
|
||||
};
|
||||
} VPU_FRAME;
|
||||
|
||||
typedef struct FrameThumbInfo {
|
||||
RK_U32 enable;
|
||||
RK_U32 yOffset;
|
||||
RK_U32 uvOffset;
|
||||
} FrameThumbInfo_t;
|
||||
|
||||
typedef struct FrameHdrInfo {
|
||||
RK_U32 isHdr;
|
||||
RK_U32 offset;
|
||||
RK_U32 size;
|
||||
} FrameHdrInfo_t;
|
||||
|
||||
typedef struct VideoFrame {
|
||||
VPU_FRAME vpuFrame;
|
||||
FrameThumbInfo_t thumbInfo;
|
||||
FrameHdrInfo_t hdrInfo;
|
||||
RK_U32 viewId;
|
||||
RK_U32 reserved[16];
|
||||
} VideoFrame_t;
|
||||
|
||||
typedef struct VideoPacket {
|
||||
RK_S64 pts; /* with unit of us*/
|
||||
RK_S64 dts; /* with unit of us*/
|
||||
RK_U8 *data;
|
||||
RK_S32 size;
|
||||
RK_U32 capability;
|
||||
RK_U32 nFlags;
|
||||
} VideoPacket_t;
|
||||
|
||||
typedef struct DecoderOut {
|
||||
RK_U8 *data;
|
||||
RK_U32 size;
|
||||
RK_S64 timeUs;
|
||||
RK_S32 nFlags;
|
||||
} DecoderOut_t;
|
||||
|
||||
typedef struct ParserOut {
|
||||
RK_U8 *data;
|
||||
RK_U32 size;
|
||||
RK_S64 timeUs;
|
||||
RK_U32 nFlags;
|
||||
RK_U32 width;
|
||||
RK_U32 height;
|
||||
} ParserOut_t;
|
||||
|
||||
typedef struct EncInputStream {
|
||||
RK_U8 *buf;
|
||||
RK_S32 size;
|
||||
RK_U32 bufPhyAddr;
|
||||
RK_S64 timeUs;
|
||||
RK_U32 nFlags;
|
||||
} EncInputStream_t;
|
||||
|
||||
typedef struct EncoderOut {
|
||||
RK_U8 *data;
|
||||
RK_S32 size;
|
||||
RK_S64 timeUs;
|
||||
RK_S32 keyFrame;
|
||||
|
||||
} EncoderOut_t;
|
||||
|
||||
/*
|
||||
* @brief Enumeration used to define the possible video compression codings.
|
||||
* @note This essentially refers to file extensions. If the coding is
|
||||
* being used to specify the ENCODE type, then additional work
|
||||
* must be done to configure the exact flavor of the compression
|
||||
* to be used. For decode cases where the user application can
|
||||
* not differentiate between MPEG-4 and H.264 bit streams, it is
|
||||
* up to the codec to handle this.
|
||||
*
|
||||
* sync with the omx_video.h
|
||||
*/
|
||||
typedef enum OMX_RK_VIDEO_CODINGTYPE {
|
||||
OMX_RK_VIDEO_CodingUnused, /**< Value when coding is N/A */
|
||||
OMX_RK_VIDEO_CodingAutoDetect, /**< Autodetection of coding type */
|
||||
OMX_RK_VIDEO_CodingMPEG2, /**< AKA: H.262 */
|
||||
OMX_RK_VIDEO_CodingH263, /**< H.263 */
|
||||
OMX_RK_VIDEO_CodingMPEG4, /**< MPEG-4 */
|
||||
OMX_RK_VIDEO_CodingWMV, /**< Windows Media Video (WMV1,WMV2,WMV3)*/
|
||||
OMX_RK_VIDEO_CodingRV, /**< all versions of Real Video */
|
||||
OMX_RK_VIDEO_CodingAVC, /**< H.264/AVC */
|
||||
OMX_RK_VIDEO_CodingMJPEG, /**< Motion JPEG */
|
||||
OMX_RK_VIDEO_CodingVP8, /**< VP8 */
|
||||
OMX_RK_VIDEO_CodingVP9, /**< VP9 */
|
||||
OMX_RK_VIDEO_CodingVC1 = 0x01000000, /**< Windows Media Video (WMV1,WMV2,WMV3)*/
|
||||
OMX_RK_VIDEO_CodingFLV1, /**< Sorenson H.263 */
|
||||
OMX_RK_VIDEO_CodingDIVX3, /**< DIVX3 */
|
||||
OMX_RK_VIDEO_CodingVP6,
|
||||
OMX_RK_VIDEO_CodingHEVC, /**< H.265/HEVC */
|
||||
OMX_RK_VIDEO_CodingAVSPLUS, /**< AVS+ profile 0x48 */
|
||||
OMX_RK_VIDEO_CodingAVS, /**< AVS profile 0x20 */
|
||||
OMX_RK_VIDEO_CodingAVS2, /**< AVS2 */
|
||||
OMX_RK_VIDEO_CodingAV1, /**< av1 */
|
||||
OMX_RK_VIDEO_CodingKhronosExtensions = 0x6F000000, /**< Reserved region for introducing Khronos Standard Extensions */
|
||||
OMX_RK_VIDEO_CodingVendorStartUnused = 0x7F000000, /**< Reserved region for introducing Vendor Extensions */
|
||||
OMX_RK_VIDEO_CodingMax = 0x7FFFFFFF
|
||||
} OMX_RK_VIDEO_CODINGTYPE;
|
||||
|
||||
typedef enum CODEC_TYPE {
|
||||
CODEC_NONE,
|
||||
CODEC_DECODER,
|
||||
CODEC_ENCODER,
|
||||
CODEC_BUTT,
|
||||
} CODEC_TYPE;
|
||||
|
||||
typedef enum VPU_API_ERR {
|
||||
VPU_API_OK = 0,
|
||||
VPU_API_ERR_UNKNOW = -1,
|
||||
VPU_API_ERR_BASE = -1000,
|
||||
VPU_API_ERR_LIST_STREAM = VPU_API_ERR_BASE - 1,
|
||||
VPU_API_ERR_INIT = VPU_API_ERR_BASE - 2,
|
||||
VPU_API_ERR_VPU_CODEC_INIT = VPU_API_ERR_BASE - 3,
|
||||
VPU_API_ERR_STREAM = VPU_API_ERR_BASE - 4,
|
||||
VPU_API_ERR_FATAL_THREAD = VPU_API_ERR_BASE - 5,
|
||||
VPU_API_EOS_STREAM_REACHED = VPU_API_ERR_BASE - 11,
|
||||
|
||||
VPU_API_ERR_BUTT,
|
||||
} VPU_API_ERR;
|
||||
|
||||
typedef enum VPU_FRAME_ERR {
|
||||
VPU_FRAME_ERR_UNKNOW = 0x0001,
|
||||
VPU_FRAME_ERR_UNSUPPORT = 0x0002,
|
||||
|
||||
} VPU_FRAME_ERR;
|
||||
|
||||
typedef struct EncParameter {
|
||||
RK_S32 width;
|
||||
RK_S32 height;
|
||||
RK_S32 rc_mode; /* 0 - CQP mode; 1 - CBR mode; 2 - FIXQP mode*/
|
||||
RK_S32 bitRate; /* target bitrate */
|
||||
RK_S32 framerate;
|
||||
RK_S32 qp;
|
||||
RK_S32 enableCabac;
|
||||
RK_S32 cabacInitIdc;
|
||||
RK_S32 format;
|
||||
RK_S32 intraPicRate;
|
||||
RK_S32 framerateout;
|
||||
RK_S32 profileIdc;
|
||||
RK_S32 levelIdc;
|
||||
RK_S32 reserved[3];
|
||||
} EncParameter_t;
|
||||
|
||||
typedef struct EXtraCfg {
|
||||
RK_S32 vc1extra_size;
|
||||
RK_S32 vp6codeid;
|
||||
RK_S32 tsformat;
|
||||
RK_U32 ori_vpu; /* use origin vpu framework */
|
||||
/* below used in decode */
|
||||
RK_U32 mpp_mode; /* use mpp framework */
|
||||
RK_U32 bit_depth; /* 8 or 10 bit */
|
||||
RK_U32 yuv_format; /* 0:420 1:422 2:444 */
|
||||
RK_U32 reserved[16];
|
||||
} EXtraCfg_t;
|
||||
|
||||
/**
|
||||
* @brief vpu function interface
|
||||
*/
|
||||
typedef struct VpuCodecContext {
|
||||
void* vpuApiObj;
|
||||
|
||||
CODEC_TYPE codecType;
|
||||
OMX_RK_VIDEO_CODINGTYPE videoCoding;
|
||||
|
||||
RK_U32 width;
|
||||
RK_U32 height;
|
||||
void *extradata;
|
||||
RK_S32 extradata_size;
|
||||
|
||||
RK_U8 enableparsing;
|
||||
|
||||
RK_S32 no_thread;
|
||||
EXtraCfg_t extra_cfg;
|
||||
|
||||
void* private_data;
|
||||
|
||||
/*
|
||||
** 1: error state(not working) 0: working
|
||||
*/
|
||||
RK_S32 decoder_err;
|
||||
|
||||
|
||||
/**
|
||||
* Allocate and initialize an VpuCodecContext.
|
||||
*
|
||||
* @param ctx The context of vpu api, allocated in this function.
|
||||
* @param extraData The extra data of codec, some codecs need / can
|
||||
* use extradata like Huffman tables, also live VC1 codec can
|
||||
* use extradata to initialize itself.
|
||||
* @param extra_size The size of extra data.
|
||||
*
|
||||
* @return 0 for init success, others for failure.
|
||||
* @note check whether ctx has been allocated success after you do init.
|
||||
*/
|
||||
RK_S32 (*init)(struct VpuCodecContext *ctx, RK_U8 *extraData, RK_U32 extra_size);
|
||||
/**
|
||||
* @brief both send video stream packet to decoder and get video frame from
|
||||
* decoder at the same time
|
||||
* @param ctx The context of vpu codec
|
||||
* @param pkt[in] Stream to be decoded
|
||||
* @param aDecOut[out] Decoding frame
|
||||
* @return 0 for decode success, others for failure.
|
||||
*/
|
||||
RK_S32 (*decode)(struct VpuCodecContext *ctx, VideoPacket_t *pkt, DecoderOut_t *aDecOut);
|
||||
/**
|
||||
* @brief both send video frame to encoder and get encoded video stream from
|
||||
* encoder at the same time.
|
||||
* @param ctx The context of vpu codec
|
||||
* @param aEncInStrm[in] Frame to be encoded
|
||||
* @param aEncOut[out] Encoding stream
|
||||
* @return 0 for encode success, others for failure.
|
||||
*/
|
||||
RK_S32 (*encode)(struct VpuCodecContext *ctx, EncInputStream_t *aEncInStrm, EncoderOut_t *aEncOut);
|
||||
/**
|
||||
* @brief flush codec while do fast forward playing.
|
||||
* @param ctx The context of vpu codec
|
||||
* @return 0 for flush success, others for failure.
|
||||
*/
|
||||
RK_S32 (*flush)(struct VpuCodecContext *ctx);
|
||||
RK_S32 (*control)(struct VpuCodecContext *ctx, VPU_API_CMD cmdType, void* param);
|
||||
/**
|
||||
* @brief send video stream packet to decoder only, async interface
|
||||
* @param ctx The context of vpu codec
|
||||
* @param pkt Stream to be decoded
|
||||
* @return 0 for success, others for failure.
|
||||
*/
|
||||
RK_S32 (*decode_sendstream)(struct VpuCodecContext *ctx, VideoPacket_t *pkt);
|
||||
/**
|
||||
* @brief get video frame from decoder only, async interface
|
||||
* @param ctx The context of vpu codec
|
||||
* @param aDecOut Decoding frame
|
||||
* @return 0 for success, others for failure.
|
||||
*/
|
||||
RK_S32 (*decode_getframe)(struct VpuCodecContext *ctx, DecoderOut_t *aDecOut);
|
||||
/**
|
||||
* @brief send video frame to encoder only, async interface
|
||||
* @param ctx The context of vpu codec
|
||||
* @param aEncInStrm Frame to be encoded
|
||||
* @return 0 for success, others for failure.
|
||||
*/
|
||||
RK_S32 (*encoder_sendframe)(struct VpuCodecContext *ctx, EncInputStream_t *aEncInStrm);
|
||||
/**
|
||||
* @brief get encoded video packet from encoder only, async interface
|
||||
* @param ctx The context of vpu codec
|
||||
* @param aEncOut Encoding stream
|
||||
* @return 0 for success, others for failure.
|
||||
*/
|
||||
RK_S32 (*encoder_getstream)(struct VpuCodecContext *ctx, EncoderOut_t *aEncOut);
|
||||
} VpuCodecContext_t;
|
||||
|
||||
/* allocated vpu codec context */
|
||||
#ifdef __cplusplus
|
||||
extern "C"
|
||||
{
|
||||
#endif
|
||||
|
||||
/**
|
||||
* @brief open context of vpu
|
||||
* @param ctx pointer of vpu codec context
|
||||
*/
|
||||
RK_S32 vpu_open_context(struct VpuCodecContext **ctx);
|
||||
/**
|
||||
* @brief close context of vpu
|
||||
* @param ctx pointer of vpu codec context
|
||||
*/
|
||||
RK_S32 vpu_close_context(struct VpuCodecContext **ctx);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
/*
|
||||
* vpu_mem api
|
||||
*/
|
||||
#define vpu_display_mem_pool_FIELDS \
|
||||
RK_S32 (*commit_hdl)(vpu_display_mem_pool *p, RK_S32 hdl, RK_S32 size); \
|
||||
void* (*get_free)(vpu_display_mem_pool *p); \
|
||||
RK_S32 (*inc_used)(vpu_display_mem_pool *p, void *hdl); \
|
||||
RK_S32 (*put_used)(vpu_display_mem_pool *p, void *hdl); \
|
||||
RK_S32 (*reset)(vpu_display_mem_pool *p); \
|
||||
RK_S32 (*get_unused_num)(vpu_display_mem_pool *p); \
|
||||
RK_S32 buff_size;\
|
||||
float version; \
|
||||
RK_S32 res[18];
|
||||
|
||||
typedef struct vpu_display_mem_pool vpu_display_mem_pool;
|
||||
|
||||
struct vpu_display_mem_pool {
|
||||
vpu_display_mem_pool_FIELDS
|
||||
};
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C"
|
||||
{
|
||||
#endif
|
||||
|
||||
/*
|
||||
* vpu memory handle interface
|
||||
*/
|
||||
RK_S32 VPUMemJudgeIommu(void);
|
||||
RK_S32 VPUMallocLinear(VPUMemLinear_t *p, RK_U32 size);
|
||||
RK_S32 VPUFreeLinear(VPUMemLinear_t *p);
|
||||
RK_S32 VPUMemDuplicate(VPUMemLinear_t *dst, VPUMemLinear_t *src);
|
||||
RK_S32 VPUMemLink(VPUMemLinear_t *p);
|
||||
RK_S32 VPUMemFlush(VPUMemLinear_t *p);
|
||||
RK_S32 VPUMemClean(VPUMemLinear_t *p);
|
||||
RK_S32 VPUMemInvalidate(VPUMemLinear_t *p);
|
||||
RK_S32 VPUMemGetFD(VPUMemLinear_t *p);
|
||||
RK_S32 VPUMallocLinearFromRender(VPUMemLinear_t *p, RK_U32 size, void *ctx);
|
||||
|
||||
/*
|
||||
* vpu memory allocator and manager interface
|
||||
*/
|
||||
vpu_display_mem_pool* open_vpu_memory_pool(void);
|
||||
void close_vpu_memory_pool(vpu_display_mem_pool *p);
|
||||
int create_vpu_memory_pool_allocator(vpu_display_mem_pool **ipool, int num, int size);
|
||||
void release_vpu_memory_pool_allocator(vpu_display_mem_pool *ipool);
|
||||
|
||||
#ifdef __cplusplus
|
||||
}
|
||||
#endif
|
||||
|
||||
#endif /*__VPU_API_H__*/
|
||||
BIN
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/opencv/opencv-linux-aarch64/bin/opencv_version
vendored
Normal file
BIN
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/opencv/opencv-linux-aarch64/bin/opencv_version
vendored
Normal file
Binary file not shown.
|
|
@ -0,0 +1,20 @@
|
|||
#!/bin/bash
|
||||
|
||||
SCRIPT_DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null && pwd )"
|
||||
|
||||
[[ ! "${OPENCV_QUIET}" ]] && ( echo "Setting vars for OpenCV 3.4.5" )
|
||||
export LD_LIBRARY_PATH="$SCRIPT_DIR/../lib:$LD_LIBRARY_PATH"
|
||||
|
||||
if [[ ! "$OPENCV_SKIP_PYTHON" ]]; then
|
||||
PYTHONPATH_OPENCV="$SCRIPT_DIR/python_loader_is_not_installed"
|
||||
[[ ! "${OPENCV_QUIET}" ]] && ( echo "Append PYTHONPATH: ${PYTHONPATH_OPENCV}" )
|
||||
export PYTHONPATH="${PYTHONPATH_OPENCV}:$PYTHONPATH"
|
||||
fi
|
||||
|
||||
# Don't exec in "sourced" mode
|
||||
if [[ "${BASH_SOURCE[0]}" == "${0}" ]]; then
|
||||
if [[ $# -ne 0 ]]; then
|
||||
[[ ! "${OPENCV_QUIET}" && "${OPENCV_VERBOSE}" ]] && ( echo "Executing: $*" )
|
||||
exec "$@"
|
||||
fi
|
||||
fi
|
||||
73
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/opencv/opencv-linux-aarch64/include/opencv/cv.h
vendored
Normal file
73
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/opencv/opencv-linux-aarch64/include/opencv/cv.h
vendored
Normal file
|
|
@ -0,0 +1,73 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_OLD_CV_H
|
||||
#define OPENCV_OLD_CV_H
|
||||
|
||||
#if defined(_MSC_VER)
|
||||
#define CV_DO_PRAGMA(x) __pragma(x)
|
||||
#define __CVSTR2__(x) #x
|
||||
#define __CVSTR1__(x) __CVSTR2__(x)
|
||||
#define __CVMSVCLOC__ __FILE__ "("__CVSTR1__(__LINE__)") : "
|
||||
#define CV_MSG_PRAGMA(_msg) CV_DO_PRAGMA(message (__CVMSVCLOC__ _msg))
|
||||
#elif defined(__GNUC__)
|
||||
#define CV_DO_PRAGMA(x) _Pragma (#x)
|
||||
#define CV_MSG_PRAGMA(_msg) CV_DO_PRAGMA(message (_msg))
|
||||
#else
|
||||
#define CV_DO_PRAGMA(x)
|
||||
#define CV_MSG_PRAGMA(_msg)
|
||||
#endif
|
||||
#define CV_WARNING(x) CV_MSG_PRAGMA("Warning: " #x)
|
||||
|
||||
//CV_WARNING("This is a deprecated opencv header provided for compatibility. Please include a header from a corresponding opencv module")
|
||||
|
||||
#include "opencv2/core/core_c.h"
|
||||
#include "opencv2/imgproc/imgproc_c.h"
|
||||
#include "opencv2/photo/photo_c.h"
|
||||
#include "opencv2/video/tracking_c.h"
|
||||
#include "opencv2/objdetect/objdetect_c.h"
|
||||
|
||||
#if !defined(CV_IMPL)
|
||||
#define CV_IMPL extern "C"
|
||||
#endif //CV_IMPL
|
||||
|
||||
#endif // __OPENCV_OLD_CV_H_
|
||||
|
|
@ -0,0 +1,60 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_OLD_CV_HPP
|
||||
#define OPENCV_OLD_CV_HPP
|
||||
|
||||
//#if defined(__GNUC__)
|
||||
//#warning "This is a deprecated opencv header provided for compatibility. Please include a header from a corresponding opencv module"
|
||||
//#endif
|
||||
|
||||
#include "cv.h"
|
||||
#include "opencv2/core.hpp"
|
||||
#include "opencv2/imgproc.hpp"
|
||||
#include "opencv2/photo.hpp"
|
||||
#include "opencv2/video.hpp"
|
||||
#include "opencv2/highgui.hpp"
|
||||
#include "opencv2/features2d.hpp"
|
||||
#include "opencv2/calib3d.hpp"
|
||||
#include "opencv2/objdetect.hpp"
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,57 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// Intel License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000, Intel Corporation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of Intel Corporation may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_OLD_AUX_H
|
||||
#define OPENCV_OLD_AUX_H
|
||||
|
||||
//#if defined(__GNUC__)
|
||||
//#warning "This is a deprecated opencv header provided for compatibility. Please include a header from a corresponding opencv module"
|
||||
//#endif
|
||||
|
||||
#include "opencv2/core/core_c.h"
|
||||
#include "opencv2/imgproc/imgproc_c.h"
|
||||
#include "opencv2/photo/photo_c.h"
|
||||
#include "opencv2/video/tracking_c.h"
|
||||
#include "opencv2/objdetect/objdetect_c.h"
|
||||
|
||||
#endif
|
||||
|
||||
/* End of file. */
|
||||
|
|
@ -0,0 +1,52 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// Intel License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000, Intel Corporation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of Intel Corporation may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_OLD_AUX_HPP
|
||||
#define OPENCV_OLD_AUX_HPP
|
||||
|
||||
//#if defined(__GNUC__)
|
||||
//#warning "This is a deprecated opencv header provided for compatibility. Please include a header from a corresponding opencv module"
|
||||
//#endif
|
||||
|
||||
#include "cvaux.h"
|
||||
#include "opencv2/core/utility.hpp"
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,46 @@
|
|||
///////////////////////////////////////////////////////////////////////////////
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to
|
||||
// this license. If you do not agree to this license, do not download,
|
||||
// install, copy or use the software.
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2008, Google, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without
|
||||
// modification, are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of Intel Corporation or contributors may not be used to endorse
|
||||
// or promote products derived from this software without specific
|
||||
// prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is"
|
||||
// and any express or implied warranties, including, but not limited to, the
|
||||
// implied warranties of merchantability and fitness for a particular purpose
|
||||
// are disclaimed. In no event shall the Intel Corporation or contributors be
|
||||
// liable for any direct, indirect, incidental, special, exemplary, or
|
||||
// consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
|
||||
|
||||
#ifndef OPENCV_OLD_WIMAGE_HPP
|
||||
#define OPENCV_OLD_WIMAGE_HPP
|
||||
|
||||
#include "opencv2/core/wimage.hpp"
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,52 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_OLD_CXCORE_H
|
||||
#define OPENCV_OLD_CXCORE_H
|
||||
|
||||
//#if defined(__GNUC__)
|
||||
//#warning "This is a deprecated opencv header provided for compatibility. Please include a header from a corresponding opencv module"
|
||||
//#endif
|
||||
|
||||
#include "opencv2/core/core_c.h"
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,53 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_OLD_CXCORE_HPP
|
||||
#define OPENCV_OLD_CXCORE_HPP
|
||||
|
||||
//#if defined(__GNUC__)
|
||||
//#warning "This is a deprecated opencv header provided for compatibility. Please include a header from a corresponding opencv module"
|
||||
//#endif
|
||||
|
||||
#include "cxcore.h"
|
||||
#include "opencv2/core.hpp"
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,48 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_OLD_EIGEN_HPP
|
||||
#define OPENCV_OLD_EIGEN_HPP
|
||||
|
||||
#include "opencv2/core/eigen.hpp"
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,8 @@
|
|||
#ifndef OPENCV_OLD_CXMISC_H
|
||||
#define OPENCV_OLD_CXMISC_H
|
||||
|
||||
#ifdef __cplusplus
|
||||
# include "opencv2/core/utility.hpp"
|
||||
#endif
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,48 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// Intel License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000, Intel Corporation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of Intel Corporation may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_OLD_HIGHGUI_H
|
||||
#define OPENCV_OLD_HIGHGUI_H
|
||||
|
||||
#include "opencv2/core/core_c.h"
|
||||
#include "opencv2/highgui/highgui_c.h"
|
||||
|
||||
#endif
|
||||
47
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/opencv/opencv-linux-aarch64/include/opencv/ml.h
vendored
Normal file
47
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/opencv/opencv-linux-aarch64/include/opencv/ml.h
vendored
Normal file
|
|
@ -0,0 +1,47 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// Intel License Agreement
|
||||
//
|
||||
// Copyright (C) 2000, Intel Corporation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of Intel Corporation may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_OLD_ML_H
|
||||
#define OPENCV_OLD_ML_H
|
||||
|
||||
#include "opencv2/core/core_c.h"
|
||||
#include "opencv2/ml.hpp"
|
||||
|
||||
#endif
|
||||
2479
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/opencv/opencv-linux-aarch64/include/opencv2/calib3d.hpp
vendored
Normal file
2479
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/opencv/opencv-linux-aarch64/include/opencv2/calib3d.hpp
vendored
Normal file
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,48 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifdef __OPENCV_BUILD
|
||||
#error this is a compatibility header which should not be used inside the OpenCV library
|
||||
#endif
|
||||
|
||||
#include "opencv2/calib3d.hpp"
|
||||
|
|
@ -0,0 +1,427 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CALIB3D_C_H
|
||||
#define OPENCV_CALIB3D_C_H
|
||||
|
||||
#include "opencv2/core/core_c.h"
|
||||
|
||||
#ifdef __cplusplus
|
||||
extern "C" {
|
||||
#endif
|
||||
|
||||
/** @addtogroup calib3d_c
|
||||
@{
|
||||
*/
|
||||
|
||||
/****************************************************************************************\
|
||||
* Camera Calibration, Pose Estimation and Stereo *
|
||||
\****************************************************************************************/
|
||||
|
||||
typedef struct CvPOSITObject CvPOSITObject;
|
||||
|
||||
/* Allocates and initializes CvPOSITObject structure before doing cvPOSIT */
|
||||
CVAPI(CvPOSITObject*) cvCreatePOSITObject( CvPoint3D32f* points, int point_count );
|
||||
|
||||
|
||||
/* Runs POSIT (POSe from ITeration) algorithm for determining 3d position of
|
||||
an object given its model and projection in a weak-perspective case */
|
||||
CVAPI(void) cvPOSIT( CvPOSITObject* posit_object, CvPoint2D32f* image_points,
|
||||
double focal_length, CvTermCriteria criteria,
|
||||
float* rotation_matrix, float* translation_vector);
|
||||
|
||||
/* Releases CvPOSITObject structure */
|
||||
CVAPI(void) cvReleasePOSITObject( CvPOSITObject** posit_object );
|
||||
|
||||
/* updates the number of RANSAC iterations */
|
||||
CVAPI(int) cvRANSACUpdateNumIters( double p, double err_prob,
|
||||
int model_points, int max_iters );
|
||||
|
||||
CVAPI(void) cvConvertPointsHomogeneous( const CvMat* src, CvMat* dst );
|
||||
|
||||
/* Calculates fundamental matrix given a set of corresponding points */
|
||||
#define CV_FM_7POINT 1
|
||||
#define CV_FM_8POINT 2
|
||||
|
||||
#define CV_LMEDS 4
|
||||
#define CV_RANSAC 8
|
||||
|
||||
#define CV_FM_LMEDS_ONLY CV_LMEDS
|
||||
#define CV_FM_RANSAC_ONLY CV_RANSAC
|
||||
#define CV_FM_LMEDS CV_LMEDS
|
||||
#define CV_FM_RANSAC CV_RANSAC
|
||||
|
||||
enum
|
||||
{
|
||||
CV_ITERATIVE = 0,
|
||||
CV_EPNP = 1, // F.Moreno-Noguer, V.Lepetit and P.Fua "EPnP: Efficient Perspective-n-Point Camera Pose Estimation"
|
||||
CV_P3P = 2, // X.S. Gao, X.-R. Hou, J. Tang, H.-F. Chang; "Complete Solution Classification for the Perspective-Three-Point Problem"
|
||||
CV_DLS = 3 // Joel A. Hesch and Stergios I. Roumeliotis. "A Direct Least-Squares (DLS) Method for PnP"
|
||||
};
|
||||
|
||||
CVAPI(int) cvFindFundamentalMat( const CvMat* points1, const CvMat* points2,
|
||||
CvMat* fundamental_matrix,
|
||||
int method CV_DEFAULT(CV_FM_RANSAC),
|
||||
double param1 CV_DEFAULT(3.), double param2 CV_DEFAULT(0.99),
|
||||
CvMat* status CV_DEFAULT(NULL) );
|
||||
|
||||
/* For each input point on one of images
|
||||
computes parameters of the corresponding
|
||||
epipolar line on the other image */
|
||||
CVAPI(void) cvComputeCorrespondEpilines( const CvMat* points,
|
||||
int which_image,
|
||||
const CvMat* fundamental_matrix,
|
||||
CvMat* correspondent_lines );
|
||||
|
||||
/* Triangulation functions */
|
||||
|
||||
CVAPI(void) cvTriangulatePoints(CvMat* projMatr1, CvMat* projMatr2,
|
||||
CvMat* projPoints1, CvMat* projPoints2,
|
||||
CvMat* points4D);
|
||||
|
||||
CVAPI(void) cvCorrectMatches(CvMat* F, CvMat* points1, CvMat* points2,
|
||||
CvMat* new_points1, CvMat* new_points2);
|
||||
|
||||
|
||||
/* Computes the optimal new camera matrix according to the free scaling parameter alpha:
|
||||
alpha=0 - only valid pixels will be retained in the undistorted image
|
||||
alpha=1 - all the source image pixels will be retained in the undistorted image
|
||||
*/
|
||||
CVAPI(void) cvGetOptimalNewCameraMatrix( const CvMat* camera_matrix,
|
||||
const CvMat* dist_coeffs,
|
||||
CvSize image_size, double alpha,
|
||||
CvMat* new_camera_matrix,
|
||||
CvSize new_imag_size CV_DEFAULT(cvSize(0,0)),
|
||||
CvRect* valid_pixel_ROI CV_DEFAULT(0),
|
||||
int center_principal_point CV_DEFAULT(0));
|
||||
|
||||
/* Converts rotation vector to rotation matrix or vice versa */
|
||||
CVAPI(int) cvRodrigues2( const CvMat* src, CvMat* dst,
|
||||
CvMat* jacobian CV_DEFAULT(0) );
|
||||
|
||||
/* Finds perspective transformation between the object plane and image (view) plane */
|
||||
CVAPI(int) cvFindHomography( const CvMat* src_points,
|
||||
const CvMat* dst_points,
|
||||
CvMat* homography,
|
||||
int method CV_DEFAULT(0),
|
||||
double ransacReprojThreshold CV_DEFAULT(3),
|
||||
CvMat* mask CV_DEFAULT(0),
|
||||
int maxIters CV_DEFAULT(2000),
|
||||
double confidence CV_DEFAULT(0.995));
|
||||
|
||||
/* Computes RQ decomposition for 3x3 matrices */
|
||||
CVAPI(void) cvRQDecomp3x3( const CvMat *matrixM, CvMat *matrixR, CvMat *matrixQ,
|
||||
CvMat *matrixQx CV_DEFAULT(NULL),
|
||||
CvMat *matrixQy CV_DEFAULT(NULL),
|
||||
CvMat *matrixQz CV_DEFAULT(NULL),
|
||||
CvPoint3D64f *eulerAngles CV_DEFAULT(NULL));
|
||||
|
||||
/* Computes projection matrix decomposition */
|
||||
CVAPI(void) cvDecomposeProjectionMatrix( const CvMat *projMatr, CvMat *calibMatr,
|
||||
CvMat *rotMatr, CvMat *posVect,
|
||||
CvMat *rotMatrX CV_DEFAULT(NULL),
|
||||
CvMat *rotMatrY CV_DEFAULT(NULL),
|
||||
CvMat *rotMatrZ CV_DEFAULT(NULL),
|
||||
CvPoint3D64f *eulerAngles CV_DEFAULT(NULL));
|
||||
|
||||
/* Computes d(AB)/dA and d(AB)/dB */
|
||||
CVAPI(void) cvCalcMatMulDeriv( const CvMat* A, const CvMat* B, CvMat* dABdA, CvMat* dABdB );
|
||||
|
||||
/* Computes r3 = rodrigues(rodrigues(r2)*rodrigues(r1)),
|
||||
t3 = rodrigues(r2)*t1 + t2 and the respective derivatives */
|
||||
CVAPI(void) cvComposeRT( const CvMat* _rvec1, const CvMat* _tvec1,
|
||||
const CvMat* _rvec2, const CvMat* _tvec2,
|
||||
CvMat* _rvec3, CvMat* _tvec3,
|
||||
CvMat* dr3dr1 CV_DEFAULT(0), CvMat* dr3dt1 CV_DEFAULT(0),
|
||||
CvMat* dr3dr2 CV_DEFAULT(0), CvMat* dr3dt2 CV_DEFAULT(0),
|
||||
CvMat* dt3dr1 CV_DEFAULT(0), CvMat* dt3dt1 CV_DEFAULT(0),
|
||||
CvMat* dt3dr2 CV_DEFAULT(0), CvMat* dt3dt2 CV_DEFAULT(0) );
|
||||
|
||||
/* Projects object points to the view plane using
|
||||
the specified extrinsic and intrinsic camera parameters */
|
||||
CVAPI(void) cvProjectPoints2( const CvMat* object_points, const CvMat* rotation_vector,
|
||||
const CvMat* translation_vector, const CvMat* camera_matrix,
|
||||
const CvMat* distortion_coeffs, CvMat* image_points,
|
||||
CvMat* dpdrot CV_DEFAULT(NULL), CvMat* dpdt CV_DEFAULT(NULL),
|
||||
CvMat* dpdf CV_DEFAULT(NULL), CvMat* dpdc CV_DEFAULT(NULL),
|
||||
CvMat* dpddist CV_DEFAULT(NULL),
|
||||
double aspect_ratio CV_DEFAULT(0));
|
||||
|
||||
/* Finds extrinsic camera parameters from
|
||||
a few known corresponding point pairs and intrinsic parameters */
|
||||
CVAPI(void) cvFindExtrinsicCameraParams2( const CvMat* object_points,
|
||||
const CvMat* image_points,
|
||||
const CvMat* camera_matrix,
|
||||
const CvMat* distortion_coeffs,
|
||||
CvMat* rotation_vector,
|
||||
CvMat* translation_vector,
|
||||
int use_extrinsic_guess CV_DEFAULT(0) );
|
||||
|
||||
/* Computes initial estimate of the intrinsic camera parameters
|
||||
in case of planar calibration target (e.g. chessboard) */
|
||||
CVAPI(void) cvInitIntrinsicParams2D( const CvMat* object_points,
|
||||
const CvMat* image_points,
|
||||
const CvMat* npoints, CvSize image_size,
|
||||
CvMat* camera_matrix,
|
||||
double aspect_ratio CV_DEFAULT(1.) );
|
||||
|
||||
#define CV_CALIB_CB_ADAPTIVE_THRESH 1
|
||||
#define CV_CALIB_CB_NORMALIZE_IMAGE 2
|
||||
#define CV_CALIB_CB_FILTER_QUADS 4
|
||||
#define CV_CALIB_CB_FAST_CHECK 8
|
||||
|
||||
// Performs a fast check if a chessboard is in the input image. This is a workaround to
|
||||
// a problem of cvFindChessboardCorners being slow on images with no chessboard
|
||||
// - src: input image
|
||||
// - size: chessboard size
|
||||
// Returns 1 if a chessboard can be in this image and findChessboardCorners should be called,
|
||||
// 0 if there is no chessboard, -1 in case of error
|
||||
CVAPI(int) cvCheckChessboard(IplImage* src, CvSize size);
|
||||
|
||||
/* Detects corners on a chessboard calibration pattern */
|
||||
CVAPI(int) cvFindChessboardCorners( const void* image, CvSize pattern_size,
|
||||
CvPoint2D32f* corners,
|
||||
int* corner_count CV_DEFAULT(NULL),
|
||||
int flags CV_DEFAULT(CV_CALIB_CB_ADAPTIVE_THRESH+CV_CALIB_CB_NORMALIZE_IMAGE) );
|
||||
|
||||
/* Draws individual chessboard corners or the whole chessboard detected */
|
||||
CVAPI(void) cvDrawChessboardCorners( CvArr* image, CvSize pattern_size,
|
||||
CvPoint2D32f* corners,
|
||||
int count, int pattern_was_found );
|
||||
|
||||
#define CV_CALIB_USE_INTRINSIC_GUESS 1
|
||||
#define CV_CALIB_FIX_ASPECT_RATIO 2
|
||||
#define CV_CALIB_FIX_PRINCIPAL_POINT 4
|
||||
#define CV_CALIB_ZERO_TANGENT_DIST 8
|
||||
#define CV_CALIB_FIX_FOCAL_LENGTH 16
|
||||
#define CV_CALIB_FIX_K1 32
|
||||
#define CV_CALIB_FIX_K2 64
|
||||
#define CV_CALIB_FIX_K3 128
|
||||
#define CV_CALIB_FIX_K4 2048
|
||||
#define CV_CALIB_FIX_K5 4096
|
||||
#define CV_CALIB_FIX_K6 8192
|
||||
#define CV_CALIB_RATIONAL_MODEL 16384
|
||||
#define CV_CALIB_THIN_PRISM_MODEL 32768
|
||||
#define CV_CALIB_FIX_S1_S2_S3_S4 65536
|
||||
#define CV_CALIB_TILTED_MODEL 262144
|
||||
#define CV_CALIB_FIX_TAUX_TAUY 524288
|
||||
#define CV_CALIB_FIX_TANGENT_DIST 2097152
|
||||
|
||||
#define CV_CALIB_NINTRINSIC 18
|
||||
|
||||
/* Finds intrinsic and extrinsic camera parameters
|
||||
from a few views of known calibration pattern */
|
||||
CVAPI(double) cvCalibrateCamera2( const CvMat* object_points,
|
||||
const CvMat* image_points,
|
||||
const CvMat* point_counts,
|
||||
CvSize image_size,
|
||||
CvMat* camera_matrix,
|
||||
CvMat* distortion_coeffs,
|
||||
CvMat* rotation_vectors CV_DEFAULT(NULL),
|
||||
CvMat* translation_vectors CV_DEFAULT(NULL),
|
||||
int flags CV_DEFAULT(0),
|
||||
CvTermCriteria term_crit CV_DEFAULT(cvTermCriteria(
|
||||
CV_TERMCRIT_ITER+CV_TERMCRIT_EPS,30,DBL_EPSILON)) );
|
||||
|
||||
/* Computes various useful characteristics of the camera from the data computed by
|
||||
cvCalibrateCamera2 */
|
||||
CVAPI(void) cvCalibrationMatrixValues( const CvMat *camera_matrix,
|
||||
CvSize image_size,
|
||||
double aperture_width CV_DEFAULT(0),
|
||||
double aperture_height CV_DEFAULT(0),
|
||||
double *fovx CV_DEFAULT(NULL),
|
||||
double *fovy CV_DEFAULT(NULL),
|
||||
double *focal_length CV_DEFAULT(NULL),
|
||||
CvPoint2D64f *principal_point CV_DEFAULT(NULL),
|
||||
double *pixel_aspect_ratio CV_DEFAULT(NULL));
|
||||
|
||||
#define CV_CALIB_FIX_INTRINSIC 256
|
||||
#define CV_CALIB_SAME_FOCAL_LENGTH 512
|
||||
|
||||
/* Computes the transformation from one camera coordinate system to another one
|
||||
from a few correspondent views of the same calibration target. Optionally, calibrates
|
||||
both cameras */
|
||||
CVAPI(double) cvStereoCalibrate( const CvMat* object_points, const CvMat* image_points1,
|
||||
const CvMat* image_points2, const CvMat* npoints,
|
||||
CvMat* camera_matrix1, CvMat* dist_coeffs1,
|
||||
CvMat* camera_matrix2, CvMat* dist_coeffs2,
|
||||
CvSize image_size, CvMat* R, CvMat* T,
|
||||
CvMat* E CV_DEFAULT(0), CvMat* F CV_DEFAULT(0),
|
||||
int flags CV_DEFAULT(CV_CALIB_FIX_INTRINSIC),
|
||||
CvTermCriteria term_crit CV_DEFAULT(cvTermCriteria(
|
||||
CV_TERMCRIT_ITER+CV_TERMCRIT_EPS,30,1e-6)) );
|
||||
|
||||
#define CV_CALIB_ZERO_DISPARITY 1024
|
||||
|
||||
/* Computes 3D rotations (+ optional shift) for each camera coordinate system to make both
|
||||
views parallel (=> to make all the epipolar lines horizontal or vertical) */
|
||||
CVAPI(void) cvStereoRectify( const CvMat* camera_matrix1, const CvMat* camera_matrix2,
|
||||
const CvMat* dist_coeffs1, const CvMat* dist_coeffs2,
|
||||
CvSize image_size, const CvMat* R, const CvMat* T,
|
||||
CvMat* R1, CvMat* R2, CvMat* P1, CvMat* P2,
|
||||
CvMat* Q CV_DEFAULT(0),
|
||||
int flags CV_DEFAULT(CV_CALIB_ZERO_DISPARITY),
|
||||
double alpha CV_DEFAULT(-1),
|
||||
CvSize new_image_size CV_DEFAULT(cvSize(0,0)),
|
||||
CvRect* valid_pix_ROI1 CV_DEFAULT(0),
|
||||
CvRect* valid_pix_ROI2 CV_DEFAULT(0));
|
||||
|
||||
/* Computes rectification transformations for uncalibrated pair of images using a set
|
||||
of point correspondences */
|
||||
CVAPI(int) cvStereoRectifyUncalibrated( const CvMat* points1, const CvMat* points2,
|
||||
const CvMat* F, CvSize img_size,
|
||||
CvMat* H1, CvMat* H2,
|
||||
double threshold CV_DEFAULT(5));
|
||||
|
||||
|
||||
|
||||
/* stereo correspondence parameters and functions */
|
||||
|
||||
#define CV_STEREO_BM_NORMALIZED_RESPONSE 0
|
||||
#define CV_STEREO_BM_XSOBEL 1
|
||||
|
||||
/* Block matching algorithm structure */
|
||||
typedef struct CvStereoBMState
|
||||
{
|
||||
// pre-filtering (normalization of input images)
|
||||
int preFilterType; // =CV_STEREO_BM_NORMALIZED_RESPONSE now
|
||||
int preFilterSize; // averaging window size: ~5x5..21x21
|
||||
int preFilterCap; // the output of pre-filtering is clipped by [-preFilterCap,preFilterCap]
|
||||
|
||||
// correspondence using Sum of Absolute Difference (SAD)
|
||||
int SADWindowSize; // ~5x5..21x21
|
||||
int minDisparity; // minimum disparity (can be negative)
|
||||
int numberOfDisparities; // maximum disparity - minimum disparity (> 0)
|
||||
|
||||
// post-filtering
|
||||
int textureThreshold; // the disparity is only computed for pixels
|
||||
// with textured enough neighborhood
|
||||
int uniquenessRatio; // accept the computed disparity d* only if
|
||||
// SAD(d) >= SAD(d*)*(1 + uniquenessRatio/100.)
|
||||
// for any d != d*+/-1 within the search range.
|
||||
int speckleWindowSize; // disparity variation window
|
||||
int speckleRange; // acceptable range of variation in window
|
||||
|
||||
int trySmallerWindows; // if 1, the results may be more accurate,
|
||||
// at the expense of slower processing
|
||||
CvRect roi1, roi2;
|
||||
int disp12MaxDiff;
|
||||
|
||||
// temporary buffers
|
||||
CvMat* preFilteredImg0;
|
||||
CvMat* preFilteredImg1;
|
||||
CvMat* slidingSumBuf;
|
||||
CvMat* cost;
|
||||
CvMat* disp;
|
||||
} CvStereoBMState;
|
||||
|
||||
#define CV_STEREO_BM_BASIC 0
|
||||
#define CV_STEREO_BM_FISH_EYE 1
|
||||
#define CV_STEREO_BM_NARROW 2
|
||||
|
||||
CVAPI(CvStereoBMState*) cvCreateStereoBMState(int preset CV_DEFAULT(CV_STEREO_BM_BASIC),
|
||||
int numberOfDisparities CV_DEFAULT(0));
|
||||
|
||||
CVAPI(void) cvReleaseStereoBMState( CvStereoBMState** state );
|
||||
|
||||
CVAPI(void) cvFindStereoCorrespondenceBM( const CvArr* left, const CvArr* right,
|
||||
CvArr* disparity, CvStereoBMState* state );
|
||||
|
||||
CVAPI(CvRect) cvGetValidDisparityROI( CvRect roi1, CvRect roi2, int minDisparity,
|
||||
int numberOfDisparities, int SADWindowSize );
|
||||
|
||||
CVAPI(void) cvValidateDisparity( CvArr* disparity, const CvArr* cost,
|
||||
int minDisparity, int numberOfDisparities,
|
||||
int disp12MaxDiff CV_DEFAULT(1) );
|
||||
|
||||
/* Reprojects the computed disparity image to the 3D space using the specified 4x4 matrix */
|
||||
CVAPI(void) cvReprojectImageTo3D( const CvArr* disparityImage,
|
||||
CvArr* _3dImage, const CvMat* Q,
|
||||
int handleMissingValues CV_DEFAULT(0) );
|
||||
|
||||
/** @} calib3d_c */
|
||||
|
||||
#ifdef __cplusplus
|
||||
} // extern "C"
|
||||
|
||||
//////////////////////////////////////////////////////////////////////////////////////////
|
||||
class CV_EXPORTS CvLevMarq
|
||||
{
|
||||
public:
|
||||
CvLevMarq();
|
||||
CvLevMarq( int nparams, int nerrs, CvTermCriteria criteria=
|
||||
cvTermCriteria(CV_TERMCRIT_EPS+CV_TERMCRIT_ITER,30,DBL_EPSILON),
|
||||
bool completeSymmFlag=false );
|
||||
~CvLevMarq();
|
||||
void init( int nparams, int nerrs, CvTermCriteria criteria=
|
||||
cvTermCriteria(CV_TERMCRIT_EPS+CV_TERMCRIT_ITER,30,DBL_EPSILON),
|
||||
bool completeSymmFlag=false );
|
||||
bool update( const CvMat*& param, CvMat*& J, CvMat*& err );
|
||||
bool updateAlt( const CvMat*& param, CvMat*& JtJ, CvMat*& JtErr, double*& errNorm );
|
||||
|
||||
void clear();
|
||||
void step();
|
||||
enum { DONE=0, STARTED=1, CALC_J=2, CHECK_ERR=3 };
|
||||
|
||||
cv::Ptr<CvMat> mask;
|
||||
cv::Ptr<CvMat> prevParam;
|
||||
cv::Ptr<CvMat> param;
|
||||
cv::Ptr<CvMat> J;
|
||||
cv::Ptr<CvMat> err;
|
||||
cv::Ptr<CvMat> JtJ;
|
||||
cv::Ptr<CvMat> JtJN;
|
||||
cv::Ptr<CvMat> JtErr;
|
||||
cv::Ptr<CvMat> JtJV;
|
||||
cv::Ptr<CvMat> JtJW;
|
||||
double prevErrNorm, errNorm;
|
||||
int lambdaLg10;
|
||||
CvTermCriteria criteria;
|
||||
int state;
|
||||
int iters;
|
||||
bool completeSymmFlag;
|
||||
int solveMethod;
|
||||
};
|
||||
|
||||
#endif
|
||||
|
||||
#endif /* OPENCV_CALIB3D_C_H */
|
||||
3285
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/opencv/opencv-linux-aarch64/include/opencv2/core.hpp
vendored
Normal file
3285
Code/RK3588/stereo_yolo/rknpu1/examples/3rdparty/opencv/opencv-linux-aarch64/include/opencv2/core.hpp
vendored
Normal file
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,678 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CORE_AFFINE3_HPP
|
||||
#define OPENCV_CORE_AFFINE3_HPP
|
||||
|
||||
#ifdef __cplusplus
|
||||
|
||||
#include <opencv2/core.hpp>
|
||||
|
||||
namespace cv
|
||||
{
|
||||
|
||||
//! @addtogroup core
|
||||
//! @{
|
||||
|
||||
/** @brief Affine transform
|
||||
*
|
||||
* It represents a 4x4 homogeneous transformation matrix \f$T\f$
|
||||
*
|
||||
* \f[T =
|
||||
* \begin{bmatrix}
|
||||
* R & t\\
|
||||
* 0 & 1\\
|
||||
* \end{bmatrix}
|
||||
* \f]
|
||||
*
|
||||
* where \f$R\f$ is a 3x3 rotation matrix and \f$t\f$ is a 3x1 translation vector.
|
||||
*
|
||||
* You can specify \f$R\f$ either by a 3x3 rotation matrix or by a 3x1 rotation vector,
|
||||
* which is converted to a 3x3 rotation matrix by the Rodrigues formula.
|
||||
*
|
||||
* To construct a matrix \f$T\f$ representing first rotation around the axis \f$r\f$ with rotation
|
||||
* angle \f$|r|\f$ in radian (right hand rule) and then translation by the vector \f$t\f$, you can use
|
||||
*
|
||||
* @code
|
||||
* cv::Vec3f r, t;
|
||||
* cv::Affine3f T(r, t);
|
||||
* @endcode
|
||||
*
|
||||
* If you already have the rotation matrix \f$R\f$, then you can use
|
||||
*
|
||||
* @code
|
||||
* cv::Matx33f R;
|
||||
* cv::Affine3f T(R, t);
|
||||
* @endcode
|
||||
*
|
||||
* To extract the rotation matrix \f$R\f$ from \f$T\f$, use
|
||||
*
|
||||
* @code
|
||||
* cv::Matx33f R = T.rotation();
|
||||
* @endcode
|
||||
*
|
||||
* To extract the translation vector \f$t\f$ from \f$T\f$, use
|
||||
*
|
||||
* @code
|
||||
* cv::Vec3f t = T.translation();
|
||||
* @endcode
|
||||
*
|
||||
* To extract the rotation vector \f$r\f$ from \f$T\f$, use
|
||||
*
|
||||
* @code
|
||||
* cv::Vec3f r = T.rvec();
|
||||
* @endcode
|
||||
*
|
||||
* Note that since the mapping from rotation vectors to rotation matrices
|
||||
* is many to one. The returned rotation vector is not necessarily the one
|
||||
* you used before to set the matrix.
|
||||
*
|
||||
* If you have two transformations \f$T = T_1 * T_2\f$, use
|
||||
*
|
||||
* @code
|
||||
* cv::Affine3f T, T1, T2;
|
||||
* T = T2.concatenate(T1);
|
||||
* @endcode
|
||||
*
|
||||
* To get the inverse transform of \f$T\f$, use
|
||||
*
|
||||
* @code
|
||||
* cv::Affine3f T, T_inv;
|
||||
* T_inv = T.inv();
|
||||
* @endcode
|
||||
*
|
||||
*/
|
||||
template<typename T>
|
||||
class Affine3
|
||||
{
|
||||
public:
|
||||
typedef T float_type;
|
||||
typedef Matx<float_type, 3, 3> Mat3;
|
||||
typedef Matx<float_type, 4, 4> Mat4;
|
||||
typedef Vec<float_type, 3> Vec3;
|
||||
|
||||
//! Default constructor. It represents a 4x4 identity matrix.
|
||||
Affine3();
|
||||
|
||||
//! Augmented affine matrix
|
||||
Affine3(const Mat4& affine);
|
||||
|
||||
/**
|
||||
* The resulting 4x4 matrix is
|
||||
*
|
||||
* \f[
|
||||
* \begin{bmatrix}
|
||||
* R & t\\
|
||||
* 0 & 1\\
|
||||
* \end{bmatrix}
|
||||
* \f]
|
||||
*
|
||||
* @param R 3x3 rotation matrix.
|
||||
* @param t 3x1 translation vector.
|
||||
*/
|
||||
Affine3(const Mat3& R, const Vec3& t = Vec3::all(0));
|
||||
|
||||
/**
|
||||
* Rodrigues vector.
|
||||
*
|
||||
* The last row of the current matrix is set to [0,0,0,1].
|
||||
*
|
||||
* @param rvec 3x1 rotation vector. Its direction indicates the rotation axis and its length
|
||||
* indicates the rotation angle in radian (using right hand rule).
|
||||
* @param t 3x1 translation vector.
|
||||
*/
|
||||
Affine3(const Vec3& rvec, const Vec3& t = Vec3::all(0));
|
||||
|
||||
/**
|
||||
* Combines all constructors above. Supports 4x4, 3x4, 3x3, 1x3, 3x1 sizes of data matrix.
|
||||
*
|
||||
* The last row of the current matrix is set to [0,0,0,1] when data is not 4x4.
|
||||
*
|
||||
* @param data 1-channel matrix.
|
||||
* when it is 4x4, it is copied to the current matrix and t is not used.
|
||||
* When it is 3x4, it is copied to the upper part 3x4 of the current matrix and t is not used.
|
||||
* When it is 3x3, it is copied to the upper left 3x3 part of the current matrix.
|
||||
* When it is 3x1 or 1x3, it is treated as a rotation vector and the Rodrigues formula is used
|
||||
* to compute a 3x3 rotation matrix.
|
||||
* @param t 3x1 translation vector. It is used only when data is neither 4x4 nor 3x4.
|
||||
*/
|
||||
explicit Affine3(const Mat& data, const Vec3& t = Vec3::all(0));
|
||||
|
||||
//! From 16-element array
|
||||
explicit Affine3(const float_type* vals);
|
||||
|
||||
//! Create an 4x4 identity transform
|
||||
static Affine3 Identity();
|
||||
|
||||
/**
|
||||
* Rotation matrix.
|
||||
*
|
||||
* Copy the rotation matrix to the upper left 3x3 part of the current matrix.
|
||||
* The remaining elements of the current matrix are not changed.
|
||||
*
|
||||
* @param R 3x3 rotation matrix.
|
||||
*
|
||||
*/
|
||||
void rotation(const Mat3& R);
|
||||
|
||||
/**
|
||||
* Rodrigues vector.
|
||||
*
|
||||
* It sets the upper left 3x3 part of the matrix. The remaining part is unaffected.
|
||||
*
|
||||
* @param rvec 3x1 rotation vector. The direction indicates the rotation axis and
|
||||
* its length indicates the rotation angle in radian (using the right thumb convention).
|
||||
*/
|
||||
void rotation(const Vec3& rvec);
|
||||
|
||||
/**
|
||||
* Combines rotation methods above. Supports 3x3, 1x3, 3x1 sizes of data matrix.
|
||||
*
|
||||
* It sets the upper left 3x3 part of the matrix. The remaining part is unaffected.
|
||||
*
|
||||
* @param data 1-channel matrix.
|
||||
* When it is a 3x3 matrix, it sets the upper left 3x3 part of the current matrix.
|
||||
* When it is a 1x3 or 3x1 matrix, it is used as a rotation vector. The Rodrigues formula
|
||||
* is used to compute the rotation matrix and sets the upper left 3x3 part of the current matrix.
|
||||
*/
|
||||
void rotation(const Mat& data);
|
||||
|
||||
/**
|
||||
* Copy the 3x3 matrix L to the upper left part of the current matrix
|
||||
*
|
||||
* It sets the upper left 3x3 part of the matrix. The remaining part is unaffected.
|
||||
*
|
||||
* @param L 3x3 matrix.
|
||||
*/
|
||||
void linear(const Mat3& L);
|
||||
|
||||
/**
|
||||
* Copy t to the first three elements of the last column of the current matrix
|
||||
*
|
||||
* It sets the upper right 3x1 part of the matrix. The remaining part is unaffected.
|
||||
*
|
||||
* @param t 3x1 translation vector.
|
||||
*/
|
||||
void translation(const Vec3& t);
|
||||
|
||||
//! @return the upper left 3x3 part
|
||||
Mat3 rotation() const;
|
||||
|
||||
//! @return the upper left 3x3 part
|
||||
Mat3 linear() const;
|
||||
|
||||
//! @return the upper right 3x1 part
|
||||
Vec3 translation() const;
|
||||
|
||||
//! Rodrigues vector.
|
||||
//! @return a vector representing the upper left 3x3 rotation matrix of the current matrix.
|
||||
//! @warning Since the mapping between rotation vectors and rotation matrices is many to one,
|
||||
//! this function returns only one rotation vector that represents the current rotation matrix,
|
||||
//! which is not necessarily the same one set by `rotation(const Vec3& rvec)`.
|
||||
Vec3 rvec() const;
|
||||
|
||||
//! @return the inverse of the current matrix.
|
||||
Affine3 inv(int method = cv::DECOMP_SVD) const;
|
||||
|
||||
//! a.rotate(R) is equivalent to Affine(R, 0) * a;
|
||||
Affine3 rotate(const Mat3& R) const;
|
||||
|
||||
//! a.rotate(rvec) is equivalent to Affine(rvec, 0) * a;
|
||||
Affine3 rotate(const Vec3& rvec) const;
|
||||
|
||||
//! a.translate(t) is equivalent to Affine(E, t) * a, where E is an identity matrix
|
||||
Affine3 translate(const Vec3& t) const;
|
||||
|
||||
//! a.concatenate(affine) is equivalent to affine * a;
|
||||
Affine3 concatenate(const Affine3& affine) const;
|
||||
|
||||
template <typename Y> operator Affine3<Y>() const;
|
||||
|
||||
template <typename Y> Affine3<Y> cast() const;
|
||||
|
||||
Mat4 matrix;
|
||||
|
||||
#if defined EIGEN_WORLD_VERSION && defined EIGEN_GEOMETRY_MODULE_H
|
||||
Affine3(const Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)>& affine);
|
||||
Affine3(const Eigen::Transform<T, 3, Eigen::Affine>& affine);
|
||||
operator Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)>() const;
|
||||
operator Eigen::Transform<T, 3, Eigen::Affine>() const;
|
||||
#endif
|
||||
};
|
||||
|
||||
template<typename T> static
|
||||
Affine3<T> operator*(const Affine3<T>& affine1, const Affine3<T>& affine2);
|
||||
|
||||
//! V is a 3-element vector with member fields x, y and z
|
||||
template<typename T, typename V> static
|
||||
V operator*(const Affine3<T>& affine, const V& vector);
|
||||
|
||||
typedef Affine3<float> Affine3f;
|
||||
typedef Affine3<double> Affine3d;
|
||||
|
||||
static Vec3f operator*(const Affine3f& affine, const Vec3f& vector);
|
||||
static Vec3d operator*(const Affine3d& affine, const Vec3d& vector);
|
||||
|
||||
template<typename _Tp> class DataType< Affine3<_Tp> >
|
||||
{
|
||||
public:
|
||||
typedef Affine3<_Tp> value_type;
|
||||
typedef Affine3<typename DataType<_Tp>::work_type> work_type;
|
||||
typedef _Tp channel_type;
|
||||
|
||||
enum { generic_type = 0,
|
||||
channels = 16,
|
||||
fmt = traits::SafeFmt<channel_type>::fmt + ((channels - 1) << 8)
|
||||
#ifdef OPENCV_TRAITS_ENABLE_DEPRECATED
|
||||
,depth = DataType<channel_type>::depth
|
||||
,type = CV_MAKETYPE(depth, channels)
|
||||
#endif
|
||||
};
|
||||
|
||||
typedef Vec<channel_type, channels> vec_type;
|
||||
};
|
||||
|
||||
namespace traits {
|
||||
template<typename _Tp>
|
||||
struct Depth< Affine3<_Tp> > { enum { value = Depth<_Tp>::value }; };
|
||||
template<typename _Tp>
|
||||
struct Type< Affine3<_Tp> > { enum { value = CV_MAKETYPE(Depth<_Tp>::value, 16) }; };
|
||||
} // namespace
|
||||
|
||||
//! @} core
|
||||
|
||||
}
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
///////////////////////////////////////////////////////////////////////////////////
|
||||
// Implementation
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T>::Affine3()
|
||||
: matrix(Mat4::eye())
|
||||
{}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T>::Affine3(const Mat4& affine)
|
||||
: matrix(affine)
|
||||
{}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T>::Affine3(const Mat3& R, const Vec3& t)
|
||||
{
|
||||
rotation(R);
|
||||
translation(t);
|
||||
matrix.val[12] = matrix.val[13] = matrix.val[14] = 0;
|
||||
matrix.val[15] = 1;
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T>::Affine3(const Vec3& _rvec, const Vec3& t)
|
||||
{
|
||||
rotation(_rvec);
|
||||
translation(t);
|
||||
matrix.val[12] = matrix.val[13] = matrix.val[14] = 0;
|
||||
matrix.val[15] = 1;
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T>::Affine3(const cv::Mat& data, const Vec3& t)
|
||||
{
|
||||
CV_Assert(data.type() == cv::traits::Type<T>::value);
|
||||
CV_Assert(data.channels() == 1);
|
||||
|
||||
if (data.cols == 4 && data.rows == 4)
|
||||
{
|
||||
data.copyTo(matrix);
|
||||
return;
|
||||
}
|
||||
else if (data.cols == 4 && data.rows == 3)
|
||||
{
|
||||
rotation(data(Rect(0, 0, 3, 3)));
|
||||
translation(data(Rect(3, 0, 1, 3)));
|
||||
}
|
||||
else
|
||||
{
|
||||
rotation(data);
|
||||
translation(t);
|
||||
}
|
||||
|
||||
matrix.val[12] = matrix.val[13] = matrix.val[14] = 0;
|
||||
matrix.val[15] = 1;
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T>::Affine3(const float_type* vals) : matrix(vals)
|
||||
{}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T> cv::Affine3<T>::Identity()
|
||||
{
|
||||
return Affine3<T>(cv::Affine3<T>::Mat4::eye());
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
void cv::Affine3<T>::rotation(const Mat3& R)
|
||||
{
|
||||
linear(R);
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
void cv::Affine3<T>::rotation(const Vec3& _rvec)
|
||||
{
|
||||
double theta = norm(_rvec);
|
||||
|
||||
if (theta < DBL_EPSILON)
|
||||
rotation(Mat3::eye());
|
||||
else
|
||||
{
|
||||
double c = std::cos(theta);
|
||||
double s = std::sin(theta);
|
||||
double c1 = 1. - c;
|
||||
double itheta = (theta != 0) ? 1./theta : 0.;
|
||||
|
||||
Point3_<T> r = _rvec*itheta;
|
||||
|
||||
Mat3 rrt( r.x*r.x, r.x*r.y, r.x*r.z, r.x*r.y, r.y*r.y, r.y*r.z, r.x*r.z, r.y*r.z, r.z*r.z );
|
||||
Mat3 r_x( 0, -r.z, r.y, r.z, 0, -r.x, -r.y, r.x, 0 );
|
||||
|
||||
// R = cos(theta)*I + (1 - cos(theta))*r*rT + sin(theta)*[r_x]
|
||||
// where [r_x] is [0 -rz ry; rz 0 -rx; -ry rx 0]
|
||||
Mat3 R = c*Mat3::eye() + c1*rrt + s*r_x;
|
||||
|
||||
rotation(R);
|
||||
}
|
||||
}
|
||||
|
||||
//Combines rotation methods above. Supports 3x3, 1x3, 3x1 sizes of data matrix;
|
||||
template<typename T> inline
|
||||
void cv::Affine3<T>::rotation(const cv::Mat& data)
|
||||
{
|
||||
CV_Assert(data.type() == cv::traits::Type<T>::value);
|
||||
CV_Assert(data.channels() == 1);
|
||||
|
||||
if (data.cols == 3 && data.rows == 3)
|
||||
{
|
||||
Mat3 R;
|
||||
data.copyTo(R);
|
||||
rotation(R);
|
||||
}
|
||||
else if ((data.cols == 3 && data.rows == 1) || (data.cols == 1 && data.rows == 3))
|
||||
{
|
||||
Vec3 _rvec;
|
||||
data.reshape(1, 3).copyTo(_rvec);
|
||||
rotation(_rvec);
|
||||
}
|
||||
else
|
||||
CV_Error(Error::StsError, "Input matrix can only be 3x3, 1x3 or 3x1");
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
void cv::Affine3<T>::linear(const Mat3& L)
|
||||
{
|
||||
matrix.val[0] = L.val[0]; matrix.val[1] = L.val[1]; matrix.val[ 2] = L.val[2];
|
||||
matrix.val[4] = L.val[3]; matrix.val[5] = L.val[4]; matrix.val[ 6] = L.val[5];
|
||||
matrix.val[8] = L.val[6]; matrix.val[9] = L.val[7]; matrix.val[10] = L.val[8];
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
void cv::Affine3<T>::translation(const Vec3& t)
|
||||
{
|
||||
matrix.val[3] = t[0]; matrix.val[7] = t[1]; matrix.val[11] = t[2];
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
typename cv::Affine3<T>::Mat3 cv::Affine3<T>::rotation() const
|
||||
{
|
||||
return linear();
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
typename cv::Affine3<T>::Mat3 cv::Affine3<T>::linear() const
|
||||
{
|
||||
typename cv::Affine3<T>::Mat3 R;
|
||||
R.val[0] = matrix.val[0]; R.val[1] = matrix.val[1]; R.val[2] = matrix.val[ 2];
|
||||
R.val[3] = matrix.val[4]; R.val[4] = matrix.val[5]; R.val[5] = matrix.val[ 6];
|
||||
R.val[6] = matrix.val[8]; R.val[7] = matrix.val[9]; R.val[8] = matrix.val[10];
|
||||
return R;
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
typename cv::Affine3<T>::Vec3 cv::Affine3<T>::translation() const
|
||||
{
|
||||
return Vec3(matrix.val[3], matrix.val[7], matrix.val[11]);
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
typename cv::Affine3<T>::Vec3 cv::Affine3<T>::rvec() const
|
||||
{
|
||||
cv::Vec3d w;
|
||||
cv::Matx33d u, vt, R = rotation();
|
||||
cv::SVD::compute(R, w, u, vt, cv::SVD::FULL_UV + cv::SVD::MODIFY_A);
|
||||
R = u * vt;
|
||||
|
||||
double rx = R.val[7] - R.val[5];
|
||||
double ry = R.val[2] - R.val[6];
|
||||
double rz = R.val[3] - R.val[1];
|
||||
|
||||
double s = std::sqrt((rx*rx + ry*ry + rz*rz)*0.25);
|
||||
double c = (R.val[0] + R.val[4] + R.val[8] - 1) * 0.5;
|
||||
c = c > 1.0 ? 1.0 : c < -1.0 ? -1.0 : c;
|
||||
double theta = acos(c);
|
||||
|
||||
if( s < 1e-5 )
|
||||
{
|
||||
if( c > 0 )
|
||||
rx = ry = rz = 0;
|
||||
else
|
||||
{
|
||||
double t;
|
||||
t = (R.val[0] + 1) * 0.5;
|
||||
rx = std::sqrt(std::max(t, 0.0));
|
||||
t = (R.val[4] + 1) * 0.5;
|
||||
ry = std::sqrt(std::max(t, 0.0)) * (R.val[1] < 0 ? -1.0 : 1.0);
|
||||
t = (R.val[8] + 1) * 0.5;
|
||||
rz = std::sqrt(std::max(t, 0.0)) * (R.val[2] < 0 ? -1.0 : 1.0);
|
||||
|
||||
if( fabs(rx) < fabs(ry) && fabs(rx) < fabs(rz) && (R.val[5] > 0) != (ry*rz > 0) )
|
||||
rz = -rz;
|
||||
theta /= std::sqrt(rx*rx + ry*ry + rz*rz);
|
||||
rx *= theta;
|
||||
ry *= theta;
|
||||
rz *= theta;
|
||||
}
|
||||
}
|
||||
else
|
||||
{
|
||||
double vth = 1/(2*s);
|
||||
vth *= theta;
|
||||
rx *= vth; ry *= vth; rz *= vth;
|
||||
}
|
||||
|
||||
return cv::Vec3d(rx, ry, rz);
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T> cv::Affine3<T>::inv(int method) const
|
||||
{
|
||||
return matrix.inv(method);
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T> cv::Affine3<T>::rotate(const Mat3& R) const
|
||||
{
|
||||
Mat3 Lc = linear();
|
||||
Vec3 tc = translation();
|
||||
Mat4 result;
|
||||
result.val[12] = result.val[13] = result.val[14] = 0;
|
||||
result.val[15] = 1;
|
||||
|
||||
for(int j = 0; j < 3; ++j)
|
||||
{
|
||||
for(int i = 0; i < 3; ++i)
|
||||
{
|
||||
float_type value = 0;
|
||||
for(int k = 0; k < 3; ++k)
|
||||
value += R(j, k) * Lc(k, i);
|
||||
result(j, i) = value;
|
||||
}
|
||||
|
||||
result(j, 3) = R.row(j).dot(tc.t());
|
||||
}
|
||||
return result;
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T> cv::Affine3<T>::rotate(const Vec3& _rvec) const
|
||||
{
|
||||
return rotate(Affine3f(_rvec).rotation());
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T> cv::Affine3<T>::translate(const Vec3& t) const
|
||||
{
|
||||
Mat4 m = matrix;
|
||||
m.val[ 3] += t[0];
|
||||
m.val[ 7] += t[1];
|
||||
m.val[11] += t[2];
|
||||
return m;
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T> cv::Affine3<T>::concatenate(const Affine3<T>& affine) const
|
||||
{
|
||||
return (*this).rotate(affine.rotation()).translate(affine.translation());
|
||||
}
|
||||
|
||||
template<typename T> template <typename Y> inline
|
||||
cv::Affine3<T>::operator Affine3<Y>() const
|
||||
{
|
||||
return Affine3<Y>(matrix);
|
||||
}
|
||||
|
||||
template<typename T> template <typename Y> inline
|
||||
cv::Affine3<Y> cv::Affine3<T>::cast() const
|
||||
{
|
||||
return Affine3<Y>(matrix);
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T> cv::operator*(const cv::Affine3<T>& affine1, const cv::Affine3<T>& affine2)
|
||||
{
|
||||
return affine2.concatenate(affine1);
|
||||
}
|
||||
|
||||
template<typename T, typename V> inline
|
||||
V cv::operator*(const cv::Affine3<T>& affine, const V& v)
|
||||
{
|
||||
const typename Affine3<T>::Mat4& m = affine.matrix;
|
||||
|
||||
V r;
|
||||
r.x = m.val[0] * v.x + m.val[1] * v.y + m.val[ 2] * v.z + m.val[ 3];
|
||||
r.y = m.val[4] * v.x + m.val[5] * v.y + m.val[ 6] * v.z + m.val[ 7];
|
||||
r.z = m.val[8] * v.x + m.val[9] * v.y + m.val[10] * v.z + m.val[11];
|
||||
return r;
|
||||
}
|
||||
|
||||
static inline
|
||||
cv::Vec3f cv::operator*(const cv::Affine3f& affine, const cv::Vec3f& v)
|
||||
{
|
||||
const cv::Matx44f& m = affine.matrix;
|
||||
cv::Vec3f r;
|
||||
r.val[0] = m.val[0] * v[0] + m.val[1] * v[1] + m.val[ 2] * v[2] + m.val[ 3];
|
||||
r.val[1] = m.val[4] * v[0] + m.val[5] * v[1] + m.val[ 6] * v[2] + m.val[ 7];
|
||||
r.val[2] = m.val[8] * v[0] + m.val[9] * v[1] + m.val[10] * v[2] + m.val[11];
|
||||
return r;
|
||||
}
|
||||
|
||||
static inline
|
||||
cv::Vec3d cv::operator*(const cv::Affine3d& affine, const cv::Vec3d& v)
|
||||
{
|
||||
const cv::Matx44d& m = affine.matrix;
|
||||
cv::Vec3d r;
|
||||
r.val[0] = m.val[0] * v[0] + m.val[1] * v[1] + m.val[ 2] * v[2] + m.val[ 3];
|
||||
r.val[1] = m.val[4] * v[0] + m.val[5] * v[1] + m.val[ 6] * v[2] + m.val[ 7];
|
||||
r.val[2] = m.val[8] * v[0] + m.val[9] * v[1] + m.val[10] * v[2] + m.val[11];
|
||||
return r;
|
||||
}
|
||||
|
||||
|
||||
|
||||
#if defined EIGEN_WORLD_VERSION && defined EIGEN_GEOMETRY_MODULE_H
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T>::Affine3(const Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)>& affine)
|
||||
{
|
||||
cv::Mat(4, 4, cv::traits::Type<T>::value, affine.matrix().data()).copyTo(matrix);
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T>::Affine3(const Eigen::Transform<T, 3, Eigen::Affine>& affine)
|
||||
{
|
||||
Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)> a = affine;
|
||||
cv::Mat(4, 4, cv::traits::Type<T>::value, a.matrix().data()).copyTo(matrix);
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T>::operator Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)>() const
|
||||
{
|
||||
Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)> r;
|
||||
cv::Mat hdr(4, 4, cv::traits::Type<T>::value, r.matrix().data());
|
||||
cv::Mat(matrix, false).copyTo(hdr);
|
||||
return r;
|
||||
}
|
||||
|
||||
template<typename T> inline
|
||||
cv::Affine3<T>::operator Eigen::Transform<T, 3, Eigen::Affine>() const
|
||||
{
|
||||
return this->operator Eigen::Transform<T, 3, Eigen::Affine, (Eigen::RowMajor)>();
|
||||
}
|
||||
|
||||
#endif /* defined EIGEN_WORLD_VERSION && defined EIGEN_GEOMETRY_MODULE_H */
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif /* __cplusplus */
|
||||
|
||||
#endif /* OPENCV_CORE_AFFINE3_HPP */
|
||||
|
|
@ -0,0 +1,707 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Copyright (C) 2014, Itseez Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CORE_BASE_HPP
|
||||
#define OPENCV_CORE_BASE_HPP
|
||||
|
||||
#ifndef __cplusplus
|
||||
# error base.hpp header must be compiled as C++
|
||||
#endif
|
||||
|
||||
#include "opencv2/opencv_modules.hpp"
|
||||
|
||||
#include <climits>
|
||||
#include <algorithm>
|
||||
|
||||
#include "opencv2/core/cvdef.h"
|
||||
#include "opencv2/core/cvstd.hpp"
|
||||
|
||||
namespace cv
|
||||
{
|
||||
|
||||
//! @addtogroup core_utils
|
||||
//! @{
|
||||
|
||||
namespace Error {
|
||||
//! error codes
|
||||
enum Code {
|
||||
StsOk= 0, //!< everything is ok
|
||||
StsBackTrace= -1, //!< pseudo error for back trace
|
||||
StsError= -2, //!< unknown /unspecified error
|
||||
StsInternal= -3, //!< internal error (bad state)
|
||||
StsNoMem= -4, //!< insufficient memory
|
||||
StsBadArg= -5, //!< function arg/param is bad
|
||||
StsBadFunc= -6, //!< unsupported function
|
||||
StsNoConv= -7, //!< iteration didn't converge
|
||||
StsAutoTrace= -8, //!< tracing
|
||||
HeaderIsNull= -9, //!< image header is NULL
|
||||
BadImageSize= -10, //!< image size is invalid
|
||||
BadOffset= -11, //!< offset is invalid
|
||||
BadDataPtr= -12, //!<
|
||||
BadStep= -13, //!< image step is wrong, this may happen for a non-continuous matrix.
|
||||
BadModelOrChSeq= -14, //!<
|
||||
BadNumChannels= -15, //!< bad number of channels, for example, some functions accept only single channel matrices.
|
||||
BadNumChannel1U= -16, //!<
|
||||
BadDepth= -17, //!< input image depth is not supported by the function
|
||||
BadAlphaChannel= -18, //!<
|
||||
BadOrder= -19, //!< number of dimensions is out of range
|
||||
BadOrigin= -20, //!< incorrect input origin
|
||||
BadAlign= -21, //!< incorrect input align
|
||||
BadCallBack= -22, //!<
|
||||
BadTileSize= -23, //!<
|
||||
BadCOI= -24, //!< input COI is not supported
|
||||
BadROISize= -25, //!< incorrect input roi
|
||||
MaskIsTiled= -26, //!<
|
||||
StsNullPtr= -27, //!< null pointer
|
||||
StsVecLengthErr= -28, //!< incorrect vector length
|
||||
StsFilterStructContentErr= -29, //!< incorrect filter structure content
|
||||
StsKernelStructContentErr= -30, //!< incorrect transform kernel content
|
||||
StsFilterOffsetErr= -31, //!< incorrect filter offset value
|
||||
StsBadSize= -201, //!< the input/output structure size is incorrect
|
||||
StsDivByZero= -202, //!< division by zero
|
||||
StsInplaceNotSupported= -203, //!< in-place operation is not supported
|
||||
StsObjectNotFound= -204, //!< request can't be completed
|
||||
StsUnmatchedFormats= -205, //!< formats of input/output arrays differ
|
||||
StsBadFlag= -206, //!< flag is wrong or not supported
|
||||
StsBadPoint= -207, //!< bad CvPoint
|
||||
StsBadMask= -208, //!< bad format of mask (neither 8uC1 nor 8sC1)
|
||||
StsUnmatchedSizes= -209, //!< sizes of input/output structures do not match
|
||||
StsUnsupportedFormat= -210, //!< the data format/type is not supported by the function
|
||||
StsOutOfRange= -211, //!< some of parameters are out of range
|
||||
StsParseError= -212, //!< invalid syntax/structure of the parsed file
|
||||
StsNotImplemented= -213, //!< the requested function/feature is not implemented
|
||||
StsBadMemBlock= -214, //!< an allocated block has been corrupted
|
||||
StsAssert= -215, //!< assertion failed
|
||||
GpuNotSupported= -216, //!< no CUDA support
|
||||
GpuApiCallError= -217, //!< GPU API call error
|
||||
OpenGlNotSupported= -218, //!< no OpenGL support
|
||||
OpenGlApiCallError= -219, //!< OpenGL API call error
|
||||
OpenCLApiCallError= -220, //!< OpenCL API call error
|
||||
OpenCLDoubleNotSupported= -221,
|
||||
OpenCLInitError= -222, //!< OpenCL initialization error
|
||||
OpenCLNoAMDBlasFft= -223
|
||||
};
|
||||
} //Error
|
||||
|
||||
//! @} core_utils
|
||||
|
||||
//! @addtogroup core_array
|
||||
//! @{
|
||||
|
||||
//! matrix decomposition types
|
||||
enum DecompTypes {
|
||||
/** Gaussian elimination with the optimal pivot element chosen. */
|
||||
DECOMP_LU = 0,
|
||||
/** singular value decomposition (SVD) method; the system can be over-defined and/or the matrix
|
||||
src1 can be singular */
|
||||
DECOMP_SVD = 1,
|
||||
/** eigenvalue decomposition; the matrix src1 must be symmetrical */
|
||||
DECOMP_EIG = 2,
|
||||
/** Cholesky \f$LL^T\f$ factorization; the matrix src1 must be symmetrical and positively
|
||||
defined */
|
||||
DECOMP_CHOLESKY = 3,
|
||||
/** QR factorization; the system can be over-defined and/or the matrix src1 can be singular */
|
||||
DECOMP_QR = 4,
|
||||
/** while all the previous flags are mutually exclusive, this flag can be used together with
|
||||
any of the previous; it means that the normal equations
|
||||
\f$\texttt{src1}^T\cdot\texttt{src1}\cdot\texttt{dst}=\texttt{src1}^T\texttt{src2}\f$ are
|
||||
solved instead of the original system
|
||||
\f$\texttt{src1}\cdot\texttt{dst}=\texttt{src2}\f$ */
|
||||
DECOMP_NORMAL = 16
|
||||
};
|
||||
|
||||
/** norm types
|
||||
|
||||
src1 and src2 denote input arrays.
|
||||
*/
|
||||
|
||||
enum NormTypes {
|
||||
/**
|
||||
\f[
|
||||
norm = \forkthree
|
||||
{\|\texttt{src1}\|_{L_{\infty}} = \max _I | \texttt{src1} (I)|}{if \(\texttt{normType} = \texttt{NORM_INF}\) }
|
||||
{\|\texttt{src1}-\texttt{src2}\|_{L_{\infty}} = \max _I | \texttt{src1} (I) - \texttt{src2} (I)|}{if \(\texttt{normType} = \texttt{NORM_INF}\) }
|
||||
{\frac{\|\texttt{src1}-\texttt{src2}\|_{L_{\infty}} }{\|\texttt{src2}\|_{L_{\infty}} }}{if \(\texttt{normType} = \texttt{NORM_RELATIVE | NORM_INF}\) }
|
||||
\f]
|
||||
*/
|
||||
NORM_INF = 1,
|
||||
/**
|
||||
\f[
|
||||
norm = \forkthree
|
||||
{\| \texttt{src1} \| _{L_1} = \sum _I | \texttt{src1} (I)|}{if \(\texttt{normType} = \texttt{NORM_L1}\)}
|
||||
{ \| \texttt{src1} - \texttt{src2} \| _{L_1} = \sum _I | \texttt{src1} (I) - \texttt{src2} (I)|}{if \(\texttt{normType} = \texttt{NORM_L1}\) }
|
||||
{ \frac{\|\texttt{src1}-\texttt{src2}\|_{L_1} }{\|\texttt{src2}\|_{L_1}} }{if \(\texttt{normType} = \texttt{NORM_RELATIVE | NORM_L1}\) }
|
||||
\f]*/
|
||||
NORM_L1 = 2,
|
||||
/**
|
||||
\f[
|
||||
norm = \forkthree
|
||||
{ \| \texttt{src1} \| _{L_2} = \sqrt{\sum_I \texttt{src1}(I)^2} }{if \(\texttt{normType} = \texttt{NORM_L2}\) }
|
||||
{ \| \texttt{src1} - \texttt{src2} \| _{L_2} = \sqrt{\sum_I (\texttt{src1}(I) - \texttt{src2}(I))^2} }{if \(\texttt{normType} = \texttt{NORM_L2}\) }
|
||||
{ \frac{\|\texttt{src1}-\texttt{src2}\|_{L_2} }{\|\texttt{src2}\|_{L_2}} }{if \(\texttt{normType} = \texttt{NORM_RELATIVE | NORM_L2}\) }
|
||||
\f]
|
||||
*/
|
||||
NORM_L2 = 4,
|
||||
/**
|
||||
\f[
|
||||
norm = \forkthree
|
||||
{ \| \texttt{src1} \| _{L_2} ^{2} = \sum_I \texttt{src1}(I)^2} {if \(\texttt{normType} = \texttt{NORM_L2SQR}\)}
|
||||
{ \| \texttt{src1} - \texttt{src2} \| _{L_2} ^{2} = \sum_I (\texttt{src1}(I) - \texttt{src2}(I))^2 }{if \(\texttt{normType} = \texttt{NORM_L2SQR}\) }
|
||||
{ \left(\frac{\|\texttt{src1}-\texttt{src2}\|_{L_2} }{\|\texttt{src2}\|_{L_2}}\right)^2 }{if \(\texttt{normType} = \texttt{NORM_RELATIVE | NORM_L2}\) }
|
||||
\f]
|
||||
*/
|
||||
NORM_L2SQR = 5,
|
||||
/**
|
||||
In the case of one input array, calculates the Hamming distance of the array from zero,
|
||||
In the case of two input arrays, calculates the Hamming distance between the arrays.
|
||||
*/
|
||||
NORM_HAMMING = 6,
|
||||
/**
|
||||
Similar to NORM_HAMMING, but in the calculation, each two bits of the input sequence will
|
||||
be added and treated as a single bit to be used in the same calculation as NORM_HAMMING.
|
||||
*/
|
||||
NORM_HAMMING2 = 7,
|
||||
NORM_TYPE_MASK = 7, //!< bit-mask which can be used to separate norm type from norm flags
|
||||
NORM_RELATIVE = 8, //!< flag
|
||||
NORM_MINMAX = 32 //!< flag
|
||||
};
|
||||
|
||||
//! comparison types
|
||||
enum CmpTypes { CMP_EQ = 0, //!< src1 is equal to src2.
|
||||
CMP_GT = 1, //!< src1 is greater than src2.
|
||||
CMP_GE = 2, //!< src1 is greater than or equal to src2.
|
||||
CMP_LT = 3, //!< src1 is less than src2.
|
||||
CMP_LE = 4, //!< src1 is less than or equal to src2.
|
||||
CMP_NE = 5 //!< src1 is unequal to src2.
|
||||
};
|
||||
|
||||
//! generalized matrix multiplication flags
|
||||
enum GemmFlags { GEMM_1_T = 1, //!< transposes src1
|
||||
GEMM_2_T = 2, //!< transposes src2
|
||||
GEMM_3_T = 4 //!< transposes src3
|
||||
};
|
||||
|
||||
enum DftFlags {
|
||||
/** performs an inverse 1D or 2D transform instead of the default forward
|
||||
transform. */
|
||||
DFT_INVERSE = 1,
|
||||
/** scales the result: divide it by the number of array elements. Normally, it is
|
||||
combined with DFT_INVERSE. */
|
||||
DFT_SCALE = 2,
|
||||
/** performs a forward or inverse transform of every individual row of the input
|
||||
matrix; this flag enables you to transform multiple vectors simultaneously and can be used to
|
||||
decrease the overhead (which is sometimes several times larger than the processing itself) to
|
||||
perform 3D and higher-dimensional transformations and so forth.*/
|
||||
DFT_ROWS = 4,
|
||||
/** performs a forward transformation of 1D or 2D real array; the result,
|
||||
though being a complex array, has complex-conjugate symmetry (*CCS*, see the function
|
||||
description below for details), and such an array can be packed into a real array of the same
|
||||
size as input, which is the fastest option and which is what the function does by default;
|
||||
however, you may wish to get a full complex array (for simpler spectrum analysis, and so on) -
|
||||
pass the flag to enable the function to produce a full-size complex output array. */
|
||||
DFT_COMPLEX_OUTPUT = 16,
|
||||
/** performs an inverse transformation of a 1D or 2D complex array; the
|
||||
result is normally a complex array of the same size, however, if the input array has
|
||||
conjugate-complex symmetry (for example, it is a result of forward transformation with
|
||||
DFT_COMPLEX_OUTPUT flag), the output is a real array; while the function itself does not
|
||||
check whether the input is symmetrical or not, you can pass the flag and then the function
|
||||
will assume the symmetry and produce the real output array (note that when the input is packed
|
||||
into a real array and inverse transformation is executed, the function treats the input as a
|
||||
packed complex-conjugate symmetrical array, and the output will also be a real array). */
|
||||
DFT_REAL_OUTPUT = 32,
|
||||
/** specifies that input is complex input. If this flag is set, the input must have 2 channels.
|
||||
On the other hand, for backwards compatibility reason, if input has 2 channels, input is
|
||||
already considered complex. */
|
||||
DFT_COMPLEX_INPUT = 64,
|
||||
/** performs an inverse 1D or 2D transform instead of the default forward transform. */
|
||||
DCT_INVERSE = DFT_INVERSE,
|
||||
/** performs a forward or inverse transform of every individual row of the input
|
||||
matrix. This flag enables you to transform multiple vectors simultaneously and can be used to
|
||||
decrease the overhead (which is sometimes several times larger than the processing itself) to
|
||||
perform 3D and higher-dimensional transforms and so forth.*/
|
||||
DCT_ROWS = DFT_ROWS
|
||||
};
|
||||
|
||||
//! Various border types, image boundaries are denoted with `|`
|
||||
//! @see borderInterpolate, copyMakeBorder
|
||||
enum BorderTypes {
|
||||
BORDER_CONSTANT = 0, //!< `iiiiii|abcdefgh|iiiiiii` with some specified `i`
|
||||
BORDER_REPLICATE = 1, //!< `aaaaaa|abcdefgh|hhhhhhh`
|
||||
BORDER_REFLECT = 2, //!< `fedcba|abcdefgh|hgfedcb`
|
||||
BORDER_WRAP = 3, //!< `cdefgh|abcdefgh|abcdefg`
|
||||
BORDER_REFLECT_101 = 4, //!< `gfedcb|abcdefgh|gfedcba`
|
||||
BORDER_TRANSPARENT = 5, //!< `uvwxyz|abcdefgh|ijklmno`
|
||||
|
||||
BORDER_REFLECT101 = BORDER_REFLECT_101, //!< same as BORDER_REFLECT_101
|
||||
BORDER_DEFAULT = BORDER_REFLECT_101, //!< same as BORDER_REFLECT_101
|
||||
BORDER_ISOLATED = 16 //!< do not look outside of ROI
|
||||
};
|
||||
|
||||
//! @} core_array
|
||||
|
||||
//! @addtogroup core_utils
|
||||
//! @{
|
||||
|
||||
/*! @brief Signals an error and raises the exception.
|
||||
|
||||
By default the function prints information about the error to stderr,
|
||||
then it either stops if setBreakOnError() had been called before or raises the exception.
|
||||
It is possible to alternate error processing by using redirectError().
|
||||
@param _code - error code (Error::Code)
|
||||
@param _err - error description
|
||||
@param _func - function name. Available only when the compiler supports getting it
|
||||
@param _file - source file name where the error has occurred
|
||||
@param _line - line number in the source file where the error has occurred
|
||||
@see CV_Error, CV_Error_, CV_Assert, CV_DbgAssert
|
||||
*/
|
||||
CV_EXPORTS void error(int _code, const String& _err, const char* _func, const char* _file, int _line);
|
||||
|
||||
#ifdef __GNUC__
|
||||
# if defined __clang__ || defined __APPLE__
|
||||
# pragma GCC diagnostic push
|
||||
# pragma GCC diagnostic ignored "-Winvalid-noreturn"
|
||||
# endif
|
||||
#endif
|
||||
|
||||
/** same as cv::error, but does not return */
|
||||
CV_INLINE CV_NORETURN void errorNoReturn(int _code, const String& _err, const char* _func, const char* _file, int _line)
|
||||
{
|
||||
error(_code, _err, _func, _file, _line);
|
||||
#ifdef __GNUC__
|
||||
# if !defined __clang__ && !defined __APPLE__
|
||||
// this suppresses this warning: "noreturn" function does return [enabled by default]
|
||||
__builtin_trap();
|
||||
// or use infinite loop: for (;;) {}
|
||||
# endif
|
||||
#endif
|
||||
}
|
||||
#ifdef __GNUC__
|
||||
# if defined __clang__ || defined __APPLE__
|
||||
# pragma GCC diagnostic pop
|
||||
# endif
|
||||
#endif
|
||||
|
||||
#ifdef CV_STATIC_ANALYSIS
|
||||
|
||||
// In practice, some macro are not processed correctly (noreturn is not detected).
|
||||
// We need to use simplified definition for them.
|
||||
#define CV_Error(...) do { abort(); } while (0)
|
||||
#define CV_Error_( code, args ) do { cv::format args; abort(); } while (0)
|
||||
#define CV_Assert( expr ) do { if (!(expr)) abort(); } while (0)
|
||||
#define CV_ErrorNoReturn CV_Error
|
||||
#define CV_ErrorNoReturn_ CV_Error_
|
||||
|
||||
#else // CV_STATIC_ANALYSIS
|
||||
|
||||
/** @brief Call the error handler.
|
||||
|
||||
Currently, the error handler prints the error code and the error message to the standard
|
||||
error stream `stderr`. In the Debug configuration, it then provokes memory access violation, so that
|
||||
the execution stack and all the parameters can be analyzed by the debugger. In the Release
|
||||
configuration, the exception is thrown.
|
||||
|
||||
@param code one of Error::Code
|
||||
@param msg error message
|
||||
*/
|
||||
#define CV_Error( code, msg ) cv::error( code, msg, CV_Func, __FILE__, __LINE__ )
|
||||
|
||||
/** @brief Call the error handler.
|
||||
|
||||
This macro can be used to construct an error message on-fly to include some dynamic information,
|
||||
for example:
|
||||
@code
|
||||
// note the extra parentheses around the formatted text message
|
||||
CV_Error_(Error::StsOutOfRange,
|
||||
("the value at (%d, %d)=%g is out of range", badPt.x, badPt.y, badValue));
|
||||
@endcode
|
||||
@param code one of Error::Code
|
||||
@param args printf-like formatted error message in parentheses
|
||||
*/
|
||||
#define CV_Error_( code, args ) cv::error( code, cv::format args, CV_Func, __FILE__, __LINE__ )
|
||||
|
||||
/** @brief Checks a condition at runtime and throws exception if it fails
|
||||
|
||||
The macros CV_Assert (and CV_DbgAssert(expr)) evaluate the specified expression. If it is 0, the macros
|
||||
raise an error (see cv::error). The macro CV_Assert checks the condition in both Debug and Release
|
||||
configurations while CV_DbgAssert is only retained in the Debug configuration.
|
||||
*/
|
||||
#define CV_Assert( expr ) do { if(!!(expr)) ; else cv::error( cv::Error::StsAssert, #expr, CV_Func, __FILE__, __LINE__ ); } while(0)
|
||||
|
||||
//! @cond IGNORED
|
||||
#define CV__ErrorNoReturn( code, msg ) cv::errorNoReturn( code, msg, CV_Func, __FILE__, __LINE__ )
|
||||
#define CV__ErrorNoReturn_( code, args ) cv::errorNoReturn( code, cv::format args, CV_Func, __FILE__, __LINE__ )
|
||||
#ifdef __OPENCV_BUILD
|
||||
#undef CV_Error
|
||||
#define CV_Error CV__ErrorNoReturn
|
||||
#undef CV_Error_
|
||||
#define CV_Error_ CV__ErrorNoReturn_
|
||||
#undef CV_Assert
|
||||
#define CV_Assert( expr ) do { if(!!(expr)) ; else cv::errorNoReturn( cv::Error::StsAssert, #expr, CV_Func, __FILE__, __LINE__ ); } while(0)
|
||||
#else
|
||||
// backward compatibility
|
||||
#define CV_ErrorNoReturn CV__ErrorNoReturn
|
||||
#define CV_ErrorNoReturn_ CV__ErrorNoReturn_
|
||||
#endif
|
||||
//! @endcond
|
||||
|
||||
#endif // CV_STATIC_ANALYSIS
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
#if defined OPENCV_FORCE_MULTIARG_ASSERT_CHECK && defined CV_STATIC_ANALYSIS
|
||||
#warning "OPENCV_FORCE_MULTIARG_ASSERT_CHECK can't be used with CV_STATIC_ANALYSIS"
|
||||
#undef OPENCV_FORCE_MULTIARG_ASSERT_CHECK
|
||||
#endif
|
||||
|
||||
#ifdef OPENCV_FORCE_MULTIARG_ASSERT_CHECK
|
||||
#define CV_Assert_1( expr ) do { if(!!(expr)) ; else cv::error( cv::Error::StsAssert, #expr, CV_Func, __FILE__, __LINE__ ); } while(0)
|
||||
#else
|
||||
#define CV_Assert_1 CV_Assert
|
||||
#endif
|
||||
#define CV_Assert_2( expr1, expr2 ) CV_Assert_1(expr1); CV_Assert_1(expr2)
|
||||
#define CV_Assert_3( expr1, expr2, expr3 ) CV_Assert_2(expr1, expr2); CV_Assert_1(expr3)
|
||||
#define CV_Assert_4( expr1, expr2, expr3, expr4 ) CV_Assert_3(expr1, expr2, expr3); CV_Assert_1(expr4)
|
||||
#define CV_Assert_5( expr1, expr2, expr3, expr4, expr5 ) CV_Assert_4(expr1, expr2, expr3, expr4); CV_Assert_1(expr5)
|
||||
#define CV_Assert_6( expr1, expr2, expr3, expr4, expr5, expr6 ) CV_Assert_5(expr1, expr2, expr3, expr4, expr5); CV_Assert_1(expr6)
|
||||
#define CV_Assert_7( expr1, expr2, expr3, expr4, expr5, expr6, expr7 ) CV_Assert_6(expr1, expr2, expr3, expr4, expr5, expr6 ); CV_Assert_1(expr7)
|
||||
#define CV_Assert_8( expr1, expr2, expr3, expr4, expr5, expr6, expr7, expr8 ) CV_Assert_7(expr1, expr2, expr3, expr4, expr5, expr6, expr7 ); CV_Assert_1(expr8)
|
||||
#define CV_Assert_9( expr1, expr2, expr3, expr4, expr5, expr6, expr7, expr8, expr9 ) CV_Assert_8(expr1, expr2, expr3, expr4, expr5, expr6, expr7, expr8 ); CV_Assert_1(expr9)
|
||||
#define CV_Assert_10( expr1, expr2, expr3, expr4, expr5, expr6, expr7, expr8, expr9, expr10 ) CV_Assert_9(expr1, expr2, expr3, expr4, expr5, expr6, expr7, expr8, expr9 ); CV_Assert_1(expr10)
|
||||
|
||||
#define CV_Assert_N(...) do { __CV_CAT(CV_Assert_, __CV_VA_NUM_ARGS(__VA_ARGS__)) (__VA_ARGS__); } while(0)
|
||||
|
||||
#ifdef OPENCV_FORCE_MULTIARG_ASSERT_CHECK
|
||||
#undef CV_Assert
|
||||
#define CV_Assert CV_Assert_N
|
||||
#endif
|
||||
//! @endcond
|
||||
|
||||
#if defined _DEBUG || defined CV_STATIC_ANALYSIS
|
||||
# define CV_DbgAssert(expr) CV_Assert(expr)
|
||||
#else
|
||||
/** replaced with CV_Assert(expr) in Debug configuration */
|
||||
# define CV_DbgAssert(expr)
|
||||
#endif
|
||||
|
||||
/*
|
||||
* Hamming distance functor - counts the bit differences between two strings - useful for the Brief descriptor
|
||||
* bit count of A exclusive XOR'ed with B
|
||||
*/
|
||||
struct CV_EXPORTS Hamming
|
||||
{
|
||||
enum { normType = NORM_HAMMING };
|
||||
typedef unsigned char ValueType;
|
||||
typedef int ResultType;
|
||||
|
||||
/** this will count the bits in a ^ b
|
||||
*/
|
||||
ResultType operator()( const unsigned char* a, const unsigned char* b, int size ) const;
|
||||
};
|
||||
|
||||
typedef Hamming HammingLUT;
|
||||
|
||||
/////////////////////////////////// inline norms ////////////////////////////////////
|
||||
|
||||
template<typename _Tp> inline _Tp cv_abs(_Tp x) { return std::abs(x); }
|
||||
inline int cv_abs(uchar x) { return x; }
|
||||
inline int cv_abs(schar x) { return std::abs(x); }
|
||||
inline int cv_abs(ushort x) { return x; }
|
||||
inline int cv_abs(short x) { return std::abs(x); }
|
||||
|
||||
template<typename _Tp, typename _AccTp> static inline
|
||||
_AccTp normL2Sqr(const _Tp* a, int n)
|
||||
{
|
||||
_AccTp s = 0;
|
||||
int i=0;
|
||||
#if CV_ENABLE_UNROLLED
|
||||
for( ; i <= n - 4; i += 4 )
|
||||
{
|
||||
_AccTp v0 = a[i], v1 = a[i+1], v2 = a[i+2], v3 = a[i+3];
|
||||
s += v0*v0 + v1*v1 + v2*v2 + v3*v3;
|
||||
}
|
||||
#endif
|
||||
for( ; i < n; i++ )
|
||||
{
|
||||
_AccTp v = a[i];
|
||||
s += v*v;
|
||||
}
|
||||
return s;
|
||||
}
|
||||
|
||||
template<typename _Tp, typename _AccTp> static inline
|
||||
_AccTp normL1(const _Tp* a, int n)
|
||||
{
|
||||
_AccTp s = 0;
|
||||
int i = 0;
|
||||
#if CV_ENABLE_UNROLLED
|
||||
for(; i <= n - 4; i += 4 )
|
||||
{
|
||||
s += (_AccTp)cv_abs(a[i]) + (_AccTp)cv_abs(a[i+1]) +
|
||||
(_AccTp)cv_abs(a[i+2]) + (_AccTp)cv_abs(a[i+3]);
|
||||
}
|
||||
#endif
|
||||
for( ; i < n; i++ )
|
||||
s += cv_abs(a[i]);
|
||||
return s;
|
||||
}
|
||||
|
||||
template<typename _Tp, typename _AccTp> static inline
|
||||
_AccTp normInf(const _Tp* a, int n)
|
||||
{
|
||||
_AccTp s = 0;
|
||||
for( int i = 0; i < n; i++ )
|
||||
s = std::max(s, (_AccTp)cv_abs(a[i]));
|
||||
return s;
|
||||
}
|
||||
|
||||
template<typename _Tp, typename _AccTp> static inline
|
||||
_AccTp normL2Sqr(const _Tp* a, const _Tp* b, int n)
|
||||
{
|
||||
_AccTp s = 0;
|
||||
int i= 0;
|
||||
#if CV_ENABLE_UNROLLED
|
||||
for(; i <= n - 4; i += 4 )
|
||||
{
|
||||
_AccTp v0 = _AccTp(a[i] - b[i]), v1 = _AccTp(a[i+1] - b[i+1]), v2 = _AccTp(a[i+2] - b[i+2]), v3 = _AccTp(a[i+3] - b[i+3]);
|
||||
s += v0*v0 + v1*v1 + v2*v2 + v3*v3;
|
||||
}
|
||||
#endif
|
||||
for( ; i < n; i++ )
|
||||
{
|
||||
_AccTp v = _AccTp(a[i] - b[i]);
|
||||
s += v*v;
|
||||
}
|
||||
return s;
|
||||
}
|
||||
|
||||
static inline float normL2Sqr(const float* a, const float* b, int n)
|
||||
{
|
||||
float s = 0.f;
|
||||
for( int i = 0; i < n; i++ )
|
||||
{
|
||||
float v = a[i] - b[i];
|
||||
s += v*v;
|
||||
}
|
||||
return s;
|
||||
}
|
||||
|
||||
template<typename _Tp, typename _AccTp> static inline
|
||||
_AccTp normL1(const _Tp* a, const _Tp* b, int n)
|
||||
{
|
||||
_AccTp s = 0;
|
||||
int i= 0;
|
||||
#if CV_ENABLE_UNROLLED
|
||||
for(; i <= n - 4; i += 4 )
|
||||
{
|
||||
_AccTp v0 = _AccTp(a[i] - b[i]), v1 = _AccTp(a[i+1] - b[i+1]), v2 = _AccTp(a[i+2] - b[i+2]), v3 = _AccTp(a[i+3] - b[i+3]);
|
||||
s += std::abs(v0) + std::abs(v1) + std::abs(v2) + std::abs(v3);
|
||||
}
|
||||
#endif
|
||||
for( ; i < n; i++ )
|
||||
{
|
||||
_AccTp v = _AccTp(a[i] - b[i]);
|
||||
s += std::abs(v);
|
||||
}
|
||||
return s;
|
||||
}
|
||||
|
||||
inline float normL1(const float* a, const float* b, int n)
|
||||
{
|
||||
float s = 0.f;
|
||||
for( int i = 0; i < n; i++ )
|
||||
{
|
||||
s += std::abs(a[i] - b[i]);
|
||||
}
|
||||
return s;
|
||||
}
|
||||
|
||||
inline int normL1(const uchar* a, const uchar* b, int n)
|
||||
{
|
||||
int s = 0;
|
||||
for( int i = 0; i < n; i++ )
|
||||
{
|
||||
s += std::abs(a[i] - b[i]);
|
||||
}
|
||||
return s;
|
||||
}
|
||||
|
||||
template<typename _Tp, typename _AccTp> static inline
|
||||
_AccTp normInf(const _Tp* a, const _Tp* b, int n)
|
||||
{
|
||||
_AccTp s = 0;
|
||||
for( int i = 0; i < n; i++ )
|
||||
{
|
||||
_AccTp v0 = a[i] - b[i];
|
||||
s = std::max(s, std::abs(v0));
|
||||
}
|
||||
return s;
|
||||
}
|
||||
|
||||
/** @brief Computes the cube root of an argument.
|
||||
|
||||
The function cubeRoot computes \f$\sqrt[3]{\texttt{val}}\f$. Negative arguments are handled correctly.
|
||||
NaN and Inf are not handled. The accuracy approaches the maximum possible accuracy for
|
||||
single-precision data.
|
||||
@param val A function argument.
|
||||
*/
|
||||
CV_EXPORTS_W float cubeRoot(float val);
|
||||
|
||||
/** @brief Calculates the angle of a 2D vector in degrees.
|
||||
|
||||
The function fastAtan2 calculates the full-range angle of an input 2D vector. The angle is measured
|
||||
in degrees and varies from 0 to 360 degrees. The accuracy is about 0.3 degrees.
|
||||
@param x x-coordinate of the vector.
|
||||
@param y y-coordinate of the vector.
|
||||
*/
|
||||
CV_EXPORTS_W float fastAtan2(float y, float x);
|
||||
|
||||
/** proxy for hal::LU */
|
||||
CV_EXPORTS int LU(float* A, size_t astep, int m, float* b, size_t bstep, int n);
|
||||
/** proxy for hal::LU */
|
||||
CV_EXPORTS int LU(double* A, size_t astep, int m, double* b, size_t bstep, int n);
|
||||
/** proxy for hal::Cholesky */
|
||||
CV_EXPORTS bool Cholesky(float* A, size_t astep, int m, float* b, size_t bstep, int n);
|
||||
/** proxy for hal::Cholesky */
|
||||
CV_EXPORTS bool Cholesky(double* A, size_t astep, int m, double* b, size_t bstep, int n);
|
||||
|
||||
////////////////// forward declarations for important OpenCV types //////////////////
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
template<typename _Tp, int cn> class Vec;
|
||||
template<typename _Tp, int m, int n> class Matx;
|
||||
|
||||
template<typename _Tp> class Complex;
|
||||
template<typename _Tp> class Point_;
|
||||
template<typename _Tp> class Point3_;
|
||||
template<typename _Tp> class Size_;
|
||||
template<typename _Tp> class Rect_;
|
||||
template<typename _Tp> class Scalar_;
|
||||
|
||||
class CV_EXPORTS RotatedRect;
|
||||
class CV_EXPORTS Range;
|
||||
class CV_EXPORTS TermCriteria;
|
||||
class CV_EXPORTS KeyPoint;
|
||||
class CV_EXPORTS DMatch;
|
||||
class CV_EXPORTS RNG;
|
||||
|
||||
class CV_EXPORTS Mat;
|
||||
class CV_EXPORTS MatExpr;
|
||||
|
||||
class CV_EXPORTS UMat;
|
||||
|
||||
class CV_EXPORTS SparseMat;
|
||||
typedef Mat MatND;
|
||||
|
||||
template<typename _Tp> class Mat_;
|
||||
template<typename _Tp> class SparseMat_;
|
||||
|
||||
class CV_EXPORTS MatConstIterator;
|
||||
class CV_EXPORTS SparseMatIterator;
|
||||
class CV_EXPORTS SparseMatConstIterator;
|
||||
template<typename _Tp> class MatIterator_;
|
||||
template<typename _Tp> class MatConstIterator_;
|
||||
template<typename _Tp> class SparseMatIterator_;
|
||||
template<typename _Tp> class SparseMatConstIterator_;
|
||||
|
||||
namespace ogl
|
||||
{
|
||||
class CV_EXPORTS Buffer;
|
||||
class CV_EXPORTS Texture2D;
|
||||
class CV_EXPORTS Arrays;
|
||||
}
|
||||
|
||||
namespace cuda
|
||||
{
|
||||
class CV_EXPORTS GpuMat;
|
||||
class CV_EXPORTS HostMem;
|
||||
class CV_EXPORTS Stream;
|
||||
class CV_EXPORTS Event;
|
||||
}
|
||||
|
||||
namespace cudev
|
||||
{
|
||||
template <typename _Tp> class GpuMat_;
|
||||
}
|
||||
|
||||
namespace ipp
|
||||
{
|
||||
#if OPENCV_ABI_COMPATIBILITY > 300
|
||||
CV_EXPORTS unsigned long long getIppFeatures();
|
||||
#else
|
||||
CV_EXPORTS int getIppFeatures();
|
||||
#endif
|
||||
CV_EXPORTS void setIppStatus(int status, const char * const funcname = NULL, const char * const filename = NULL,
|
||||
int line = 0);
|
||||
CV_EXPORTS int getIppStatus();
|
||||
CV_EXPORTS String getIppErrorLocation();
|
||||
CV_EXPORTS_W bool useIPP();
|
||||
CV_EXPORTS_W void setUseIPP(bool flag);
|
||||
CV_EXPORTS_W String getIppVersion();
|
||||
|
||||
// IPP Not-Exact mode. This function may force use of IPP then both IPP and OpenCV provide proper results
|
||||
// but have internal accuracy differences which have too much direct or indirect impact on accuracy tests.
|
||||
CV_EXPORTS_W bool useIPP_NotExact();
|
||||
CV_EXPORTS_W void setUseIPP_NotExact(bool flag);
|
||||
#if OPENCV_ABI_COMPATIBILITY < 400
|
||||
CV_EXPORTS_W bool useIPP_NE();
|
||||
CV_EXPORTS_W void setUseIPP_NE(bool flag);
|
||||
#endif
|
||||
|
||||
} // ipp
|
||||
|
||||
//! @endcond
|
||||
|
||||
//! @} core_utils
|
||||
|
||||
|
||||
|
||||
|
||||
} // cv
|
||||
|
||||
#include "opencv2/core/neon_utils.hpp"
|
||||
#include "opencv2/core/vsx_utils.hpp"
|
||||
#include "opencv2/core/check.hpp"
|
||||
|
||||
#endif //OPENCV_CORE_BASE_HPP
|
||||
|
|
@ -0,0 +1,23 @@
|
|||
// This file is part of OpenCV project.
|
||||
// It is subject to the license terms in the LICENSE file found in the top-level directory
|
||||
// of this distribution and at http://opencv.org/license.html.
|
||||
|
||||
#ifndef OPENCV_CORE_BINDINGS_UTILS_HPP
|
||||
#define OPENCV_CORE_BINDINGS_UTILS_HPP
|
||||
|
||||
namespace cv { namespace utils {
|
||||
//! @addtogroup core_utils
|
||||
//! @{
|
||||
|
||||
CV_EXPORTS_W String dumpInputArray(InputArray argument);
|
||||
|
||||
CV_EXPORTS_W String dumpInputArrayOfArrays(InputArrayOfArrays argument);
|
||||
|
||||
CV_EXPORTS_W String dumpInputOutputArray(InputOutputArray argument);
|
||||
|
||||
CV_EXPORTS_W String dumpInputOutputArrayOfArrays(InputOutputArrayOfArrays argument);
|
||||
|
||||
//! @}
|
||||
}} // namespace
|
||||
|
||||
#endif // OPENCV_CORE_BINDINGS_UTILS_HPP
|
||||
|
|
@ -0,0 +1,40 @@
|
|||
// This file is part of OpenCV project.
|
||||
// It is subject to the license terms in the LICENSE file found in the top-level directory
|
||||
// of this distribution and at http://opencv.org/license.html.
|
||||
//
|
||||
// Copyright (C) 2014, Advanced Micro Devices, Inc., all rights reserved.
|
||||
|
||||
#ifndef OPENCV_CORE_BUFFER_POOL_HPP
|
||||
#define OPENCV_CORE_BUFFER_POOL_HPP
|
||||
|
||||
#ifdef _MSC_VER
|
||||
#pragma warning(push)
|
||||
#pragma warning(disable: 4265)
|
||||
#endif
|
||||
|
||||
namespace cv
|
||||
{
|
||||
|
||||
//! @addtogroup core
|
||||
//! @{
|
||||
|
||||
class BufferPoolController
|
||||
{
|
||||
protected:
|
||||
~BufferPoolController() { }
|
||||
public:
|
||||
virtual size_t getReservedSize() const = 0;
|
||||
virtual size_t getMaxReservedSize() const = 0;
|
||||
virtual void setMaxReservedSize(size_t size) = 0;
|
||||
virtual void freeAllReservedBuffers() = 0;
|
||||
};
|
||||
|
||||
//! @}
|
||||
|
||||
}
|
||||
|
||||
#ifdef _MSC_VER
|
||||
#pragma warning(pop)
|
||||
#endif
|
||||
|
||||
#endif // OPENCV_CORE_BUFFER_POOL_HPP
|
||||
|
|
@ -0,0 +1,157 @@
|
|||
// This file is part of OpenCV project.
|
||||
// It is subject to the license terms in the LICENSE file found in the top-level directory
|
||||
// of this distribution and at http://opencv.org/license.html.
|
||||
|
||||
#ifndef OPENCV_CORE_CHECK_HPP
|
||||
#define OPENCV_CORE_CHECK_HPP
|
||||
|
||||
#include <opencv2/core/base.hpp>
|
||||
|
||||
namespace cv {
|
||||
|
||||
/** Returns string of cv::Mat depth value: CV_8U -> "CV_8U" or "<invalid depth>" */
|
||||
CV_EXPORTS const char* depthToString(int depth);
|
||||
|
||||
/** Returns string of cv::Mat depth value: CV_8UC3 -> "CV_8UC3" or "<invalid type>" */
|
||||
CV_EXPORTS const String typeToString(int type);
|
||||
|
||||
|
||||
//! @cond IGNORED
|
||||
namespace detail {
|
||||
|
||||
/** Returns string of cv::Mat depth value: CV_8U -> "CV_8U" or NULL */
|
||||
CV_EXPORTS const char* depthToString_(int depth);
|
||||
|
||||
/** Returns string of cv::Mat depth value: CV_8UC3 -> "CV_8UC3" or cv::String() */
|
||||
CV_EXPORTS const cv::String typeToString_(int type);
|
||||
|
||||
enum TestOp {
|
||||
TEST_CUSTOM = 0,
|
||||
TEST_EQ = 1,
|
||||
TEST_NE = 2,
|
||||
TEST_LE = 3,
|
||||
TEST_LT = 4,
|
||||
TEST_GE = 5,
|
||||
TEST_GT = 6,
|
||||
CV__LAST_TEST_OP
|
||||
};
|
||||
|
||||
struct CheckContext {
|
||||
const char* func;
|
||||
const char* file;
|
||||
int line;
|
||||
enum TestOp testOp;
|
||||
const char* message;
|
||||
const char* p1_str;
|
||||
const char* p2_str;
|
||||
};
|
||||
|
||||
#ifndef CV__CHECK_FILENAME
|
||||
# define CV__CHECK_FILENAME __FILE__
|
||||
#endif
|
||||
|
||||
#ifndef CV__CHECK_FUNCTION
|
||||
# if defined _MSC_VER
|
||||
# define CV__CHECK_FUNCTION __FUNCSIG__
|
||||
# elif defined __GNUC__
|
||||
# define CV__CHECK_FUNCTION __PRETTY_FUNCTION__
|
||||
# else
|
||||
# define CV__CHECK_FUNCTION "<unknown>"
|
||||
# endif
|
||||
#endif
|
||||
|
||||
#define CV__CHECK_LOCATION_VARNAME(id) CVAUX_CONCAT(CVAUX_CONCAT(__cv_check_, id), __LINE__)
|
||||
#define CV__DEFINE_CHECK_CONTEXT(id, message, testOp, p1_str, p2_str) \
|
||||
static const cv::detail::CheckContext CV__CHECK_LOCATION_VARNAME(id) = \
|
||||
{ CV__CHECK_FUNCTION, CV__CHECK_FILENAME, __LINE__, testOp, message, p1_str, p2_str }
|
||||
|
||||
CV_EXPORTS void CV_NORETURN check_failed_auto(const int v1, const int v2, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_auto(const size_t v1, const size_t v2, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_auto(const float v1, const float v2, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_auto(const double v1, const double v2, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_MatDepth(const int v1, const int v2, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_MatType(const int v1, const int v2, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_MatChannels(const int v1, const int v2, const CheckContext& ctx);
|
||||
|
||||
CV_EXPORTS void CV_NORETURN check_failed_auto(const int v, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_auto(const size_t v, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_auto(const float v, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_auto(const double v, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_MatDepth(const int v, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_MatType(const int v, const CheckContext& ctx);
|
||||
CV_EXPORTS void CV_NORETURN check_failed_MatChannels(const int v, const CheckContext& ctx);
|
||||
|
||||
|
||||
#define CV__TEST_EQ(v1, v2) ((v1) == (v2))
|
||||
#define CV__TEST_NE(v1, v2) ((v1) != (v2))
|
||||
#define CV__TEST_LE(v1, v2) ((v1) <= (v2))
|
||||
#define CV__TEST_LT(v1, v2) ((v1) < (v2))
|
||||
#define CV__TEST_GE(v1, v2) ((v1) >= (v2))
|
||||
#define CV__TEST_GT(v1, v2) ((v1) > (v2))
|
||||
|
||||
#define CV__CHECK(id, op, type, v1, v2, v1_str, v2_str, msg_str) do { \
|
||||
if(CV__TEST_##op((v1), (v2))) ; else { \
|
||||
CV__DEFINE_CHECK_CONTEXT(id, msg_str, cv::detail::TEST_ ## op, v1_str, v2_str); \
|
||||
cv::detail::check_failed_ ## type((v1), (v2), CV__CHECK_LOCATION_VARNAME(id)); \
|
||||
} \
|
||||
} while (0)
|
||||
|
||||
#define CV__CHECK_CUSTOM_TEST(id, type, v, test_expr, v_str, test_expr_str, msg_str) do { \
|
||||
if(!!(test_expr)) ; else { \
|
||||
CV__DEFINE_CHECK_CONTEXT(id, msg_str, cv::detail::TEST_CUSTOM, v_str, test_expr_str); \
|
||||
cv::detail::check_failed_ ## type((v), CV__CHECK_LOCATION_VARNAME(id)); \
|
||||
} \
|
||||
} while (0)
|
||||
|
||||
} // namespace
|
||||
//! @endcond
|
||||
|
||||
|
||||
/// Supported values of these types: int, float, double
|
||||
#define CV_CheckEQ(v1, v2, msg) CV__CHECK(_, EQ, auto, v1, v2, #v1, #v2, msg)
|
||||
#define CV_CheckNE(v1, v2, msg) CV__CHECK(_, NE, auto, v1, v2, #v1, #v2, msg)
|
||||
#define CV_CheckLE(v1, v2, msg) CV__CHECK(_, LE, auto, v1, v2, #v1, #v2, msg)
|
||||
#define CV_CheckLT(v1, v2, msg) CV__CHECK(_, LT, auto, v1, v2, #v1, #v2, msg)
|
||||
#define CV_CheckGE(v1, v2, msg) CV__CHECK(_, GE, auto, v1, v2, #v1, #v2, msg)
|
||||
#define CV_CheckGT(v1, v2, msg) CV__CHECK(_, GT, auto, v1, v2, #v1, #v2, msg)
|
||||
|
||||
/// Check with additional "decoding" of type values in error message
|
||||
#define CV_CheckTypeEQ(t1, t2, msg) CV__CHECK(_, EQ, MatType, t1, t2, #t1, #t2, msg)
|
||||
/// Check with additional "decoding" of depth values in error message
|
||||
#define CV_CheckDepthEQ(d1, d2, msg) CV__CHECK(_, EQ, MatDepth, d1, d2, #d1, #d2, msg)
|
||||
|
||||
#define CV_CheckChannelsEQ(c1, c2, msg) CV__CHECK(_, EQ, MatChannels, c1, c2, #c1, #c2, msg)
|
||||
|
||||
/// Example: type == CV_8UC1 || type == CV_8UC3
|
||||
#define CV_CheckType(t, test_expr, msg) CV__CHECK_CUSTOM_TEST(_, MatType, t, (test_expr), #t, #test_expr, msg)
|
||||
|
||||
/// Example: depth == CV_32F || depth == CV_64F
|
||||
#define CV_CheckDepth(t, test_expr, msg) CV__CHECK_CUSTOM_TEST(_, MatDepth, t, (test_expr), #t, #test_expr, msg)
|
||||
|
||||
/// Example: v == A || v == B
|
||||
#define CV_Check(v, test_expr, msg) CV__CHECK_CUSTOM_TEST(_, auto, v, (test_expr), #v, #test_expr, msg)
|
||||
|
||||
/// Some complex conditions: CV_Check(src2, src2.empty() || (src2.type() == src1.type() && src2.size() == src1.size()), "src2 should have same size/type as src1")
|
||||
// TODO define pretty-printers
|
||||
|
||||
#ifndef NDEBUG
|
||||
#define CV_DbgCheck(v, test_expr, msg) CV__CHECK_CUSTOM_TEST(_, auto, v, (test_expr), #v, #test_expr, msg)
|
||||
#define CV_DbgCheckEQ(v1, v2, msg) CV__CHECK(_, EQ, auto, v1, v2, #v1, #v2, msg)
|
||||
#define CV_DbgCheckNE(v1, v2, msg) CV__CHECK(_, NE, auto, v1, v2, #v1, #v2, msg)
|
||||
#define CV_DbgCheckLE(v1, v2, msg) CV__CHECK(_, LE, auto, v1, v2, #v1, #v2, msg)
|
||||
#define CV_DbgCheckLT(v1, v2, msg) CV__CHECK(_, LT, auto, v1, v2, #v1, #v2, msg)
|
||||
#define CV_DbgCheckGE(v1, v2, msg) CV__CHECK(_, GE, auto, v1, v2, #v1, #v2, msg)
|
||||
#define CV_DbgCheckGT(v1, v2, msg) CV__CHECK(_, GT, auto, v1, v2, #v1, #v2, msg)
|
||||
#else
|
||||
#define CV_DbgCheck(v, test_expr, msg) do { } while (0)
|
||||
#define CV_DbgCheckEQ(v1, v2, msg) do { } while (0)
|
||||
#define CV_DbgCheckNE(v1, v2, msg) do { } while (0)
|
||||
#define CV_DbgCheckLE(v1, v2, msg) do { } while (0)
|
||||
#define CV_DbgCheckLT(v1, v2, msg) do { } while (0)
|
||||
#define CV_DbgCheckGE(v1, v2, msg) do { } while (0)
|
||||
#define CV_DbgCheckGT(v1, v2, msg) do { } while (0)
|
||||
#endif
|
||||
|
||||
} // namespace
|
||||
|
||||
#endif // OPENCV_CORE_CHECK_HPP
|
||||
|
|
@ -0,0 +1,48 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifdef __OPENCV_BUILD
|
||||
#error this is a compatibility header which should not be used inside the OpenCV library
|
||||
#endif
|
||||
|
||||
#include "opencv2/core.hpp"
|
||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,631 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CORE_CUDAINL_HPP
|
||||
#define OPENCV_CORE_CUDAINL_HPP
|
||||
|
||||
#include "opencv2/core/cuda.hpp"
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda {
|
||||
|
||||
//===================================================================================
|
||||
// GpuMat
|
||||
//===================================================================================
|
||||
|
||||
inline
|
||||
GpuMat::GpuMat(Allocator* allocator_)
|
||||
: flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
|
||||
{}
|
||||
|
||||
inline
|
||||
GpuMat::GpuMat(int rows_, int cols_, int type_, Allocator* allocator_)
|
||||
: flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
|
||||
{
|
||||
if (rows_ > 0 && cols_ > 0)
|
||||
create(rows_, cols_, type_);
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat::GpuMat(Size size_, int type_, Allocator* allocator_)
|
||||
: flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
|
||||
{
|
||||
if (size_.height > 0 && size_.width > 0)
|
||||
create(size_.height, size_.width, type_);
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat::GpuMat(int rows_, int cols_, int type_, Scalar s_, Allocator* allocator_)
|
||||
: flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
|
||||
{
|
||||
if (rows_ > 0 && cols_ > 0)
|
||||
{
|
||||
create(rows_, cols_, type_);
|
||||
setTo(s_);
|
||||
}
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat::GpuMat(Size size_, int type_, Scalar s_, Allocator* allocator_)
|
||||
: flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
|
||||
{
|
||||
if (size_.height > 0 && size_.width > 0)
|
||||
{
|
||||
create(size_.height, size_.width, type_);
|
||||
setTo(s_);
|
||||
}
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat::GpuMat(const GpuMat& m)
|
||||
: flags(m.flags), rows(m.rows), cols(m.cols), step(m.step), data(m.data), refcount(m.refcount), datastart(m.datastart), dataend(m.dataend), allocator(m.allocator)
|
||||
{
|
||||
if (refcount)
|
||||
CV_XADD(refcount, 1);
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat::GpuMat(InputArray arr, Allocator* allocator_) :
|
||||
flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), allocator(allocator_)
|
||||
{
|
||||
upload(arr);
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat::~GpuMat()
|
||||
{
|
||||
release();
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat& GpuMat::operator =(const GpuMat& m)
|
||||
{
|
||||
if (this != &m)
|
||||
{
|
||||
GpuMat temp(m);
|
||||
swap(temp);
|
||||
}
|
||||
|
||||
return *this;
|
||||
}
|
||||
|
||||
inline
|
||||
void GpuMat::create(Size size_, int type_)
|
||||
{
|
||||
create(size_.height, size_.width, type_);
|
||||
}
|
||||
|
||||
inline
|
||||
void GpuMat::swap(GpuMat& b)
|
||||
{
|
||||
std::swap(flags, b.flags);
|
||||
std::swap(rows, b.rows);
|
||||
std::swap(cols, b.cols);
|
||||
std::swap(step, b.step);
|
||||
std::swap(data, b.data);
|
||||
std::swap(datastart, b.datastart);
|
||||
std::swap(dataend, b.dataend);
|
||||
std::swap(refcount, b.refcount);
|
||||
std::swap(allocator, b.allocator);
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat GpuMat::clone() const
|
||||
{
|
||||
GpuMat m;
|
||||
copyTo(m);
|
||||
return m;
|
||||
}
|
||||
|
||||
inline
|
||||
void GpuMat::copyTo(OutputArray dst, InputArray mask) const
|
||||
{
|
||||
copyTo(dst, mask, Stream::Null());
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat& GpuMat::setTo(Scalar s)
|
||||
{
|
||||
return setTo(s, Stream::Null());
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat& GpuMat::setTo(Scalar s, InputArray mask)
|
||||
{
|
||||
return setTo(s, mask, Stream::Null());
|
||||
}
|
||||
|
||||
inline
|
||||
void GpuMat::convertTo(OutputArray dst, int rtype) const
|
||||
{
|
||||
convertTo(dst, rtype, Stream::Null());
|
||||
}
|
||||
|
||||
inline
|
||||
void GpuMat::convertTo(OutputArray dst, int rtype, double alpha, double beta) const
|
||||
{
|
||||
convertTo(dst, rtype, alpha, beta, Stream::Null());
|
||||
}
|
||||
|
||||
inline
|
||||
void GpuMat::convertTo(OutputArray dst, int rtype, double alpha, Stream& stream) const
|
||||
{
|
||||
convertTo(dst, rtype, alpha, 0.0, stream);
|
||||
}
|
||||
|
||||
inline
|
||||
void GpuMat::assignTo(GpuMat& m, int _type) const
|
||||
{
|
||||
if (_type < 0)
|
||||
m = *this;
|
||||
else
|
||||
convertTo(m, _type);
|
||||
}
|
||||
|
||||
inline
|
||||
uchar* GpuMat::ptr(int y)
|
||||
{
|
||||
CV_DbgAssert( (unsigned)y < (unsigned)rows );
|
||||
return data + step * y;
|
||||
}
|
||||
|
||||
inline
|
||||
const uchar* GpuMat::ptr(int y) const
|
||||
{
|
||||
CV_DbgAssert( (unsigned)y < (unsigned)rows );
|
||||
return data + step * y;
|
||||
}
|
||||
|
||||
template<typename _Tp> inline
|
||||
_Tp* GpuMat::ptr(int y)
|
||||
{
|
||||
return (_Tp*)ptr(y);
|
||||
}
|
||||
|
||||
template<typename _Tp> inline
|
||||
const _Tp* GpuMat::ptr(int y) const
|
||||
{
|
||||
return (const _Tp*)ptr(y);
|
||||
}
|
||||
|
||||
template <class T> inline
|
||||
GpuMat::operator PtrStepSz<T>() const
|
||||
{
|
||||
return PtrStepSz<T>(rows, cols, (T*)data, step);
|
||||
}
|
||||
|
||||
template <class T> inline
|
||||
GpuMat::operator PtrStep<T>() const
|
||||
{
|
||||
return PtrStep<T>((T*)data, step);
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat GpuMat::row(int y) const
|
||||
{
|
||||
return GpuMat(*this, Range(y, y+1), Range::all());
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat GpuMat::col(int x) const
|
||||
{
|
||||
return GpuMat(*this, Range::all(), Range(x, x+1));
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat GpuMat::rowRange(int startrow, int endrow) const
|
||||
{
|
||||
return GpuMat(*this, Range(startrow, endrow), Range::all());
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat GpuMat::rowRange(Range r) const
|
||||
{
|
||||
return GpuMat(*this, r, Range::all());
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat GpuMat::colRange(int startcol, int endcol) const
|
||||
{
|
||||
return GpuMat(*this, Range::all(), Range(startcol, endcol));
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat GpuMat::colRange(Range r) const
|
||||
{
|
||||
return GpuMat(*this, Range::all(), r);
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat GpuMat::operator ()(Range rowRange_, Range colRange_) const
|
||||
{
|
||||
return GpuMat(*this, rowRange_, colRange_);
|
||||
}
|
||||
|
||||
inline
|
||||
GpuMat GpuMat::operator ()(Rect roi) const
|
||||
{
|
||||
return GpuMat(*this, roi);
|
||||
}
|
||||
|
||||
inline
|
||||
bool GpuMat::isContinuous() const
|
||||
{
|
||||
return (flags & Mat::CONTINUOUS_FLAG) != 0;
|
||||
}
|
||||
|
||||
inline
|
||||
size_t GpuMat::elemSize() const
|
||||
{
|
||||
return CV_ELEM_SIZE(flags);
|
||||
}
|
||||
|
||||
inline
|
||||
size_t GpuMat::elemSize1() const
|
||||
{
|
||||
return CV_ELEM_SIZE1(flags);
|
||||
}
|
||||
|
||||
inline
|
||||
int GpuMat::type() const
|
||||
{
|
||||
return CV_MAT_TYPE(flags);
|
||||
}
|
||||
|
||||
inline
|
||||
int GpuMat::depth() const
|
||||
{
|
||||
return CV_MAT_DEPTH(flags);
|
||||
}
|
||||
|
||||
inline
|
||||
int GpuMat::channels() const
|
||||
{
|
||||
return CV_MAT_CN(flags);
|
||||
}
|
||||
|
||||
inline
|
||||
size_t GpuMat::step1() const
|
||||
{
|
||||
return step / elemSize1();
|
||||
}
|
||||
|
||||
inline
|
||||
Size GpuMat::size() const
|
||||
{
|
||||
return Size(cols, rows);
|
||||
}
|
||||
|
||||
inline
|
||||
bool GpuMat::empty() const
|
||||
{
|
||||
return data == 0;
|
||||
}
|
||||
|
||||
static inline
|
||||
GpuMat createContinuous(int rows, int cols, int type)
|
||||
{
|
||||
GpuMat m;
|
||||
createContinuous(rows, cols, type, m);
|
||||
return m;
|
||||
}
|
||||
|
||||
static inline
|
||||
void createContinuous(Size size, int type, OutputArray arr)
|
||||
{
|
||||
createContinuous(size.height, size.width, type, arr);
|
||||
}
|
||||
|
||||
static inline
|
||||
GpuMat createContinuous(Size size, int type)
|
||||
{
|
||||
GpuMat m;
|
||||
createContinuous(size, type, m);
|
||||
return m;
|
||||
}
|
||||
|
||||
static inline
|
||||
void ensureSizeIsEnough(Size size, int type, OutputArray arr)
|
||||
{
|
||||
ensureSizeIsEnough(size.height, size.width, type, arr);
|
||||
}
|
||||
|
||||
static inline
|
||||
void swap(GpuMat& a, GpuMat& b)
|
||||
{
|
||||
a.swap(b);
|
||||
}
|
||||
|
||||
//===================================================================================
|
||||
// HostMem
|
||||
//===================================================================================
|
||||
|
||||
inline
|
||||
HostMem::HostMem(AllocType alloc_type_)
|
||||
: flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), alloc_type(alloc_type_)
|
||||
{
|
||||
}
|
||||
|
||||
inline
|
||||
HostMem::HostMem(const HostMem& m)
|
||||
: flags(m.flags), rows(m.rows), cols(m.cols), step(m.step), data(m.data), refcount(m.refcount), datastart(m.datastart), dataend(m.dataend), alloc_type(m.alloc_type)
|
||||
{
|
||||
if( refcount )
|
||||
CV_XADD(refcount, 1);
|
||||
}
|
||||
|
||||
inline
|
||||
HostMem::HostMem(int rows_, int cols_, int type_, AllocType alloc_type_)
|
||||
: flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), alloc_type(alloc_type_)
|
||||
{
|
||||
if (rows_ > 0 && cols_ > 0)
|
||||
create(rows_, cols_, type_);
|
||||
}
|
||||
|
||||
inline
|
||||
HostMem::HostMem(Size size_, int type_, AllocType alloc_type_)
|
||||
: flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), alloc_type(alloc_type_)
|
||||
{
|
||||
if (size_.height > 0 && size_.width > 0)
|
||||
create(size_.height, size_.width, type_);
|
||||
}
|
||||
|
||||
inline
|
||||
HostMem::HostMem(InputArray arr, AllocType alloc_type_)
|
||||
: flags(0), rows(0), cols(0), step(0), data(0), refcount(0), datastart(0), dataend(0), alloc_type(alloc_type_)
|
||||
{
|
||||
arr.getMat().copyTo(*this);
|
||||
}
|
||||
|
||||
inline
|
||||
HostMem::~HostMem()
|
||||
{
|
||||
release();
|
||||
}
|
||||
|
||||
inline
|
||||
HostMem& HostMem::operator =(const HostMem& m)
|
||||
{
|
||||
if (this != &m)
|
||||
{
|
||||
HostMem temp(m);
|
||||
swap(temp);
|
||||
}
|
||||
|
||||
return *this;
|
||||
}
|
||||
|
||||
inline
|
||||
void HostMem::swap(HostMem& b)
|
||||
{
|
||||
std::swap(flags, b.flags);
|
||||
std::swap(rows, b.rows);
|
||||
std::swap(cols, b.cols);
|
||||
std::swap(step, b.step);
|
||||
std::swap(data, b.data);
|
||||
std::swap(datastart, b.datastart);
|
||||
std::swap(dataend, b.dataend);
|
||||
std::swap(refcount, b.refcount);
|
||||
std::swap(alloc_type, b.alloc_type);
|
||||
}
|
||||
|
||||
inline
|
||||
HostMem HostMem::clone() const
|
||||
{
|
||||
HostMem m(size(), type(), alloc_type);
|
||||
createMatHeader().copyTo(m);
|
||||
return m;
|
||||
}
|
||||
|
||||
inline
|
||||
void HostMem::create(Size size_, int type_)
|
||||
{
|
||||
create(size_.height, size_.width, type_);
|
||||
}
|
||||
|
||||
inline
|
||||
Mat HostMem::createMatHeader() const
|
||||
{
|
||||
return Mat(size(), type(), data, step);
|
||||
}
|
||||
|
||||
inline
|
||||
bool HostMem::isContinuous() const
|
||||
{
|
||||
return (flags & Mat::CONTINUOUS_FLAG) != 0;
|
||||
}
|
||||
|
||||
inline
|
||||
size_t HostMem::elemSize() const
|
||||
{
|
||||
return CV_ELEM_SIZE(flags);
|
||||
}
|
||||
|
||||
inline
|
||||
size_t HostMem::elemSize1() const
|
||||
{
|
||||
return CV_ELEM_SIZE1(flags);
|
||||
}
|
||||
|
||||
inline
|
||||
int HostMem::type() const
|
||||
{
|
||||
return CV_MAT_TYPE(flags);
|
||||
}
|
||||
|
||||
inline
|
||||
int HostMem::depth() const
|
||||
{
|
||||
return CV_MAT_DEPTH(flags);
|
||||
}
|
||||
|
||||
inline
|
||||
int HostMem::channels() const
|
||||
{
|
||||
return CV_MAT_CN(flags);
|
||||
}
|
||||
|
||||
inline
|
||||
size_t HostMem::step1() const
|
||||
{
|
||||
return step / elemSize1();
|
||||
}
|
||||
|
||||
inline
|
||||
Size HostMem::size() const
|
||||
{
|
||||
return Size(cols, rows);
|
||||
}
|
||||
|
||||
inline
|
||||
bool HostMem::empty() const
|
||||
{
|
||||
return data == 0;
|
||||
}
|
||||
|
||||
static inline
|
||||
void swap(HostMem& a, HostMem& b)
|
||||
{
|
||||
a.swap(b);
|
||||
}
|
||||
|
||||
//===================================================================================
|
||||
// Stream
|
||||
//===================================================================================
|
||||
|
||||
inline
|
||||
Stream::Stream(const Ptr<Impl>& impl)
|
||||
: impl_(impl)
|
||||
{
|
||||
}
|
||||
|
||||
//===================================================================================
|
||||
// Event
|
||||
//===================================================================================
|
||||
|
||||
inline
|
||||
Event::Event(const Ptr<Impl>& impl)
|
||||
: impl_(impl)
|
||||
{
|
||||
}
|
||||
|
||||
//===================================================================================
|
||||
// Initialization & Info
|
||||
//===================================================================================
|
||||
|
||||
inline
|
||||
bool TargetArchs::has(int major, int minor)
|
||||
{
|
||||
return hasPtx(major, minor) || hasBin(major, minor);
|
||||
}
|
||||
|
||||
inline
|
||||
bool TargetArchs::hasEqualOrGreater(int major, int minor)
|
||||
{
|
||||
return hasEqualOrGreaterPtx(major, minor) || hasEqualOrGreaterBin(major, minor);
|
||||
}
|
||||
|
||||
inline
|
||||
DeviceInfo::DeviceInfo()
|
||||
{
|
||||
device_id_ = getDevice();
|
||||
}
|
||||
|
||||
inline
|
||||
DeviceInfo::DeviceInfo(int device_id)
|
||||
{
|
||||
CV_Assert( device_id >= 0 && device_id < getCudaEnabledDeviceCount() );
|
||||
device_id_ = device_id;
|
||||
}
|
||||
|
||||
inline
|
||||
int DeviceInfo::deviceID() const
|
||||
{
|
||||
return device_id_;
|
||||
}
|
||||
|
||||
inline
|
||||
size_t DeviceInfo::freeMemory() const
|
||||
{
|
||||
size_t _totalMemory = 0, _freeMemory = 0;
|
||||
queryMemory(_totalMemory, _freeMemory);
|
||||
return _freeMemory;
|
||||
}
|
||||
|
||||
inline
|
||||
size_t DeviceInfo::totalMemory() const
|
||||
{
|
||||
size_t _totalMemory = 0, _freeMemory = 0;
|
||||
queryMemory(_totalMemory, _freeMemory);
|
||||
return _totalMemory;
|
||||
}
|
||||
|
||||
inline
|
||||
bool DeviceInfo::supports(FeatureSet feature_set) const
|
||||
{
|
||||
int version = majorVersion() * 10 + minorVersion();
|
||||
return version >= feature_set;
|
||||
}
|
||||
|
||||
|
||||
}} // namespace cv { namespace cuda {
|
||||
|
||||
//===================================================================================
|
||||
// Mat
|
||||
//===================================================================================
|
||||
|
||||
namespace cv {
|
||||
|
||||
inline
|
||||
Mat::Mat(const cuda::GpuMat& m)
|
||||
: flags(0), dims(0), rows(0), cols(0), data(0), datastart(0), dataend(0), datalimit(0), allocator(0), u(0), size(&rows)
|
||||
{
|
||||
m.download(*this);
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CORE_CUDAINL_HPP
|
||||
|
|
@ -0,0 +1,211 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_DEVICE_BLOCK_HPP
|
||||
#define OPENCV_CUDA_DEVICE_BLOCK_HPP
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
struct Block
|
||||
{
|
||||
static __device__ __forceinline__ unsigned int id()
|
||||
{
|
||||
return blockIdx.x;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int stride()
|
||||
{
|
||||
return blockDim.x * blockDim.y * blockDim.z;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ void sync()
|
||||
{
|
||||
__syncthreads();
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ int flattenedThreadId()
|
||||
{
|
||||
return threadIdx.z * blockDim.x * blockDim.y + threadIdx.y * blockDim.x + threadIdx.x;
|
||||
}
|
||||
|
||||
template<typename It, typename T>
|
||||
static __device__ __forceinline__ void fill(It beg, It end, const T& value)
|
||||
{
|
||||
int STRIDE = stride();
|
||||
It t = beg + flattenedThreadId();
|
||||
|
||||
for(; t < end; t += STRIDE)
|
||||
*t = value;
|
||||
}
|
||||
|
||||
template<typename OutIt, typename T>
|
||||
static __device__ __forceinline__ void yota(OutIt beg, OutIt end, T value)
|
||||
{
|
||||
int STRIDE = stride();
|
||||
int tid = flattenedThreadId();
|
||||
value += tid;
|
||||
|
||||
for(OutIt t = beg + tid; t < end; t += STRIDE, value += STRIDE)
|
||||
*t = value;
|
||||
}
|
||||
|
||||
template<typename InIt, typename OutIt>
|
||||
static __device__ __forceinline__ void copy(InIt beg, InIt end, OutIt out)
|
||||
{
|
||||
int STRIDE = stride();
|
||||
InIt t = beg + flattenedThreadId();
|
||||
OutIt o = out + (t - beg);
|
||||
|
||||
for(; t < end; t += STRIDE, o += STRIDE)
|
||||
*o = *t;
|
||||
}
|
||||
|
||||
template<typename InIt, typename OutIt, class UnOp>
|
||||
static __device__ __forceinline__ void transform(InIt beg, InIt end, OutIt out, UnOp op)
|
||||
{
|
||||
int STRIDE = stride();
|
||||
InIt t = beg + flattenedThreadId();
|
||||
OutIt o = out + (t - beg);
|
||||
|
||||
for(; t < end; t += STRIDE, o += STRIDE)
|
||||
*o = op(*t);
|
||||
}
|
||||
|
||||
template<typename InIt1, typename InIt2, typename OutIt, class BinOp>
|
||||
static __device__ __forceinline__ void transform(InIt1 beg1, InIt1 end1, InIt2 beg2, OutIt out, BinOp op)
|
||||
{
|
||||
int STRIDE = stride();
|
||||
InIt1 t1 = beg1 + flattenedThreadId();
|
||||
InIt2 t2 = beg2 + flattenedThreadId();
|
||||
OutIt o = out + (t1 - beg1);
|
||||
|
||||
for(; t1 < end1; t1 += STRIDE, t2 += STRIDE, o += STRIDE)
|
||||
*o = op(*t1, *t2);
|
||||
}
|
||||
|
||||
template<int CTA_SIZE, typename T, class BinOp>
|
||||
static __device__ __forceinline__ void reduce(volatile T* buffer, BinOp op)
|
||||
{
|
||||
int tid = flattenedThreadId();
|
||||
T val = buffer[tid];
|
||||
|
||||
if (CTA_SIZE >= 1024) { if (tid < 512) buffer[tid] = val = op(val, buffer[tid + 512]); __syncthreads(); }
|
||||
if (CTA_SIZE >= 512) { if (tid < 256) buffer[tid] = val = op(val, buffer[tid + 256]); __syncthreads(); }
|
||||
if (CTA_SIZE >= 256) { if (tid < 128) buffer[tid] = val = op(val, buffer[tid + 128]); __syncthreads(); }
|
||||
if (CTA_SIZE >= 128) { if (tid < 64) buffer[tid] = val = op(val, buffer[tid + 64]); __syncthreads(); }
|
||||
|
||||
if (tid < 32)
|
||||
{
|
||||
if (CTA_SIZE >= 64) { buffer[tid] = val = op(val, buffer[tid + 32]); }
|
||||
if (CTA_SIZE >= 32) { buffer[tid] = val = op(val, buffer[tid + 16]); }
|
||||
if (CTA_SIZE >= 16) { buffer[tid] = val = op(val, buffer[tid + 8]); }
|
||||
if (CTA_SIZE >= 8) { buffer[tid] = val = op(val, buffer[tid + 4]); }
|
||||
if (CTA_SIZE >= 4) { buffer[tid] = val = op(val, buffer[tid + 2]); }
|
||||
if (CTA_SIZE >= 2) { buffer[tid] = val = op(val, buffer[tid + 1]); }
|
||||
}
|
||||
}
|
||||
|
||||
template<int CTA_SIZE, typename T, class BinOp>
|
||||
static __device__ __forceinline__ T reduce(volatile T* buffer, T init, BinOp op)
|
||||
{
|
||||
int tid = flattenedThreadId();
|
||||
T val = buffer[tid] = init;
|
||||
__syncthreads();
|
||||
|
||||
if (CTA_SIZE >= 1024) { if (tid < 512) buffer[tid] = val = op(val, buffer[tid + 512]); __syncthreads(); }
|
||||
if (CTA_SIZE >= 512) { if (tid < 256) buffer[tid] = val = op(val, buffer[tid + 256]); __syncthreads(); }
|
||||
if (CTA_SIZE >= 256) { if (tid < 128) buffer[tid] = val = op(val, buffer[tid + 128]); __syncthreads(); }
|
||||
if (CTA_SIZE >= 128) { if (tid < 64) buffer[tid] = val = op(val, buffer[tid + 64]); __syncthreads(); }
|
||||
|
||||
if (tid < 32)
|
||||
{
|
||||
if (CTA_SIZE >= 64) { buffer[tid] = val = op(val, buffer[tid + 32]); }
|
||||
if (CTA_SIZE >= 32) { buffer[tid] = val = op(val, buffer[tid + 16]); }
|
||||
if (CTA_SIZE >= 16) { buffer[tid] = val = op(val, buffer[tid + 8]); }
|
||||
if (CTA_SIZE >= 8) { buffer[tid] = val = op(val, buffer[tid + 4]); }
|
||||
if (CTA_SIZE >= 4) { buffer[tid] = val = op(val, buffer[tid + 2]); }
|
||||
if (CTA_SIZE >= 2) { buffer[tid] = val = op(val, buffer[tid + 1]); }
|
||||
}
|
||||
__syncthreads();
|
||||
return buffer[0];
|
||||
}
|
||||
|
||||
template <typename T, class BinOp>
|
||||
static __device__ __forceinline__ void reduce_n(T* data, unsigned int n, BinOp op)
|
||||
{
|
||||
int ftid = flattenedThreadId();
|
||||
int sft = stride();
|
||||
|
||||
if (sft < n)
|
||||
{
|
||||
for (unsigned int i = sft + ftid; i < n; i += sft)
|
||||
data[ftid] = op(data[ftid], data[i]);
|
||||
|
||||
__syncthreads();
|
||||
|
||||
n = sft;
|
||||
}
|
||||
|
||||
while (n > 1)
|
||||
{
|
||||
unsigned int half = n/2;
|
||||
|
||||
if (ftid < half)
|
||||
data[ftid] = op(data[ftid], data[n - ftid - 1]);
|
||||
|
||||
__syncthreads();
|
||||
|
||||
n = n - half;
|
||||
}
|
||||
}
|
||||
};
|
||||
}}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif /* OPENCV_CUDA_DEVICE_BLOCK_HPP */
|
||||
|
|
@ -0,0 +1,722 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_BORDER_INTERPOLATE_HPP
|
||||
#define OPENCV_CUDA_BORDER_INTERPOLATE_HPP
|
||||
|
||||
#include "saturate_cast.hpp"
|
||||
#include "vec_traits.hpp"
|
||||
#include "vec_math.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
//////////////////////////////////////////////////////////////
|
||||
// BrdConstant
|
||||
|
||||
template <typename D> struct BrdRowConstant
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ BrdRowConstant(int width_, const D& val_ = VecTraits<D>::all(0)) : width(width_), val(val_) {}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_low(int x, const T* data) const
|
||||
{
|
||||
return x >= 0 ? saturate_cast<D>(data[x]) : val;
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_high(int x, const T* data) const
|
||||
{
|
||||
return x < width ? saturate_cast<D>(data[x]) : val;
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int x, const T* data) const
|
||||
{
|
||||
return (x >= 0 && x < width) ? saturate_cast<D>(data[x]) : val;
|
||||
}
|
||||
|
||||
int width;
|
||||
D val;
|
||||
};
|
||||
|
||||
template <typename D> struct BrdColConstant
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ BrdColConstant(int height_, const D& val_ = VecTraits<D>::all(0)) : height(height_), val(val_) {}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_low(int y, const T* data, size_t step) const
|
||||
{
|
||||
return y >= 0 ? saturate_cast<D>(*(const T*)((const char*)data + y * step)) : val;
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_high(int y, const T* data, size_t step) const
|
||||
{
|
||||
return y < height ? saturate_cast<D>(*(const T*)((const char*)data + y * step)) : val;
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int y, const T* data, size_t step) const
|
||||
{
|
||||
return (y >= 0 && y < height) ? saturate_cast<D>(*(const T*)((const char*)data + y * step)) : val;
|
||||
}
|
||||
|
||||
int height;
|
||||
D val;
|
||||
};
|
||||
|
||||
template <typename D> struct BrdConstant
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
__host__ __device__ __forceinline__ BrdConstant(int height_, int width_, const D& val_ = VecTraits<D>::all(0)) : height(height_), width(width_), val(val_)
|
||||
{
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int y, int x, const T* data, size_t step) const
|
||||
{
|
||||
return (x >= 0 && x < width && y >= 0 && y < height) ? saturate_cast<D>(((const T*)((const uchar*)data + y * step))[x]) : val;
|
||||
}
|
||||
|
||||
template <typename Ptr2D> __device__ __forceinline__ D at(typename Ptr2D::index_type y, typename Ptr2D::index_type x, const Ptr2D& src) const
|
||||
{
|
||||
return (x >= 0 && x < width && y >= 0 && y < height) ? saturate_cast<D>(src(y, x)) : val;
|
||||
}
|
||||
|
||||
int height;
|
||||
int width;
|
||||
D val;
|
||||
};
|
||||
|
||||
//////////////////////////////////////////////////////////////
|
||||
// BrdReplicate
|
||||
|
||||
template <typename D> struct BrdRowReplicate
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ BrdRowReplicate(int width) : last_col(width - 1) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdRowReplicate(int width, U) : last_col(width - 1) {}
|
||||
|
||||
__device__ __forceinline__ int idx_col_low(int x) const
|
||||
{
|
||||
return ::max(x, 0);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_high(int x) const
|
||||
{
|
||||
return ::min(x, last_col);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col(int x) const
|
||||
{
|
||||
return idx_col_low(idx_col_high(x));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_low(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col_low(x)]);
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_high(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col_high(x)]);
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col(x)]);
|
||||
}
|
||||
|
||||
int last_col;
|
||||
};
|
||||
|
||||
template <typename D> struct BrdColReplicate
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ BrdColReplicate(int height) : last_row(height - 1) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdColReplicate(int height, U) : last_row(height - 1) {}
|
||||
|
||||
__device__ __forceinline__ int idx_row_low(int y) const
|
||||
{
|
||||
return ::max(y, 0);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row_high(int y) const
|
||||
{
|
||||
return ::min(y, last_row);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row(int y) const
|
||||
{
|
||||
return idx_row_low(idx_row_high(y));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_low(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const T*)((const char*)data + idx_row_low(y) * step));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_high(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const T*)((const char*)data + idx_row_high(y) * step));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const T*)((const char*)data + idx_row(y) * step));
|
||||
}
|
||||
|
||||
int last_row;
|
||||
};
|
||||
|
||||
template <typename D> struct BrdReplicate
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
__host__ __device__ __forceinline__ BrdReplicate(int height, int width) : last_row(height - 1), last_col(width - 1) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdReplicate(int height, int width, U) : last_row(height - 1), last_col(width - 1) {}
|
||||
|
||||
__device__ __forceinline__ int idx_row_low(int y) const
|
||||
{
|
||||
return ::max(y, 0);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row_high(int y) const
|
||||
{
|
||||
return ::min(y, last_row);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row(int y) const
|
||||
{
|
||||
return idx_row_low(idx_row_high(y));
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_low(int x) const
|
||||
{
|
||||
return ::max(x, 0);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_high(int x) const
|
||||
{
|
||||
return ::min(x, last_col);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col(int x) const
|
||||
{
|
||||
return idx_col_low(idx_col_high(x));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int y, int x, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(((const T*)((const char*)data + idx_row(y) * step))[idx_col(x)]);
|
||||
}
|
||||
|
||||
template <typename Ptr2D> __device__ __forceinline__ D at(typename Ptr2D::index_type y, typename Ptr2D::index_type x, const Ptr2D& src) const
|
||||
{
|
||||
return saturate_cast<D>(src(idx_row(y), idx_col(x)));
|
||||
}
|
||||
|
||||
int last_row;
|
||||
int last_col;
|
||||
};
|
||||
|
||||
//////////////////////////////////////////////////////////////
|
||||
// BrdReflect101
|
||||
|
||||
template <typename D> struct BrdRowReflect101
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ BrdRowReflect101(int width) : last_col(width - 1) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdRowReflect101(int width, U) : last_col(width - 1) {}
|
||||
|
||||
__device__ __forceinline__ int idx_col_low(int x) const
|
||||
{
|
||||
return ::abs(x) % (last_col + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_high(int x) const
|
||||
{
|
||||
return ::abs(last_col - ::abs(last_col - x)) % (last_col + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col(int x) const
|
||||
{
|
||||
return idx_col_low(idx_col_high(x));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_low(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col_low(x)]);
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_high(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col_high(x)]);
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col(x)]);
|
||||
}
|
||||
|
||||
int last_col;
|
||||
};
|
||||
|
||||
template <typename D> struct BrdColReflect101
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ BrdColReflect101(int height) : last_row(height - 1) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdColReflect101(int height, U) : last_row(height - 1) {}
|
||||
|
||||
__device__ __forceinline__ int idx_row_low(int y) const
|
||||
{
|
||||
return ::abs(y) % (last_row + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row_high(int y) const
|
||||
{
|
||||
return ::abs(last_row - ::abs(last_row - y)) % (last_row + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row(int y) const
|
||||
{
|
||||
return idx_row_low(idx_row_high(y));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_low(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const D*)((const char*)data + idx_row_low(y) * step));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_high(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const D*)((const char*)data + idx_row_high(y) * step));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const D*)((const char*)data + idx_row(y) * step));
|
||||
}
|
||||
|
||||
int last_row;
|
||||
};
|
||||
|
||||
template <typename D> struct BrdReflect101
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
__host__ __device__ __forceinline__ BrdReflect101(int height, int width) : last_row(height - 1), last_col(width - 1) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdReflect101(int height, int width, U) : last_row(height - 1), last_col(width - 1) {}
|
||||
|
||||
__device__ __forceinline__ int idx_row_low(int y) const
|
||||
{
|
||||
return ::abs(y) % (last_row + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row_high(int y) const
|
||||
{
|
||||
return ::abs(last_row - ::abs(last_row - y)) % (last_row + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row(int y) const
|
||||
{
|
||||
return idx_row_low(idx_row_high(y));
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_low(int x) const
|
||||
{
|
||||
return ::abs(x) % (last_col + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_high(int x) const
|
||||
{
|
||||
return ::abs(last_col - ::abs(last_col - x)) % (last_col + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col(int x) const
|
||||
{
|
||||
return idx_col_low(idx_col_high(x));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int y, int x, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(((const T*)((const char*)data + idx_row(y) * step))[idx_col(x)]);
|
||||
}
|
||||
|
||||
template <typename Ptr2D> __device__ __forceinline__ D at(typename Ptr2D::index_type y, typename Ptr2D::index_type x, const Ptr2D& src) const
|
||||
{
|
||||
return saturate_cast<D>(src(idx_row(y), idx_col(x)));
|
||||
}
|
||||
|
||||
int last_row;
|
||||
int last_col;
|
||||
};
|
||||
|
||||
//////////////////////////////////////////////////////////////
|
||||
// BrdReflect
|
||||
|
||||
template <typename D> struct BrdRowReflect
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ BrdRowReflect(int width) : last_col(width - 1) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdRowReflect(int width, U) : last_col(width - 1) {}
|
||||
|
||||
__device__ __forceinline__ int idx_col_low(int x) const
|
||||
{
|
||||
return (::abs(x) - (x < 0)) % (last_col + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_high(int x) const
|
||||
{
|
||||
return ::abs(last_col - ::abs(last_col - x) + (x > last_col)) % (last_col + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col(int x) const
|
||||
{
|
||||
return idx_col_high(::abs(x) - (x < 0));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_low(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col_low(x)]);
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_high(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col_high(x)]);
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col(x)]);
|
||||
}
|
||||
|
||||
int last_col;
|
||||
};
|
||||
|
||||
template <typename D> struct BrdColReflect
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ BrdColReflect(int height) : last_row(height - 1) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdColReflect(int height, U) : last_row(height - 1) {}
|
||||
|
||||
__device__ __forceinline__ int idx_row_low(int y) const
|
||||
{
|
||||
return (::abs(y) - (y < 0)) % (last_row + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row_high(int y) const
|
||||
{
|
||||
return ::abs(last_row - ::abs(last_row - y) + (y > last_row)) % (last_row + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row(int y) const
|
||||
{
|
||||
return idx_row_high(::abs(y) - (y < 0));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_low(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const D*)((const char*)data + idx_row_low(y) * step));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_high(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const D*)((const char*)data + idx_row_high(y) * step));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const D*)((const char*)data + idx_row(y) * step));
|
||||
}
|
||||
|
||||
int last_row;
|
||||
};
|
||||
|
||||
template <typename D> struct BrdReflect
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
__host__ __device__ __forceinline__ BrdReflect(int height, int width) : last_row(height - 1), last_col(width - 1) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdReflect(int height, int width, U) : last_row(height - 1), last_col(width - 1) {}
|
||||
|
||||
__device__ __forceinline__ int idx_row_low(int y) const
|
||||
{
|
||||
return (::abs(y) - (y < 0)) % (last_row + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row_high(int y) const
|
||||
{
|
||||
return /*::abs*/(last_row - ::abs(last_row - y) + (y > last_row)) /*% (last_row + 1)*/;
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row(int y) const
|
||||
{
|
||||
return idx_row_low(idx_row_high(y));
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_low(int x) const
|
||||
{
|
||||
return (::abs(x) - (x < 0)) % (last_col + 1);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_high(int x) const
|
||||
{
|
||||
return (last_col - ::abs(last_col - x) + (x > last_col));
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col(int x) const
|
||||
{
|
||||
return idx_col_low(idx_col_high(x));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int y, int x, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(((const T*)((const char*)data + idx_row(y) * step))[idx_col(x)]);
|
||||
}
|
||||
|
||||
template <typename Ptr2D> __device__ __forceinline__ D at(typename Ptr2D::index_type y, typename Ptr2D::index_type x, const Ptr2D& src) const
|
||||
{
|
||||
return saturate_cast<D>(src(idx_row(y), idx_col(x)));
|
||||
}
|
||||
|
||||
int last_row;
|
||||
int last_col;
|
||||
};
|
||||
|
||||
//////////////////////////////////////////////////////////////
|
||||
// BrdWrap
|
||||
|
||||
template <typename D> struct BrdRowWrap
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ BrdRowWrap(int width_) : width(width_) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdRowWrap(int width_, U) : width(width_) {}
|
||||
|
||||
__device__ __forceinline__ int idx_col_low(int x) const
|
||||
{
|
||||
return (x >= 0) * x + (x < 0) * (x - ((x - width + 1) / width) * width);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_high(int x) const
|
||||
{
|
||||
return (x < width) * x + (x >= width) * (x % width);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col(int x) const
|
||||
{
|
||||
return idx_col_high(idx_col_low(x));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_low(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col_low(x)]);
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_high(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col_high(x)]);
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int x, const T* data) const
|
||||
{
|
||||
return saturate_cast<D>(data[idx_col(x)]);
|
||||
}
|
||||
|
||||
int width;
|
||||
};
|
||||
|
||||
template <typename D> struct BrdColWrap
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ BrdColWrap(int height_) : height(height_) {}
|
||||
template <typename U> __host__ __device__ __forceinline__ BrdColWrap(int height_, U) : height(height_) {}
|
||||
|
||||
__device__ __forceinline__ int idx_row_low(int y) const
|
||||
{
|
||||
return (y >= 0) * y + (y < 0) * (y - ((y - height + 1) / height) * height);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row_high(int y) const
|
||||
{
|
||||
return (y < height) * y + (y >= height) * (y % height);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row(int y) const
|
||||
{
|
||||
return idx_row_high(idx_row_low(y));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_low(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const D*)((const char*)data + idx_row_low(y) * step));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at_high(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const D*)((const char*)data + idx_row_high(y) * step));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int y, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(*(const D*)((const char*)data + idx_row(y) * step));
|
||||
}
|
||||
|
||||
int height;
|
||||
};
|
||||
|
||||
template <typename D> struct BrdWrap
|
||||
{
|
||||
typedef D result_type;
|
||||
|
||||
__host__ __device__ __forceinline__ BrdWrap(int height_, int width_) :
|
||||
height(height_), width(width_)
|
||||
{
|
||||
}
|
||||
template <typename U>
|
||||
__host__ __device__ __forceinline__ BrdWrap(int height_, int width_, U) :
|
||||
height(height_), width(width_)
|
||||
{
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row_low(int y) const
|
||||
{
|
||||
return (y >= 0) ? y : (y - ((y - height + 1) / height) * height);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row_high(int y) const
|
||||
{
|
||||
return (y < height) ? y : (y % height);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_row(int y) const
|
||||
{
|
||||
return idx_row_high(idx_row_low(y));
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_low(int x) const
|
||||
{
|
||||
return (x >= 0) ? x : (x - ((x - width + 1) / width) * width);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col_high(int x) const
|
||||
{
|
||||
return (x < width) ? x : (x % width);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ int idx_col(int x) const
|
||||
{
|
||||
return idx_col_high(idx_col_low(x));
|
||||
}
|
||||
|
||||
template <typename T> __device__ __forceinline__ D at(int y, int x, const T* data, size_t step) const
|
||||
{
|
||||
return saturate_cast<D>(((const T*)((const char*)data + idx_row(y) * step))[idx_col(x)]);
|
||||
}
|
||||
|
||||
template <typename Ptr2D> __device__ __forceinline__ D at(typename Ptr2D::index_type y, typename Ptr2D::index_type x, const Ptr2D& src) const
|
||||
{
|
||||
return saturate_cast<D>(src(idx_row(y), idx_col(x)));
|
||||
}
|
||||
|
||||
int height;
|
||||
int width;
|
||||
};
|
||||
|
||||
//////////////////////////////////////////////////////////////
|
||||
// BorderReader
|
||||
|
||||
template <typename Ptr2D, typename B> struct BorderReader
|
||||
{
|
||||
typedef typename B::result_type elem_type;
|
||||
typedef typename Ptr2D::index_type index_type;
|
||||
|
||||
__host__ __device__ __forceinline__ BorderReader(const Ptr2D& ptr_, const B& b_) : ptr(ptr_), b(b_) {}
|
||||
|
||||
__device__ __forceinline__ elem_type operator ()(index_type y, index_type x) const
|
||||
{
|
||||
return b.at(y, x, ptr);
|
||||
}
|
||||
|
||||
Ptr2D ptr;
|
||||
B b;
|
||||
};
|
||||
|
||||
// under win32 there is some bug with templated types that passed as kernel parameters
|
||||
// with this specialization all works fine
|
||||
template <typename Ptr2D, typename D> struct BorderReader< Ptr2D, BrdConstant<D> >
|
||||
{
|
||||
typedef typename BrdConstant<D>::result_type elem_type;
|
||||
typedef typename Ptr2D::index_type index_type;
|
||||
|
||||
__host__ __device__ __forceinline__ BorderReader(const Ptr2D& src_, const BrdConstant<D>& b) :
|
||||
src(src_), height(b.height), width(b.width), val(b.val)
|
||||
{
|
||||
}
|
||||
|
||||
__device__ __forceinline__ D operator ()(index_type y, index_type x) const
|
||||
{
|
||||
return (x >= 0 && x < width && y >= 0 && y < height) ? saturate_cast<D>(src(y, x)) : val;
|
||||
}
|
||||
|
||||
Ptr2D src;
|
||||
int height;
|
||||
int width;
|
||||
D val;
|
||||
};
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_BORDER_INTERPOLATE_HPP
|
||||
|
|
@ -0,0 +1,309 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_COLOR_HPP
|
||||
#define OPENCV_CUDA_COLOR_HPP
|
||||
|
||||
#include "detail/color_detail.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
// All OPENCV_CUDA_IMPLEMENT_*_TRAITS(ColorSpace1_to_ColorSpace2, ...) macros implements
|
||||
// template <typename T> class ColorSpace1_to_ColorSpace2_traits
|
||||
// {
|
||||
// typedef ... functor_type;
|
||||
// static __host__ __device__ functor_type create_functor();
|
||||
// };
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB_TRAITS(bgr_to_rgb, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB_TRAITS(bgr_to_bgra, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB_TRAITS(bgr_to_rgba, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB_TRAITS(bgra_to_bgr, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB_TRAITS(bgra_to_rgb, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB_TRAITS(bgra_to_rgba, 4, 4, 2)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB2RGB_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB5x5_TRAITS(bgr_to_bgr555, 3, 0, 5)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB5x5_TRAITS(bgr_to_bgr565, 3, 0, 6)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB5x5_TRAITS(rgb_to_bgr555, 3, 2, 5)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB5x5_TRAITS(rgb_to_bgr565, 3, 2, 6)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB5x5_TRAITS(bgra_to_bgr555, 4, 0, 5)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB5x5_TRAITS(bgra_to_bgr565, 4, 0, 6)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB5x5_TRAITS(rgba_to_bgr555, 4, 2, 5)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2RGB5x5_TRAITS(rgba_to_bgr565, 4, 2, 6)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB2RGB5x5_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB5x52RGB_TRAITS(bgr555_to_rgb, 3, 2, 5)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB5x52RGB_TRAITS(bgr565_to_rgb, 3, 2, 6)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB5x52RGB_TRAITS(bgr555_to_bgr, 3, 0, 5)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB5x52RGB_TRAITS(bgr565_to_bgr, 3, 0, 6)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB5x52RGB_TRAITS(bgr555_to_rgba, 4, 2, 5)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB5x52RGB_TRAITS(bgr565_to_rgba, 4, 2, 6)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB5x52RGB_TRAITS(bgr555_to_bgra, 4, 0, 5)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB5x52RGB_TRAITS(bgr565_to_bgra, 4, 0, 6)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB5x52RGB_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_GRAY2RGB_TRAITS(gray_to_bgr, 3)
|
||||
OPENCV_CUDA_IMPLEMENT_GRAY2RGB_TRAITS(gray_to_bgra, 4)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_GRAY2RGB_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_GRAY2RGB5x5_TRAITS(gray_to_bgr555, 5)
|
||||
OPENCV_CUDA_IMPLEMENT_GRAY2RGB5x5_TRAITS(gray_to_bgr565, 6)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_GRAY2RGB5x5_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB5x52GRAY_TRAITS(bgr555_to_gray, 5)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB5x52GRAY_TRAITS(bgr565_to_gray, 6)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB5x52GRAY_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2GRAY_TRAITS(rgb_to_gray, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2GRAY_TRAITS(bgr_to_gray, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2GRAY_TRAITS(rgba_to_gray, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2GRAY_TRAITS(bgra_to_gray, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB2GRAY_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YUV_TRAITS(rgb_to_yuv, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YUV_TRAITS(rgba_to_yuv, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YUV_TRAITS(rgb_to_yuv4, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YUV_TRAITS(rgba_to_yuv4, 4, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YUV_TRAITS(bgr_to_yuv, 3, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YUV_TRAITS(bgra_to_yuv, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YUV_TRAITS(bgr_to_yuv4, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YUV_TRAITS(bgra_to_yuv4, 4, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB2YUV_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_YUV2RGB_TRAITS(yuv_to_rgb, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_YUV2RGB_TRAITS(yuv_to_rgba, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_YUV2RGB_TRAITS(yuv4_to_rgb, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_YUV2RGB_TRAITS(yuv4_to_rgba, 4, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_YUV2RGB_TRAITS(yuv_to_bgr, 3, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_YUV2RGB_TRAITS(yuv_to_bgra, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_YUV2RGB_TRAITS(yuv4_to_bgr, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_YUV2RGB_TRAITS(yuv4_to_bgra, 4, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_YUV2RGB_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YCrCb_TRAITS(rgb_to_YCrCb, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YCrCb_TRAITS(rgba_to_YCrCb, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YCrCb_TRAITS(rgb_to_YCrCb4, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YCrCb_TRAITS(rgba_to_YCrCb4, 4, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YCrCb_TRAITS(bgr_to_YCrCb, 3, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YCrCb_TRAITS(bgra_to_YCrCb, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YCrCb_TRAITS(bgr_to_YCrCb4, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2YCrCb_TRAITS(bgra_to_YCrCb4, 4, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB2YCrCb_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_YCrCb2RGB_TRAITS(YCrCb_to_rgb, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_YCrCb2RGB_TRAITS(YCrCb_to_rgba, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_YCrCb2RGB_TRAITS(YCrCb4_to_rgb, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_YCrCb2RGB_TRAITS(YCrCb4_to_rgba, 4, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_YCrCb2RGB_TRAITS(YCrCb_to_bgr, 3, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_YCrCb2RGB_TRAITS(YCrCb_to_bgra, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_YCrCb2RGB_TRAITS(YCrCb4_to_bgr, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_YCrCb2RGB_TRAITS(YCrCb4_to_bgra, 4, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_YCrCb2RGB_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2XYZ_TRAITS(rgb_to_xyz, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2XYZ_TRAITS(rgba_to_xyz, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2XYZ_TRAITS(rgb_to_xyz4, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2XYZ_TRAITS(rgba_to_xyz4, 4, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2XYZ_TRAITS(bgr_to_xyz, 3, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2XYZ_TRAITS(bgra_to_xyz, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2XYZ_TRAITS(bgr_to_xyz4, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2XYZ_TRAITS(bgra_to_xyz4, 4, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB2XYZ_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_XYZ2RGB_TRAITS(xyz_to_rgb, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_XYZ2RGB_TRAITS(xyz4_to_rgb, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_XYZ2RGB_TRAITS(xyz_to_rgba, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_XYZ2RGB_TRAITS(xyz4_to_rgba, 4, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_XYZ2RGB_TRAITS(xyz_to_bgr, 3, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_XYZ2RGB_TRAITS(xyz4_to_bgr, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_XYZ2RGB_TRAITS(xyz_to_bgra, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_XYZ2RGB_TRAITS(xyz4_to_bgra, 4, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_XYZ2RGB_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HSV_TRAITS(rgb_to_hsv, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HSV_TRAITS(rgba_to_hsv, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HSV_TRAITS(rgb_to_hsv4, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HSV_TRAITS(rgba_to_hsv4, 4, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HSV_TRAITS(bgr_to_hsv, 3, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HSV_TRAITS(bgra_to_hsv, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HSV_TRAITS(bgr_to_hsv4, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HSV_TRAITS(bgra_to_hsv4, 4, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB2HSV_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_HSV2RGB_TRAITS(hsv_to_rgb, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_HSV2RGB_TRAITS(hsv_to_rgba, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_HSV2RGB_TRAITS(hsv4_to_rgb, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_HSV2RGB_TRAITS(hsv4_to_rgba, 4, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_HSV2RGB_TRAITS(hsv_to_bgr, 3, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_HSV2RGB_TRAITS(hsv_to_bgra, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_HSV2RGB_TRAITS(hsv4_to_bgr, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_HSV2RGB_TRAITS(hsv4_to_bgra, 4, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_HSV2RGB_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HLS_TRAITS(rgb_to_hls, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HLS_TRAITS(rgba_to_hls, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HLS_TRAITS(rgb_to_hls4, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HLS_TRAITS(rgba_to_hls4, 4, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HLS_TRAITS(bgr_to_hls, 3, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HLS_TRAITS(bgra_to_hls, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HLS_TRAITS(bgr_to_hls4, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2HLS_TRAITS(bgra_to_hls4, 4, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB2HLS_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_HLS2RGB_TRAITS(hls_to_rgb, 3, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_HLS2RGB_TRAITS(hls_to_rgba, 3, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_HLS2RGB_TRAITS(hls4_to_rgb, 4, 3, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_HLS2RGB_TRAITS(hls4_to_rgba, 4, 4, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_HLS2RGB_TRAITS(hls_to_bgr, 3, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_HLS2RGB_TRAITS(hls_to_bgra, 3, 4, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_HLS2RGB_TRAITS(hls4_to_bgr, 4, 3, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_HLS2RGB_TRAITS(hls4_to_bgra, 4, 4, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_HLS2RGB_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(rgb_to_lab, 3, 3, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(rgba_to_lab, 4, 3, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(rgb_to_lab4, 3, 4, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(rgba_to_lab4, 4, 4, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(bgr_to_lab, 3, 3, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(bgra_to_lab, 4, 3, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(bgr_to_lab4, 3, 4, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(bgra_to_lab4, 4, 4, true, 0)
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(lrgb_to_lab, 3, 3, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(lrgba_to_lab, 4, 3, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(lrgb_to_lab4, 3, 4, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(lrgba_to_lab4, 4, 4, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(lbgr_to_lab, 3, 3, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(lbgra_to_lab, 4, 3, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(lbgr_to_lab4, 3, 4, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS(lbgra_to_lab4, 4, 4, false, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB2Lab_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab_to_rgb, 3, 3, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab4_to_rgb, 4, 3, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab_to_rgba, 3, 4, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab4_to_rgba, 4, 4, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab_to_bgr, 3, 3, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab4_to_bgr, 4, 3, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab_to_bgra, 3, 4, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab4_to_bgra, 4, 4, true, 0)
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab_to_lrgb, 3, 3, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab4_to_lrgb, 4, 3, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab_to_lrgba, 3, 4, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab4_to_lrgba, 4, 4, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab_to_lbgr, 3, 3, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab4_to_lbgr, 4, 3, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab_to_lbgra, 3, 4, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS(lab4_to_lbgra, 4, 4, false, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_Lab2RGB_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(rgb_to_luv, 3, 3, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(rgba_to_luv, 4, 3, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(rgb_to_luv4, 3, 4, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(rgba_to_luv4, 4, 4, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(bgr_to_luv, 3, 3, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(bgra_to_luv, 4, 3, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(bgr_to_luv4, 3, 4, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(bgra_to_luv4, 4, 4, true, 0)
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(lrgb_to_luv, 3, 3, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(lrgba_to_luv, 4, 3, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(lrgb_to_luv4, 3, 4, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(lrgba_to_luv4, 4, 4, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(lbgr_to_luv, 3, 3, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(lbgra_to_luv, 4, 3, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(lbgr_to_luv4, 3, 4, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS(lbgra_to_luv4, 4, 4, false, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_RGB2Luv_TRAITS
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv_to_rgb, 3, 3, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv4_to_rgb, 4, 3, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv_to_rgba, 3, 4, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv4_to_rgba, 4, 4, true, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv_to_bgr, 3, 3, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv4_to_bgr, 4, 3, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv_to_bgra, 3, 4, true, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv4_to_bgra, 4, 4, true, 0)
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv_to_lrgb, 3, 3, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv4_to_lrgb, 4, 3, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv_to_lrgba, 3, 4, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv4_to_lrgba, 4, 4, false, 2)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv_to_lbgr, 3, 3, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv4_to_lbgr, 4, 3, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv_to_lbgra, 3, 4, false, 0)
|
||||
OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS(luv4_to_lbgra, 4, 4, false, 0)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_Luv2RGB_TRAITS
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_COLOR_HPP
|
||||
|
|
@ -0,0 +1,109 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_COMMON_HPP
|
||||
#define OPENCV_CUDA_COMMON_HPP
|
||||
|
||||
#include <cuda_runtime.h>
|
||||
#include "opencv2/core/cuda_types.hpp"
|
||||
#include "opencv2/core/cvdef.h"
|
||||
#include "opencv2/core/base.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
#ifndef CV_PI_F
|
||||
#ifndef CV_PI
|
||||
#define CV_PI_F 3.14159265f
|
||||
#else
|
||||
#define CV_PI_F ((float)CV_PI)
|
||||
#endif
|
||||
#endif
|
||||
|
||||
namespace cv { namespace cuda {
|
||||
static inline void checkCudaError(cudaError_t err, const char* file, const int line, const char* func)
|
||||
{
|
||||
if (cudaSuccess != err)
|
||||
cv::error(cv::Error::GpuApiCallError, cudaGetErrorString(err), func, file, line);
|
||||
}
|
||||
}}
|
||||
|
||||
#ifndef cudaSafeCall
|
||||
#define cudaSafeCall(expr) cv::cuda::checkCudaError(expr, __FILE__, __LINE__, CV_Func)
|
||||
#endif
|
||||
|
||||
namespace cv { namespace cuda
|
||||
{
|
||||
template <typename T> static inline bool isAligned(const T* ptr, size_t size)
|
||||
{
|
||||
return reinterpret_cast<size_t>(ptr) % size == 0;
|
||||
}
|
||||
|
||||
static inline bool isAligned(size_t step, size_t size)
|
||||
{
|
||||
return step % size == 0;
|
||||
}
|
||||
}}
|
||||
|
||||
namespace cv { namespace cuda
|
||||
{
|
||||
namespace device
|
||||
{
|
||||
__host__ __device__ __forceinline__ int divUp(int total, int grain)
|
||||
{
|
||||
return (total + grain - 1) / grain;
|
||||
}
|
||||
|
||||
template<class T> inline void bindTexture(const textureReference* tex, const PtrStepSz<T>& img)
|
||||
{
|
||||
cudaChannelFormatDesc desc = cudaCreateChannelDesc<T>();
|
||||
cudaSafeCall( cudaBindTexture2D(0, tex, img.ptr(), &desc, img.cols, img.rows, img.step) );
|
||||
}
|
||||
}
|
||||
}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_COMMON_HPP
|
||||
|
|
@ -0,0 +1,113 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_DATAMOV_UTILS_HPP
|
||||
#define OPENCV_CUDA_DATAMOV_UTILS_HPP
|
||||
|
||||
#include "common.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 200
|
||||
|
||||
// for Fermi memory space is detected automatically
|
||||
template <typename T> struct ForceGlob
|
||||
{
|
||||
__device__ __forceinline__ static void Load(const T* ptr, int offset, T& val) { val = ptr[offset]; }
|
||||
};
|
||||
|
||||
#else // __CUDA_ARCH__ >= 200
|
||||
|
||||
#if defined(_WIN64) || defined(__LP64__)
|
||||
// 64-bit register modifier for inlined asm
|
||||
#define OPENCV_CUDA_ASM_PTR "l"
|
||||
#else
|
||||
// 32-bit register modifier for inlined asm
|
||||
#define OPENCV_CUDA_ASM_PTR "r"
|
||||
#endif
|
||||
|
||||
template<class T> struct ForceGlob;
|
||||
|
||||
#define OPENCV_CUDA_DEFINE_FORCE_GLOB(base_type, ptx_type, reg_mod) \
|
||||
template <> struct ForceGlob<base_type> \
|
||||
{ \
|
||||
__device__ __forceinline__ static void Load(const base_type* ptr, int offset, base_type& val) \
|
||||
{ \
|
||||
asm("ld.global."#ptx_type" %0, [%1];" : "="#reg_mod(val) : OPENCV_CUDA_ASM_PTR(ptr + offset)); \
|
||||
} \
|
||||
};
|
||||
|
||||
#define OPENCV_CUDA_DEFINE_FORCE_GLOB_B(base_type, ptx_type) \
|
||||
template <> struct ForceGlob<base_type> \
|
||||
{ \
|
||||
__device__ __forceinline__ static void Load(const base_type* ptr, int offset, base_type& val) \
|
||||
{ \
|
||||
asm("ld.global."#ptx_type" %0, [%1];" : "=r"(*reinterpret_cast<uint*>(&val)) : OPENCV_CUDA_ASM_PTR(ptr + offset)); \
|
||||
} \
|
||||
};
|
||||
|
||||
OPENCV_CUDA_DEFINE_FORCE_GLOB_B(uchar, u8)
|
||||
OPENCV_CUDA_DEFINE_FORCE_GLOB_B(schar, s8)
|
||||
OPENCV_CUDA_DEFINE_FORCE_GLOB_B(char, b8)
|
||||
OPENCV_CUDA_DEFINE_FORCE_GLOB (ushort, u16, h)
|
||||
OPENCV_CUDA_DEFINE_FORCE_GLOB (short, s16, h)
|
||||
OPENCV_CUDA_DEFINE_FORCE_GLOB (uint, u32, r)
|
||||
OPENCV_CUDA_DEFINE_FORCE_GLOB (int, s32, r)
|
||||
OPENCV_CUDA_DEFINE_FORCE_GLOB (float, f32, f)
|
||||
OPENCV_CUDA_DEFINE_FORCE_GLOB (double, f64, d)
|
||||
|
||||
#undef OPENCV_CUDA_DEFINE_FORCE_GLOB
|
||||
#undef OPENCV_CUDA_DEFINE_FORCE_GLOB_B
|
||||
#undef OPENCV_CUDA_ASM_PTR
|
||||
|
||||
#endif // __CUDA_ARCH__ >= 200
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_DATAMOV_UTILS_HPP
|
||||
File diff suppressed because one or more lines are too long
|
|
@ -0,0 +1,365 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_REDUCE_DETAIL_HPP
|
||||
#define OPENCV_CUDA_REDUCE_DETAIL_HPP
|
||||
|
||||
#include <thrust/tuple.h>
|
||||
#include "../warp.hpp"
|
||||
#include "../warp_shuffle.hpp"
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
namespace reduce_detail
|
||||
{
|
||||
template <typename T> struct GetType;
|
||||
template <typename T> struct GetType<T*>
|
||||
{
|
||||
typedef T type;
|
||||
};
|
||||
template <typename T> struct GetType<volatile T*>
|
||||
{
|
||||
typedef T type;
|
||||
};
|
||||
template <typename T> struct GetType<T&>
|
||||
{
|
||||
typedef T type;
|
||||
};
|
||||
|
||||
template <unsigned int I, unsigned int N>
|
||||
struct For
|
||||
{
|
||||
template <class PointerTuple, class ValTuple>
|
||||
static __device__ void loadToSmem(const PointerTuple& smem, const ValTuple& val, unsigned int tid)
|
||||
{
|
||||
thrust::get<I>(smem)[tid] = thrust::get<I>(val);
|
||||
|
||||
For<I + 1, N>::loadToSmem(smem, val, tid);
|
||||
}
|
||||
template <class PointerTuple, class ValTuple>
|
||||
static __device__ void loadFromSmem(const PointerTuple& smem, const ValTuple& val, unsigned int tid)
|
||||
{
|
||||
thrust::get<I>(val) = thrust::get<I>(smem)[tid];
|
||||
|
||||
For<I + 1, N>::loadFromSmem(smem, val, tid);
|
||||
}
|
||||
|
||||
template <class PointerTuple, class ValTuple, class OpTuple>
|
||||
static __device__ void merge(const PointerTuple& smem, const ValTuple& val, unsigned int tid, unsigned int delta, const OpTuple& op)
|
||||
{
|
||||
typename GetType<typename thrust::tuple_element<I, PointerTuple>::type>::type reg = thrust::get<I>(smem)[tid + delta];
|
||||
thrust::get<I>(smem)[tid] = thrust::get<I>(val) = thrust::get<I>(op)(thrust::get<I>(val), reg);
|
||||
|
||||
For<I + 1, N>::merge(smem, val, tid, delta, op);
|
||||
}
|
||||
template <class ValTuple, class OpTuple>
|
||||
static __device__ void mergeShfl(const ValTuple& val, unsigned int delta, unsigned int width, const OpTuple& op)
|
||||
{
|
||||
typename GetType<typename thrust::tuple_element<I, ValTuple>::type>::type reg = shfl_down(thrust::get<I>(val), delta, width);
|
||||
thrust::get<I>(val) = thrust::get<I>(op)(thrust::get<I>(val), reg);
|
||||
|
||||
For<I + 1, N>::mergeShfl(val, delta, width, op);
|
||||
}
|
||||
};
|
||||
template <unsigned int N>
|
||||
struct For<N, N>
|
||||
{
|
||||
template <class PointerTuple, class ValTuple>
|
||||
static __device__ void loadToSmem(const PointerTuple&, const ValTuple&, unsigned int)
|
||||
{
|
||||
}
|
||||
template <class PointerTuple, class ValTuple>
|
||||
static __device__ void loadFromSmem(const PointerTuple&, const ValTuple&, unsigned int)
|
||||
{
|
||||
}
|
||||
|
||||
template <class PointerTuple, class ValTuple, class OpTuple>
|
||||
static __device__ void merge(const PointerTuple&, const ValTuple&, unsigned int, unsigned int, const OpTuple&)
|
||||
{
|
||||
}
|
||||
template <class ValTuple, class OpTuple>
|
||||
static __device__ void mergeShfl(const ValTuple&, unsigned int, unsigned int, const OpTuple&)
|
||||
{
|
||||
}
|
||||
};
|
||||
|
||||
template <typename T>
|
||||
__device__ __forceinline__ void loadToSmem(volatile T* smem, T& val, unsigned int tid)
|
||||
{
|
||||
smem[tid] = val;
|
||||
}
|
||||
template <typename T>
|
||||
__device__ __forceinline__ void loadFromSmem(volatile T* smem, T& val, unsigned int tid)
|
||||
{
|
||||
val = smem[tid];
|
||||
}
|
||||
template <typename P0, typename P1, typename P2, typename P3, typename P4, typename P5, typename P6, typename P7, typename P8, typename P9,
|
||||
typename R0, typename R1, typename R2, typename R3, typename R4, typename R5, typename R6, typename R7, typename R8, typename R9>
|
||||
__device__ __forceinline__ void loadToSmem(const thrust::tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9>& smem,
|
||||
const thrust::tuple<R0, R1, R2, R3, R4, R5, R6, R7, R8, R9>& val,
|
||||
unsigned int tid)
|
||||
{
|
||||
For<0, thrust::tuple_size<thrust::tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9> >::value>::loadToSmem(smem, val, tid);
|
||||
}
|
||||
template <typename P0, typename P1, typename P2, typename P3, typename P4, typename P5, typename P6, typename P7, typename P8, typename P9,
|
||||
typename R0, typename R1, typename R2, typename R3, typename R4, typename R5, typename R6, typename R7, typename R8, typename R9>
|
||||
__device__ __forceinline__ void loadFromSmem(const thrust::tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9>& smem,
|
||||
const thrust::tuple<R0, R1, R2, R3, R4, R5, R6, R7, R8, R9>& val,
|
||||
unsigned int tid)
|
||||
{
|
||||
For<0, thrust::tuple_size<thrust::tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9> >::value>::loadFromSmem(smem, val, tid);
|
||||
}
|
||||
|
||||
template <typename T, class Op>
|
||||
__device__ __forceinline__ void merge(volatile T* smem, T& val, unsigned int tid, unsigned int delta, const Op& op)
|
||||
{
|
||||
T reg = smem[tid + delta];
|
||||
smem[tid] = val = op(val, reg);
|
||||
}
|
||||
template <typename T, class Op>
|
||||
__device__ __forceinline__ void mergeShfl(T& val, unsigned int delta, unsigned int width, const Op& op)
|
||||
{
|
||||
T reg = shfl_down(val, delta, width);
|
||||
val = op(val, reg);
|
||||
}
|
||||
template <typename P0, typename P1, typename P2, typename P3, typename P4, typename P5, typename P6, typename P7, typename P8, typename P9,
|
||||
typename R0, typename R1, typename R2, typename R3, typename R4, typename R5, typename R6, typename R7, typename R8, typename R9,
|
||||
class Op0, class Op1, class Op2, class Op3, class Op4, class Op5, class Op6, class Op7, class Op8, class Op9>
|
||||
__device__ __forceinline__ void merge(const thrust::tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9>& smem,
|
||||
const thrust::tuple<R0, R1, R2, R3, R4, R5, R6, R7, R8, R9>& val,
|
||||
unsigned int tid,
|
||||
unsigned int delta,
|
||||
const thrust::tuple<Op0, Op1, Op2, Op3, Op4, Op5, Op6, Op7, Op8, Op9>& op)
|
||||
{
|
||||
For<0, thrust::tuple_size<thrust::tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9> >::value>::merge(smem, val, tid, delta, op);
|
||||
}
|
||||
template <typename R0, typename R1, typename R2, typename R3, typename R4, typename R5, typename R6, typename R7, typename R8, typename R9,
|
||||
class Op0, class Op1, class Op2, class Op3, class Op4, class Op5, class Op6, class Op7, class Op8, class Op9>
|
||||
__device__ __forceinline__ void mergeShfl(const thrust::tuple<R0, R1, R2, R3, R4, R5, R6, R7, R8, R9>& val,
|
||||
unsigned int delta,
|
||||
unsigned int width,
|
||||
const thrust::tuple<Op0, Op1, Op2, Op3, Op4, Op5, Op6, Op7, Op8, Op9>& op)
|
||||
{
|
||||
For<0, thrust::tuple_size<thrust::tuple<R0, R1, R2, R3, R4, R5, R6, R7, R8, R9> >::value>::mergeShfl(val, delta, width, op);
|
||||
}
|
||||
|
||||
template <unsigned int N> struct Generic
|
||||
{
|
||||
template <typename Pointer, typename Reference, class Op>
|
||||
static __device__ void reduce(Pointer smem, Reference val, unsigned int tid, Op op)
|
||||
{
|
||||
loadToSmem(smem, val, tid);
|
||||
if (N >= 32)
|
||||
__syncthreads();
|
||||
|
||||
if (N >= 2048)
|
||||
{
|
||||
if (tid < 1024)
|
||||
merge(smem, val, tid, 1024, op);
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
if (N >= 1024)
|
||||
{
|
||||
if (tid < 512)
|
||||
merge(smem, val, tid, 512, op);
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
if (N >= 512)
|
||||
{
|
||||
if (tid < 256)
|
||||
merge(smem, val, tid, 256, op);
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
if (N >= 256)
|
||||
{
|
||||
if (tid < 128)
|
||||
merge(smem, val, tid, 128, op);
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
if (N >= 128)
|
||||
{
|
||||
if (tid < 64)
|
||||
merge(smem, val, tid, 64, op);
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
if (N >= 64)
|
||||
{
|
||||
if (tid < 32)
|
||||
merge(smem, val, tid, 32, op);
|
||||
}
|
||||
|
||||
if (tid < 16)
|
||||
{
|
||||
merge(smem, val, tid, 16, op);
|
||||
merge(smem, val, tid, 8, op);
|
||||
merge(smem, val, tid, 4, op);
|
||||
merge(smem, val, tid, 2, op);
|
||||
merge(smem, val, tid, 1, op);
|
||||
}
|
||||
}
|
||||
};
|
||||
|
||||
template <unsigned int I, typename Pointer, typename Reference, class Op>
|
||||
struct Unroll
|
||||
{
|
||||
static __device__ void loopShfl(Reference val, Op op, unsigned int N)
|
||||
{
|
||||
mergeShfl(val, I, N, op);
|
||||
Unroll<I / 2, Pointer, Reference, Op>::loopShfl(val, op, N);
|
||||
}
|
||||
static __device__ void loop(Pointer smem, Reference val, unsigned int tid, Op op)
|
||||
{
|
||||
merge(smem, val, tid, I, op);
|
||||
Unroll<I / 2, Pointer, Reference, Op>::loop(smem, val, tid, op);
|
||||
}
|
||||
};
|
||||
template <typename Pointer, typename Reference, class Op>
|
||||
struct Unroll<0, Pointer, Reference, Op>
|
||||
{
|
||||
static __device__ void loopShfl(Reference, Op, unsigned int)
|
||||
{
|
||||
}
|
||||
static __device__ void loop(Pointer, Reference, unsigned int, Op)
|
||||
{
|
||||
}
|
||||
};
|
||||
|
||||
template <unsigned int N> struct WarpOptimized
|
||||
{
|
||||
template <typename Pointer, typename Reference, class Op>
|
||||
static __device__ void reduce(Pointer smem, Reference val, unsigned int tid, Op op)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
CV_UNUSED(smem);
|
||||
CV_UNUSED(tid);
|
||||
|
||||
Unroll<N / 2, Pointer, Reference, Op>::loopShfl(val, op, N);
|
||||
#else
|
||||
loadToSmem(smem, val, tid);
|
||||
|
||||
if (tid < N / 2)
|
||||
Unroll<N / 2, Pointer, Reference, Op>::loop(smem, val, tid, op);
|
||||
#endif
|
||||
}
|
||||
};
|
||||
|
||||
template <unsigned int N> struct GenericOptimized32
|
||||
{
|
||||
enum { M = N / 32 };
|
||||
|
||||
template <typename Pointer, typename Reference, class Op>
|
||||
static __device__ void reduce(Pointer smem, Reference val, unsigned int tid, Op op)
|
||||
{
|
||||
const unsigned int laneId = Warp::laneId();
|
||||
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
Unroll<16, Pointer, Reference, Op>::loopShfl(val, op, warpSize);
|
||||
|
||||
if (laneId == 0)
|
||||
loadToSmem(smem, val, tid / 32);
|
||||
#else
|
||||
loadToSmem(smem, val, tid);
|
||||
|
||||
if (laneId < 16)
|
||||
Unroll<16, Pointer, Reference, Op>::loop(smem, val, tid, op);
|
||||
|
||||
__syncthreads();
|
||||
|
||||
if (laneId == 0)
|
||||
loadToSmem(smem, val, tid / 32);
|
||||
#endif
|
||||
|
||||
__syncthreads();
|
||||
|
||||
loadFromSmem(smem, val, tid);
|
||||
|
||||
if (tid < 32)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
Unroll<M / 2, Pointer, Reference, Op>::loopShfl(val, op, M);
|
||||
#else
|
||||
Unroll<M / 2, Pointer, Reference, Op>::loop(smem, val, tid, op);
|
||||
#endif
|
||||
}
|
||||
}
|
||||
};
|
||||
|
||||
template <bool val, class T1, class T2> struct StaticIf;
|
||||
template <class T1, class T2> struct StaticIf<true, T1, T2>
|
||||
{
|
||||
typedef T1 type;
|
||||
};
|
||||
template <class T1, class T2> struct StaticIf<false, T1, T2>
|
||||
{
|
||||
typedef T2 type;
|
||||
};
|
||||
|
||||
template <unsigned int N> struct IsPowerOf2
|
||||
{
|
||||
enum { value = ((N != 0) && !(N & (N - 1))) };
|
||||
};
|
||||
|
||||
template <unsigned int N> struct Dispatcher
|
||||
{
|
||||
typedef typename StaticIf<
|
||||
(N <= 32) && IsPowerOf2<N>::value,
|
||||
WarpOptimized<N>,
|
||||
typename StaticIf<
|
||||
(N <= 1024) && IsPowerOf2<N>::value,
|
||||
GenericOptimized32<N>,
|
||||
Generic<N>
|
||||
>::type
|
||||
>::type reductor;
|
||||
};
|
||||
}
|
||||
}}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_REDUCE_DETAIL_HPP
|
||||
|
|
@ -0,0 +1,502 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_PRED_VAL_REDUCE_DETAIL_HPP
|
||||
#define OPENCV_CUDA_PRED_VAL_REDUCE_DETAIL_HPP
|
||||
|
||||
#include <thrust/tuple.h>
|
||||
#include "../warp.hpp"
|
||||
#include "../warp_shuffle.hpp"
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
namespace reduce_key_val_detail
|
||||
{
|
||||
template <typename T> struct GetType;
|
||||
template <typename T> struct GetType<T*>
|
||||
{
|
||||
typedef T type;
|
||||
};
|
||||
template <typename T> struct GetType<volatile T*>
|
||||
{
|
||||
typedef T type;
|
||||
};
|
||||
template <typename T> struct GetType<T&>
|
||||
{
|
||||
typedef T type;
|
||||
};
|
||||
|
||||
template <unsigned int I, unsigned int N>
|
||||
struct For
|
||||
{
|
||||
template <class PointerTuple, class ReferenceTuple>
|
||||
static __device__ void loadToSmem(const PointerTuple& smem, const ReferenceTuple& data, unsigned int tid)
|
||||
{
|
||||
thrust::get<I>(smem)[tid] = thrust::get<I>(data);
|
||||
|
||||
For<I + 1, N>::loadToSmem(smem, data, tid);
|
||||
}
|
||||
template <class PointerTuple, class ReferenceTuple>
|
||||
static __device__ void loadFromSmem(const PointerTuple& smem, const ReferenceTuple& data, unsigned int tid)
|
||||
{
|
||||
thrust::get<I>(data) = thrust::get<I>(smem)[tid];
|
||||
|
||||
For<I + 1, N>::loadFromSmem(smem, data, tid);
|
||||
}
|
||||
|
||||
template <class ReferenceTuple>
|
||||
static __device__ void copyShfl(const ReferenceTuple& val, unsigned int delta, int width)
|
||||
{
|
||||
thrust::get<I>(val) = shfl_down(thrust::get<I>(val), delta, width);
|
||||
|
||||
For<I + 1, N>::copyShfl(val, delta, width);
|
||||
}
|
||||
template <class PointerTuple, class ReferenceTuple>
|
||||
static __device__ void copy(const PointerTuple& svals, const ReferenceTuple& val, unsigned int tid, unsigned int delta)
|
||||
{
|
||||
thrust::get<I>(svals)[tid] = thrust::get<I>(val) = thrust::get<I>(svals)[tid + delta];
|
||||
|
||||
For<I + 1, N>::copy(svals, val, tid, delta);
|
||||
}
|
||||
|
||||
template <class KeyReferenceTuple, class ValReferenceTuple, class CmpTuple>
|
||||
static __device__ void mergeShfl(const KeyReferenceTuple& key, const ValReferenceTuple& val, const CmpTuple& cmp, unsigned int delta, int width)
|
||||
{
|
||||
typename GetType<typename thrust::tuple_element<I, KeyReferenceTuple>::type>::type reg = shfl_down(thrust::get<I>(key), delta, width);
|
||||
|
||||
if (thrust::get<I>(cmp)(reg, thrust::get<I>(key)))
|
||||
{
|
||||
thrust::get<I>(key) = reg;
|
||||
thrust::get<I>(val) = shfl_down(thrust::get<I>(val), delta, width);
|
||||
}
|
||||
|
||||
For<I + 1, N>::mergeShfl(key, val, cmp, delta, width);
|
||||
}
|
||||
template <class KeyPointerTuple, class KeyReferenceTuple, class ValPointerTuple, class ValReferenceTuple, class CmpTuple>
|
||||
static __device__ void merge(const KeyPointerTuple& skeys, const KeyReferenceTuple& key,
|
||||
const ValPointerTuple& svals, const ValReferenceTuple& val,
|
||||
const CmpTuple& cmp,
|
||||
unsigned int tid, unsigned int delta)
|
||||
{
|
||||
typename GetType<typename thrust::tuple_element<I, KeyPointerTuple>::type>::type reg = thrust::get<I>(skeys)[tid + delta];
|
||||
|
||||
if (thrust::get<I>(cmp)(reg, thrust::get<I>(key)))
|
||||
{
|
||||
thrust::get<I>(skeys)[tid] = thrust::get<I>(key) = reg;
|
||||
thrust::get<I>(svals)[tid] = thrust::get<I>(val) = thrust::get<I>(svals)[tid + delta];
|
||||
}
|
||||
|
||||
For<I + 1, N>::merge(skeys, key, svals, val, cmp, tid, delta);
|
||||
}
|
||||
};
|
||||
template <unsigned int N>
|
||||
struct For<N, N>
|
||||
{
|
||||
template <class PointerTuple, class ReferenceTuple>
|
||||
static __device__ void loadToSmem(const PointerTuple&, const ReferenceTuple&, unsigned int)
|
||||
{
|
||||
}
|
||||
template <class PointerTuple, class ReferenceTuple>
|
||||
static __device__ void loadFromSmem(const PointerTuple&, const ReferenceTuple&, unsigned int)
|
||||
{
|
||||
}
|
||||
|
||||
template <class ReferenceTuple>
|
||||
static __device__ void copyShfl(const ReferenceTuple&, unsigned int, int)
|
||||
{
|
||||
}
|
||||
template <class PointerTuple, class ReferenceTuple>
|
||||
static __device__ void copy(const PointerTuple&, const ReferenceTuple&, unsigned int, unsigned int)
|
||||
{
|
||||
}
|
||||
|
||||
template <class KeyReferenceTuple, class ValReferenceTuple, class CmpTuple>
|
||||
static __device__ void mergeShfl(const KeyReferenceTuple&, const ValReferenceTuple&, const CmpTuple&, unsigned int, int)
|
||||
{
|
||||
}
|
||||
template <class KeyPointerTuple, class KeyReferenceTuple, class ValPointerTuple, class ValReferenceTuple, class CmpTuple>
|
||||
static __device__ void merge(const KeyPointerTuple&, const KeyReferenceTuple&,
|
||||
const ValPointerTuple&, const ValReferenceTuple&,
|
||||
const CmpTuple&,
|
||||
unsigned int, unsigned int)
|
||||
{
|
||||
}
|
||||
};
|
||||
|
||||
//////////////////////////////////////////////////////
|
||||
// loadToSmem
|
||||
|
||||
template <typename T>
|
||||
__device__ __forceinline__ void loadToSmem(volatile T* smem, T& data, unsigned int tid)
|
||||
{
|
||||
smem[tid] = data;
|
||||
}
|
||||
template <typename T>
|
||||
__device__ __forceinline__ void loadFromSmem(volatile T* smem, T& data, unsigned int tid)
|
||||
{
|
||||
data = smem[tid];
|
||||
}
|
||||
template <typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
||||
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9>
|
||||
__device__ __forceinline__ void loadToSmem(const thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& smem,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& data,
|
||||
unsigned int tid)
|
||||
{
|
||||
For<0, thrust::tuple_size<thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9> >::value>::loadToSmem(smem, data, tid);
|
||||
}
|
||||
template <typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
||||
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9>
|
||||
__device__ __forceinline__ void loadFromSmem(const thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& smem,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& data,
|
||||
unsigned int tid)
|
||||
{
|
||||
For<0, thrust::tuple_size<thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9> >::value>::loadFromSmem(smem, data, tid);
|
||||
}
|
||||
|
||||
//////////////////////////////////////////////////////
|
||||
// copyVals
|
||||
|
||||
template <typename V>
|
||||
__device__ __forceinline__ void copyValsShfl(V& val, unsigned int delta, int width)
|
||||
{
|
||||
val = shfl_down(val, delta, width);
|
||||
}
|
||||
template <typename V>
|
||||
__device__ __forceinline__ void copyVals(volatile V* svals, V& val, unsigned int tid, unsigned int delta)
|
||||
{
|
||||
svals[tid] = val = svals[tid + delta];
|
||||
}
|
||||
template <typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9>
|
||||
__device__ __forceinline__ void copyValsShfl(const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
||||
unsigned int delta,
|
||||
int width)
|
||||
{
|
||||
For<0, thrust::tuple_size<thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9> >::value>::copyShfl(val, delta, width);
|
||||
}
|
||||
template <typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
||||
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9>
|
||||
__device__ __forceinline__ void copyVals(const thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& svals,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
||||
unsigned int tid, unsigned int delta)
|
||||
{
|
||||
For<0, thrust::tuple_size<thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9> >::value>::copy(svals, val, tid, delta);
|
||||
}
|
||||
|
||||
//////////////////////////////////////////////////////
|
||||
// merge
|
||||
|
||||
template <typename K, typename V, class Cmp>
|
||||
__device__ __forceinline__ void mergeShfl(K& key, V& val, const Cmp& cmp, unsigned int delta, int width)
|
||||
{
|
||||
K reg = shfl_down(key, delta, width);
|
||||
|
||||
if (cmp(reg, key))
|
||||
{
|
||||
key = reg;
|
||||
copyValsShfl(val, delta, width);
|
||||
}
|
||||
}
|
||||
template <typename K, typename V, class Cmp>
|
||||
__device__ __forceinline__ void merge(volatile K* skeys, K& key, volatile V* svals, V& val, const Cmp& cmp, unsigned int tid, unsigned int delta)
|
||||
{
|
||||
K reg = skeys[tid + delta];
|
||||
|
||||
if (cmp(reg, key))
|
||||
{
|
||||
skeys[tid] = key = reg;
|
||||
copyVals(svals, val, tid, delta);
|
||||
}
|
||||
}
|
||||
template <typename K,
|
||||
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9,
|
||||
class Cmp>
|
||||
__device__ __forceinline__ void mergeShfl(K& key,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
||||
const Cmp& cmp,
|
||||
unsigned int delta, int width)
|
||||
{
|
||||
K reg = shfl_down(key, delta, width);
|
||||
|
||||
if (cmp(reg, key))
|
||||
{
|
||||
key = reg;
|
||||
copyValsShfl(val, delta, width);
|
||||
}
|
||||
}
|
||||
template <typename K,
|
||||
typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
||||
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9,
|
||||
class Cmp>
|
||||
__device__ __forceinline__ void merge(volatile K* skeys, K& key,
|
||||
const thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& svals,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
||||
const Cmp& cmp, unsigned int tid, unsigned int delta)
|
||||
{
|
||||
K reg = skeys[tid + delta];
|
||||
|
||||
if (cmp(reg, key))
|
||||
{
|
||||
skeys[tid] = key = reg;
|
||||
copyVals(svals, val, tid, delta);
|
||||
}
|
||||
}
|
||||
template <typename KR0, typename KR1, typename KR2, typename KR3, typename KR4, typename KR5, typename KR6, typename KR7, typename KR8, typename KR9,
|
||||
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9,
|
||||
class Cmp0, class Cmp1, class Cmp2, class Cmp3, class Cmp4, class Cmp5, class Cmp6, class Cmp7, class Cmp8, class Cmp9>
|
||||
__device__ __forceinline__ void mergeShfl(const thrust::tuple<KR0, KR1, KR2, KR3, KR4, KR5, KR6, KR7, KR8, KR9>& key,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
||||
const thrust::tuple<Cmp0, Cmp1, Cmp2, Cmp3, Cmp4, Cmp5, Cmp6, Cmp7, Cmp8, Cmp9>& cmp,
|
||||
unsigned int delta, int width)
|
||||
{
|
||||
For<0, thrust::tuple_size<thrust::tuple<KR0, KR1, KR2, KR3, KR4, KR5, KR6, KR7, KR8, KR9> >::value>::mergeShfl(key, val, cmp, delta, width);
|
||||
}
|
||||
template <typename KP0, typename KP1, typename KP2, typename KP3, typename KP4, typename KP5, typename KP6, typename KP7, typename KP8, typename KP9,
|
||||
typename KR0, typename KR1, typename KR2, typename KR3, typename KR4, typename KR5, typename KR6, typename KR7, typename KR8, typename KR9,
|
||||
typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
||||
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9,
|
||||
class Cmp0, class Cmp1, class Cmp2, class Cmp3, class Cmp4, class Cmp5, class Cmp6, class Cmp7, class Cmp8, class Cmp9>
|
||||
__device__ __forceinline__ void merge(const thrust::tuple<KP0, KP1, KP2, KP3, KP4, KP5, KP6, KP7, KP8, KP9>& skeys,
|
||||
const thrust::tuple<KR0, KR1, KR2, KR3, KR4, KR5, KR6, KR7, KR8, KR9>& key,
|
||||
const thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& svals,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
||||
const thrust::tuple<Cmp0, Cmp1, Cmp2, Cmp3, Cmp4, Cmp5, Cmp6, Cmp7, Cmp8, Cmp9>& cmp,
|
||||
unsigned int tid, unsigned int delta)
|
||||
{
|
||||
For<0, thrust::tuple_size<thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9> >::value>::merge(skeys, key, svals, val, cmp, tid, delta);
|
||||
}
|
||||
|
||||
//////////////////////////////////////////////////////
|
||||
// Generic
|
||||
|
||||
template <unsigned int N> struct Generic
|
||||
{
|
||||
template <class KP, class KR, class VP, class VR, class Cmp>
|
||||
static __device__ void reduce(KP skeys, KR key, VP svals, VR val, unsigned int tid, Cmp cmp)
|
||||
{
|
||||
loadToSmem(skeys, key, tid);
|
||||
loadValsToSmem(svals, val, tid);
|
||||
if (N >= 32)
|
||||
__syncthreads();
|
||||
|
||||
if (N >= 2048)
|
||||
{
|
||||
if (tid < 1024)
|
||||
merge(skeys, key, svals, val, cmp, tid, 1024);
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
if (N >= 1024)
|
||||
{
|
||||
if (tid < 512)
|
||||
merge(skeys, key, svals, val, cmp, tid, 512);
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
if (N >= 512)
|
||||
{
|
||||
if (tid < 256)
|
||||
merge(skeys, key, svals, val, cmp, tid, 256);
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
if (N >= 256)
|
||||
{
|
||||
if (tid < 128)
|
||||
merge(skeys, key, svals, val, cmp, tid, 128);
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
if (N >= 128)
|
||||
{
|
||||
if (tid < 64)
|
||||
merge(skeys, key, svals, val, cmp, tid, 64);
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
if (N >= 64)
|
||||
{
|
||||
if (tid < 32)
|
||||
merge(skeys, key, svals, val, cmp, tid, 32);
|
||||
}
|
||||
|
||||
if (tid < 16)
|
||||
{
|
||||
merge(skeys, key, svals, val, cmp, tid, 16);
|
||||
merge(skeys, key, svals, val, cmp, tid, 8);
|
||||
merge(skeys, key, svals, val, cmp, tid, 4);
|
||||
merge(skeys, key, svals, val, cmp, tid, 2);
|
||||
merge(skeys, key, svals, val, cmp, tid, 1);
|
||||
}
|
||||
}
|
||||
};
|
||||
|
||||
template <unsigned int I, class KP, class KR, class VP, class VR, class Cmp>
|
||||
struct Unroll
|
||||
{
|
||||
static __device__ void loopShfl(KR key, VR val, Cmp cmp, unsigned int N)
|
||||
{
|
||||
mergeShfl(key, val, cmp, I, N);
|
||||
Unroll<I / 2, KP, KR, VP, VR, Cmp>::loopShfl(key, val, cmp, N);
|
||||
}
|
||||
static __device__ void loop(KP skeys, KR key, VP svals, VR val, unsigned int tid, Cmp cmp)
|
||||
{
|
||||
merge(skeys, key, svals, val, cmp, tid, I);
|
||||
Unroll<I / 2, KP, KR, VP, VR, Cmp>::loop(skeys, key, svals, val, tid, cmp);
|
||||
}
|
||||
};
|
||||
template <class KP, class KR, class VP, class VR, class Cmp>
|
||||
struct Unroll<0, KP, KR, VP, VR, Cmp>
|
||||
{
|
||||
static __device__ void loopShfl(KR, VR, Cmp, unsigned int)
|
||||
{
|
||||
}
|
||||
static __device__ void loop(KP, KR, VP, VR, unsigned int, Cmp)
|
||||
{
|
||||
}
|
||||
};
|
||||
|
||||
template <unsigned int N> struct WarpOptimized
|
||||
{
|
||||
template <class KP, class KR, class VP, class VR, class Cmp>
|
||||
static __device__ void reduce(KP skeys, KR key, VP svals, VR val, unsigned int tid, Cmp cmp)
|
||||
{
|
||||
#if 0 // __CUDA_ARCH__ >= 300
|
||||
CV_UNUSED(skeys);
|
||||
CV_UNUSED(svals);
|
||||
CV_UNUSED(tid);
|
||||
|
||||
Unroll<N / 2, KP, KR, VP, VR, Cmp>::loopShfl(key, val, cmp, N);
|
||||
#else
|
||||
loadToSmem(skeys, key, tid);
|
||||
loadToSmem(svals, val, tid);
|
||||
|
||||
if (tid < N / 2)
|
||||
Unroll<N / 2, KP, KR, VP, VR, Cmp>::loop(skeys, key, svals, val, tid, cmp);
|
||||
#endif
|
||||
}
|
||||
};
|
||||
|
||||
template <unsigned int N> struct GenericOptimized32
|
||||
{
|
||||
enum { M = N / 32 };
|
||||
|
||||
template <class KP, class KR, class VP, class VR, class Cmp>
|
||||
static __device__ void reduce(KP skeys, KR key, VP svals, VR val, unsigned int tid, Cmp cmp)
|
||||
{
|
||||
const unsigned int laneId = Warp::laneId();
|
||||
|
||||
#if 0 // __CUDA_ARCH__ >= 300
|
||||
Unroll<16, KP, KR, VP, VR, Cmp>::loopShfl(key, val, cmp, warpSize);
|
||||
|
||||
if (laneId == 0)
|
||||
{
|
||||
loadToSmem(skeys, key, tid / 32);
|
||||
loadToSmem(svals, val, tid / 32);
|
||||
}
|
||||
#else
|
||||
loadToSmem(skeys, key, tid);
|
||||
loadToSmem(svals, val, tid);
|
||||
|
||||
if (laneId < 16)
|
||||
Unroll<16, KP, KR, VP, VR, Cmp>::loop(skeys, key, svals, val, tid, cmp);
|
||||
|
||||
__syncthreads();
|
||||
|
||||
if (laneId == 0)
|
||||
{
|
||||
loadToSmem(skeys, key, tid / 32);
|
||||
loadToSmem(svals, val, tid / 32);
|
||||
}
|
||||
#endif
|
||||
|
||||
__syncthreads();
|
||||
|
||||
loadFromSmem(skeys, key, tid);
|
||||
|
||||
if (tid < 32)
|
||||
{
|
||||
#if 0 // __CUDA_ARCH__ >= 300
|
||||
loadFromSmem(svals, val, tid);
|
||||
|
||||
Unroll<M / 2, KP, KR, VP, VR, Cmp>::loopShfl(key, val, cmp, M);
|
||||
#else
|
||||
Unroll<M / 2, KP, KR, VP, VR, Cmp>::loop(skeys, key, svals, val, tid, cmp);
|
||||
#endif
|
||||
}
|
||||
}
|
||||
};
|
||||
|
||||
template <bool val, class T1, class T2> struct StaticIf;
|
||||
template <class T1, class T2> struct StaticIf<true, T1, T2>
|
||||
{
|
||||
typedef T1 type;
|
||||
};
|
||||
template <class T1, class T2> struct StaticIf<false, T1, T2>
|
||||
{
|
||||
typedef T2 type;
|
||||
};
|
||||
|
||||
template <unsigned int N> struct IsPowerOf2
|
||||
{
|
||||
enum { value = ((N != 0) && !(N & (N - 1))) };
|
||||
};
|
||||
|
||||
template <unsigned int N> struct Dispatcher
|
||||
{
|
||||
typedef typename StaticIf<
|
||||
(N <= 32) && IsPowerOf2<N>::value,
|
||||
WarpOptimized<N>,
|
||||
typename StaticIf<
|
||||
(N <= 1024) && IsPowerOf2<N>::value,
|
||||
GenericOptimized32<N>,
|
||||
Generic<N>
|
||||
>::type
|
||||
>::type reductor;
|
||||
};
|
||||
}
|
||||
}}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_PRED_VAL_REDUCE_DETAIL_HPP
|
||||
|
|
@ -0,0 +1,392 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_TRANSFORM_DETAIL_HPP
|
||||
#define OPENCV_CUDA_TRANSFORM_DETAIL_HPP
|
||||
|
||||
#include "../common.hpp"
|
||||
#include "../vec_traits.hpp"
|
||||
#include "../functional.hpp"
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
namespace transform_detail
|
||||
{
|
||||
//! Read Write Traits
|
||||
|
||||
template <typename T, typename D, int shift> struct UnaryReadWriteTraits
|
||||
{
|
||||
typedef typename TypeVec<T, shift>::vec_type read_type;
|
||||
typedef typename TypeVec<D, shift>::vec_type write_type;
|
||||
};
|
||||
|
||||
template <typename T1, typename T2, typename D, int shift> struct BinaryReadWriteTraits
|
||||
{
|
||||
typedef typename TypeVec<T1, shift>::vec_type read_type1;
|
||||
typedef typename TypeVec<T2, shift>::vec_type read_type2;
|
||||
typedef typename TypeVec<D, shift>::vec_type write_type;
|
||||
};
|
||||
|
||||
//! Transform kernels
|
||||
|
||||
template <int shift> struct OpUnroller;
|
||||
template <> struct OpUnroller<1>
|
||||
{
|
||||
template <typename T, typename D, typename UnOp, typename Mask>
|
||||
static __device__ __forceinline__ void unroll(const T& src, D& dst, const Mask& mask, UnOp& op, int x_shifted, int y)
|
||||
{
|
||||
if (mask(y, x_shifted))
|
||||
dst.x = op(src.x);
|
||||
}
|
||||
|
||||
template <typename T1, typename T2, typename D, typename BinOp, typename Mask>
|
||||
static __device__ __forceinline__ void unroll(const T1& src1, const T2& src2, D& dst, const Mask& mask, BinOp& op, int x_shifted, int y)
|
||||
{
|
||||
if (mask(y, x_shifted))
|
||||
dst.x = op(src1.x, src2.x);
|
||||
}
|
||||
};
|
||||
template <> struct OpUnroller<2>
|
||||
{
|
||||
template <typename T, typename D, typename UnOp, typename Mask>
|
||||
static __device__ __forceinline__ void unroll(const T& src, D& dst, const Mask& mask, UnOp& op, int x_shifted, int y)
|
||||
{
|
||||
if (mask(y, x_shifted))
|
||||
dst.x = op(src.x);
|
||||
if (mask(y, x_shifted + 1))
|
||||
dst.y = op(src.y);
|
||||
}
|
||||
|
||||
template <typename T1, typename T2, typename D, typename BinOp, typename Mask>
|
||||
static __device__ __forceinline__ void unroll(const T1& src1, const T2& src2, D& dst, const Mask& mask, BinOp& op, int x_shifted, int y)
|
||||
{
|
||||
if (mask(y, x_shifted))
|
||||
dst.x = op(src1.x, src2.x);
|
||||
if (mask(y, x_shifted + 1))
|
||||
dst.y = op(src1.y, src2.y);
|
||||
}
|
||||
};
|
||||
template <> struct OpUnroller<3>
|
||||
{
|
||||
template <typename T, typename D, typename UnOp, typename Mask>
|
||||
static __device__ __forceinline__ void unroll(const T& src, D& dst, const Mask& mask, const UnOp& op, int x_shifted, int y)
|
||||
{
|
||||
if (mask(y, x_shifted))
|
||||
dst.x = op(src.x);
|
||||
if (mask(y, x_shifted + 1))
|
||||
dst.y = op(src.y);
|
||||
if (mask(y, x_shifted + 2))
|
||||
dst.z = op(src.z);
|
||||
}
|
||||
|
||||
template <typename T1, typename T2, typename D, typename BinOp, typename Mask>
|
||||
static __device__ __forceinline__ void unroll(const T1& src1, const T2& src2, D& dst, const Mask& mask, const BinOp& op, int x_shifted, int y)
|
||||
{
|
||||
if (mask(y, x_shifted))
|
||||
dst.x = op(src1.x, src2.x);
|
||||
if (mask(y, x_shifted + 1))
|
||||
dst.y = op(src1.y, src2.y);
|
||||
if (mask(y, x_shifted + 2))
|
||||
dst.z = op(src1.z, src2.z);
|
||||
}
|
||||
};
|
||||
template <> struct OpUnroller<4>
|
||||
{
|
||||
template <typename T, typename D, typename UnOp, typename Mask>
|
||||
static __device__ __forceinline__ void unroll(const T& src, D& dst, const Mask& mask, const UnOp& op, int x_shifted, int y)
|
||||
{
|
||||
if (mask(y, x_shifted))
|
||||
dst.x = op(src.x);
|
||||
if (mask(y, x_shifted + 1))
|
||||
dst.y = op(src.y);
|
||||
if (mask(y, x_shifted + 2))
|
||||
dst.z = op(src.z);
|
||||
if (mask(y, x_shifted + 3))
|
||||
dst.w = op(src.w);
|
||||
}
|
||||
|
||||
template <typename T1, typename T2, typename D, typename BinOp, typename Mask>
|
||||
static __device__ __forceinline__ void unroll(const T1& src1, const T2& src2, D& dst, const Mask& mask, const BinOp& op, int x_shifted, int y)
|
||||
{
|
||||
if (mask(y, x_shifted))
|
||||
dst.x = op(src1.x, src2.x);
|
||||
if (mask(y, x_shifted + 1))
|
||||
dst.y = op(src1.y, src2.y);
|
||||
if (mask(y, x_shifted + 2))
|
||||
dst.z = op(src1.z, src2.z);
|
||||
if (mask(y, x_shifted + 3))
|
||||
dst.w = op(src1.w, src2.w);
|
||||
}
|
||||
};
|
||||
template <> struct OpUnroller<8>
|
||||
{
|
||||
template <typename T, typename D, typename UnOp, typename Mask>
|
||||
static __device__ __forceinline__ void unroll(const T& src, D& dst, const Mask& mask, const UnOp& op, int x_shifted, int y)
|
||||
{
|
||||
if (mask(y, x_shifted))
|
||||
dst.a0 = op(src.a0);
|
||||
if (mask(y, x_shifted + 1))
|
||||
dst.a1 = op(src.a1);
|
||||
if (mask(y, x_shifted + 2))
|
||||
dst.a2 = op(src.a2);
|
||||
if (mask(y, x_shifted + 3))
|
||||
dst.a3 = op(src.a3);
|
||||
if (mask(y, x_shifted + 4))
|
||||
dst.a4 = op(src.a4);
|
||||
if (mask(y, x_shifted + 5))
|
||||
dst.a5 = op(src.a5);
|
||||
if (mask(y, x_shifted + 6))
|
||||
dst.a6 = op(src.a6);
|
||||
if (mask(y, x_shifted + 7))
|
||||
dst.a7 = op(src.a7);
|
||||
}
|
||||
|
||||
template <typename T1, typename T2, typename D, typename BinOp, typename Mask>
|
||||
static __device__ __forceinline__ void unroll(const T1& src1, const T2& src2, D& dst, const Mask& mask, const BinOp& op, int x_shifted, int y)
|
||||
{
|
||||
if (mask(y, x_shifted))
|
||||
dst.a0 = op(src1.a0, src2.a0);
|
||||
if (mask(y, x_shifted + 1))
|
||||
dst.a1 = op(src1.a1, src2.a1);
|
||||
if (mask(y, x_shifted + 2))
|
||||
dst.a2 = op(src1.a2, src2.a2);
|
||||
if (mask(y, x_shifted + 3))
|
||||
dst.a3 = op(src1.a3, src2.a3);
|
||||
if (mask(y, x_shifted + 4))
|
||||
dst.a4 = op(src1.a4, src2.a4);
|
||||
if (mask(y, x_shifted + 5))
|
||||
dst.a5 = op(src1.a5, src2.a5);
|
||||
if (mask(y, x_shifted + 6))
|
||||
dst.a6 = op(src1.a6, src2.a6);
|
||||
if (mask(y, x_shifted + 7))
|
||||
dst.a7 = op(src1.a7, src2.a7);
|
||||
}
|
||||
};
|
||||
|
||||
template <typename T, typename D, typename UnOp, typename Mask>
|
||||
static __global__ void transformSmart(const PtrStepSz<T> src_, PtrStep<D> dst_, const Mask mask, const UnOp op)
|
||||
{
|
||||
typedef TransformFunctorTraits<UnOp> ft;
|
||||
typedef typename UnaryReadWriteTraits<T, D, ft::smart_shift>::read_type read_type;
|
||||
typedef typename UnaryReadWriteTraits<T, D, ft::smart_shift>::write_type write_type;
|
||||
|
||||
const int x = threadIdx.x + blockIdx.x * blockDim.x;
|
||||
const int y = threadIdx.y + blockIdx.y * blockDim.y;
|
||||
const int x_shifted = x * ft::smart_shift;
|
||||
|
||||
if (y < src_.rows)
|
||||
{
|
||||
const T* src = src_.ptr(y);
|
||||
D* dst = dst_.ptr(y);
|
||||
|
||||
if (x_shifted + ft::smart_shift - 1 < src_.cols)
|
||||
{
|
||||
const read_type src_n_el = ((const read_type*)src)[x];
|
||||
OpUnroller<ft::smart_shift>::unroll(src_n_el, ((write_type*)dst)[x], mask, op, x_shifted, y);
|
||||
}
|
||||
else
|
||||
{
|
||||
for (int real_x = x_shifted; real_x < src_.cols; ++real_x)
|
||||
{
|
||||
if (mask(y, real_x))
|
||||
dst[real_x] = op(src[real_x]);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
template <typename T, typename D, typename UnOp, typename Mask>
|
||||
__global__ static void transformSimple(const PtrStepSz<T> src, PtrStep<D> dst, const Mask mask, const UnOp op)
|
||||
{
|
||||
const int x = blockDim.x * blockIdx.x + threadIdx.x;
|
||||
const int y = blockDim.y * blockIdx.y + threadIdx.y;
|
||||
|
||||
if (x < src.cols && y < src.rows && mask(y, x))
|
||||
{
|
||||
dst.ptr(y)[x] = op(src.ptr(y)[x]);
|
||||
}
|
||||
}
|
||||
|
||||
template <typename T1, typename T2, typename D, typename BinOp, typename Mask>
|
||||
static __global__ void transformSmart(const PtrStepSz<T1> src1_, const PtrStep<T2> src2_, PtrStep<D> dst_,
|
||||
const Mask mask, const BinOp op)
|
||||
{
|
||||
typedef TransformFunctorTraits<BinOp> ft;
|
||||
typedef typename BinaryReadWriteTraits<T1, T2, D, ft::smart_shift>::read_type1 read_type1;
|
||||
typedef typename BinaryReadWriteTraits<T1, T2, D, ft::smart_shift>::read_type2 read_type2;
|
||||
typedef typename BinaryReadWriteTraits<T1, T2, D, ft::smart_shift>::write_type write_type;
|
||||
|
||||
const int x = threadIdx.x + blockIdx.x * blockDim.x;
|
||||
const int y = threadIdx.y + blockIdx.y * blockDim.y;
|
||||
const int x_shifted = x * ft::smart_shift;
|
||||
|
||||
if (y < src1_.rows)
|
||||
{
|
||||
const T1* src1 = src1_.ptr(y);
|
||||
const T2* src2 = src2_.ptr(y);
|
||||
D* dst = dst_.ptr(y);
|
||||
|
||||
if (x_shifted + ft::smart_shift - 1 < src1_.cols)
|
||||
{
|
||||
const read_type1 src1_n_el = ((const read_type1*)src1)[x];
|
||||
const read_type2 src2_n_el = ((const read_type2*)src2)[x];
|
||||
|
||||
OpUnroller<ft::smart_shift>::unroll(src1_n_el, src2_n_el, ((write_type*)dst)[x], mask, op, x_shifted, y);
|
||||
}
|
||||
else
|
||||
{
|
||||
for (int real_x = x_shifted; real_x < src1_.cols; ++real_x)
|
||||
{
|
||||
if (mask(y, real_x))
|
||||
dst[real_x] = op(src1[real_x], src2[real_x]);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
template <typename T1, typename T2, typename D, typename BinOp, typename Mask>
|
||||
static __global__ void transformSimple(const PtrStepSz<T1> src1, const PtrStep<T2> src2, PtrStep<D> dst,
|
||||
const Mask mask, const BinOp op)
|
||||
{
|
||||
const int x = blockDim.x * blockIdx.x + threadIdx.x;
|
||||
const int y = blockDim.y * blockIdx.y + threadIdx.y;
|
||||
|
||||
if (x < src1.cols && y < src1.rows && mask(y, x))
|
||||
{
|
||||
const T1 src1_data = src1.ptr(y)[x];
|
||||
const T2 src2_data = src2.ptr(y)[x];
|
||||
dst.ptr(y)[x] = op(src1_data, src2_data);
|
||||
}
|
||||
}
|
||||
|
||||
template <bool UseSmart> struct TransformDispatcher;
|
||||
template<> struct TransformDispatcher<false>
|
||||
{
|
||||
template <typename T, typename D, typename UnOp, typename Mask>
|
||||
static void call(PtrStepSz<T> src, PtrStepSz<D> dst, UnOp op, Mask mask, cudaStream_t stream)
|
||||
{
|
||||
typedef TransformFunctorTraits<UnOp> ft;
|
||||
|
||||
const dim3 threads(ft::simple_block_dim_x, ft::simple_block_dim_y, 1);
|
||||
const dim3 grid(divUp(src.cols, threads.x), divUp(src.rows, threads.y), 1);
|
||||
|
||||
transformSimple<T, D><<<grid, threads, 0, stream>>>(src, dst, mask, op);
|
||||
cudaSafeCall( cudaGetLastError() );
|
||||
|
||||
if (stream == 0)
|
||||
cudaSafeCall( cudaDeviceSynchronize() );
|
||||
}
|
||||
|
||||
template <typename T1, typename T2, typename D, typename BinOp, typename Mask>
|
||||
static void call(PtrStepSz<T1> src1, PtrStepSz<T2> src2, PtrStepSz<D> dst, BinOp op, Mask mask, cudaStream_t stream)
|
||||
{
|
||||
typedef TransformFunctorTraits<BinOp> ft;
|
||||
|
||||
const dim3 threads(ft::simple_block_dim_x, ft::simple_block_dim_y, 1);
|
||||
const dim3 grid(divUp(src1.cols, threads.x), divUp(src1.rows, threads.y), 1);
|
||||
|
||||
transformSimple<T1, T2, D><<<grid, threads, 0, stream>>>(src1, src2, dst, mask, op);
|
||||
cudaSafeCall( cudaGetLastError() );
|
||||
|
||||
if (stream == 0)
|
||||
cudaSafeCall( cudaDeviceSynchronize() );
|
||||
}
|
||||
};
|
||||
template<> struct TransformDispatcher<true>
|
||||
{
|
||||
template <typename T, typename D, typename UnOp, typename Mask>
|
||||
static void call(PtrStepSz<T> src, PtrStepSz<D> dst, UnOp op, Mask mask, cudaStream_t stream)
|
||||
{
|
||||
typedef TransformFunctorTraits<UnOp> ft;
|
||||
|
||||
CV_StaticAssert(ft::smart_shift != 1, "");
|
||||
|
||||
if (!isAligned(src.data, ft::smart_shift * sizeof(T)) || !isAligned(src.step, ft::smart_shift * sizeof(T)) ||
|
||||
!isAligned(dst.data, ft::smart_shift * sizeof(D)) || !isAligned(dst.step, ft::smart_shift * sizeof(D)))
|
||||
{
|
||||
TransformDispatcher<false>::call(src, dst, op, mask, stream);
|
||||
return;
|
||||
}
|
||||
|
||||
const dim3 threads(ft::smart_block_dim_x, ft::smart_block_dim_y, 1);
|
||||
const dim3 grid(divUp(src.cols, threads.x * ft::smart_shift), divUp(src.rows, threads.y), 1);
|
||||
|
||||
transformSmart<T, D><<<grid, threads, 0, stream>>>(src, dst, mask, op);
|
||||
cudaSafeCall( cudaGetLastError() );
|
||||
|
||||
if (stream == 0)
|
||||
cudaSafeCall( cudaDeviceSynchronize() );
|
||||
}
|
||||
|
||||
template <typename T1, typename T2, typename D, typename BinOp, typename Mask>
|
||||
static void call(PtrStepSz<T1> src1, PtrStepSz<T2> src2, PtrStepSz<D> dst, BinOp op, Mask mask, cudaStream_t stream)
|
||||
{
|
||||
typedef TransformFunctorTraits<BinOp> ft;
|
||||
|
||||
CV_StaticAssert(ft::smart_shift != 1, "");
|
||||
|
||||
if (!isAligned(src1.data, ft::smart_shift * sizeof(T1)) || !isAligned(src1.step, ft::smart_shift * sizeof(T1)) ||
|
||||
!isAligned(src2.data, ft::smart_shift * sizeof(T2)) || !isAligned(src2.step, ft::smart_shift * sizeof(T2)) ||
|
||||
!isAligned(dst.data, ft::smart_shift * sizeof(D)) || !isAligned(dst.step, ft::smart_shift * sizeof(D)))
|
||||
{
|
||||
TransformDispatcher<false>::call(src1, src2, dst, op, mask, stream);
|
||||
return;
|
||||
}
|
||||
|
||||
const dim3 threads(ft::smart_block_dim_x, ft::smart_block_dim_y, 1);
|
||||
const dim3 grid(divUp(src1.cols, threads.x * ft::smart_shift), divUp(src1.rows, threads.y), 1);
|
||||
|
||||
transformSmart<T1, T2, D><<<grid, threads, 0, stream>>>(src1, src2, dst, mask, op);
|
||||
cudaSafeCall( cudaGetLastError() );
|
||||
|
||||
if (stream == 0)
|
||||
cudaSafeCall( cudaDeviceSynchronize() );
|
||||
}
|
||||
};
|
||||
} // namespace transform_detail
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_TRANSFORM_DETAIL_HPP
|
||||
|
|
@ -0,0 +1,191 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_TYPE_TRAITS_DETAIL_HPP
|
||||
#define OPENCV_CUDA_TYPE_TRAITS_DETAIL_HPP
|
||||
|
||||
#include "../common.hpp"
|
||||
#include "../vec_traits.hpp"
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
namespace type_traits_detail
|
||||
{
|
||||
template <bool, typename T1, typename T2> struct Select { typedef T1 type; };
|
||||
template <typename T1, typename T2> struct Select<false, T1, T2> { typedef T2 type; };
|
||||
|
||||
template <typename T> struct IsSignedIntergral { enum {value = 0}; };
|
||||
template <> struct IsSignedIntergral<schar> { enum {value = 1}; };
|
||||
template <> struct IsSignedIntergral<char1> { enum {value = 1}; };
|
||||
template <> struct IsSignedIntergral<short> { enum {value = 1}; };
|
||||
template <> struct IsSignedIntergral<short1> { enum {value = 1}; };
|
||||
template <> struct IsSignedIntergral<int> { enum {value = 1}; };
|
||||
template <> struct IsSignedIntergral<int1> { enum {value = 1}; };
|
||||
|
||||
template <typename T> struct IsUnsignedIntegral { enum {value = 0}; };
|
||||
template <> struct IsUnsignedIntegral<uchar> { enum {value = 1}; };
|
||||
template <> struct IsUnsignedIntegral<uchar1> { enum {value = 1}; };
|
||||
template <> struct IsUnsignedIntegral<ushort> { enum {value = 1}; };
|
||||
template <> struct IsUnsignedIntegral<ushort1> { enum {value = 1}; };
|
||||
template <> struct IsUnsignedIntegral<uint> { enum {value = 1}; };
|
||||
template <> struct IsUnsignedIntegral<uint1> { enum {value = 1}; };
|
||||
|
||||
template <typename T> struct IsIntegral { enum {value = IsSignedIntergral<T>::value || IsUnsignedIntegral<T>::value}; };
|
||||
template <> struct IsIntegral<char> { enum {value = 1}; };
|
||||
template <> struct IsIntegral<bool> { enum {value = 1}; };
|
||||
|
||||
template <typename T> struct IsFloat { enum {value = 0}; };
|
||||
template <> struct IsFloat<float> { enum {value = 1}; };
|
||||
template <> struct IsFloat<double> { enum {value = 1}; };
|
||||
|
||||
template <typename T> struct IsVec { enum {value = 0}; };
|
||||
template <> struct IsVec<uchar1> { enum {value = 1}; };
|
||||
template <> struct IsVec<uchar2> { enum {value = 1}; };
|
||||
template <> struct IsVec<uchar3> { enum {value = 1}; };
|
||||
template <> struct IsVec<uchar4> { enum {value = 1}; };
|
||||
template <> struct IsVec<uchar8> { enum {value = 1}; };
|
||||
template <> struct IsVec<char1> { enum {value = 1}; };
|
||||
template <> struct IsVec<char2> { enum {value = 1}; };
|
||||
template <> struct IsVec<char3> { enum {value = 1}; };
|
||||
template <> struct IsVec<char4> { enum {value = 1}; };
|
||||
template <> struct IsVec<char8> { enum {value = 1}; };
|
||||
template <> struct IsVec<ushort1> { enum {value = 1}; };
|
||||
template <> struct IsVec<ushort2> { enum {value = 1}; };
|
||||
template <> struct IsVec<ushort3> { enum {value = 1}; };
|
||||
template <> struct IsVec<ushort4> { enum {value = 1}; };
|
||||
template <> struct IsVec<ushort8> { enum {value = 1}; };
|
||||
template <> struct IsVec<short1> { enum {value = 1}; };
|
||||
template <> struct IsVec<short2> { enum {value = 1}; };
|
||||
template <> struct IsVec<short3> { enum {value = 1}; };
|
||||
template <> struct IsVec<short4> { enum {value = 1}; };
|
||||
template <> struct IsVec<short8> { enum {value = 1}; };
|
||||
template <> struct IsVec<uint1> { enum {value = 1}; };
|
||||
template <> struct IsVec<uint2> { enum {value = 1}; };
|
||||
template <> struct IsVec<uint3> { enum {value = 1}; };
|
||||
template <> struct IsVec<uint4> { enum {value = 1}; };
|
||||
template <> struct IsVec<uint8> { enum {value = 1}; };
|
||||
template <> struct IsVec<int1> { enum {value = 1}; };
|
||||
template <> struct IsVec<int2> { enum {value = 1}; };
|
||||
template <> struct IsVec<int3> { enum {value = 1}; };
|
||||
template <> struct IsVec<int4> { enum {value = 1}; };
|
||||
template <> struct IsVec<int8> { enum {value = 1}; };
|
||||
template <> struct IsVec<float1> { enum {value = 1}; };
|
||||
template <> struct IsVec<float2> { enum {value = 1}; };
|
||||
template <> struct IsVec<float3> { enum {value = 1}; };
|
||||
template <> struct IsVec<float4> { enum {value = 1}; };
|
||||
template <> struct IsVec<float8> { enum {value = 1}; };
|
||||
template <> struct IsVec<double1> { enum {value = 1}; };
|
||||
template <> struct IsVec<double2> { enum {value = 1}; };
|
||||
template <> struct IsVec<double3> { enum {value = 1}; };
|
||||
template <> struct IsVec<double4> { enum {value = 1}; };
|
||||
template <> struct IsVec<double8> { enum {value = 1}; };
|
||||
|
||||
template <class U> struct AddParameterType { typedef const U& type; };
|
||||
template <class U> struct AddParameterType<U&> { typedef U& type; };
|
||||
template <> struct AddParameterType<void> { typedef void type; };
|
||||
|
||||
template <class U> struct ReferenceTraits
|
||||
{
|
||||
enum { value = false };
|
||||
typedef U type;
|
||||
};
|
||||
template <class U> struct ReferenceTraits<U&>
|
||||
{
|
||||
enum { value = true };
|
||||
typedef U type;
|
||||
};
|
||||
|
||||
template <class U> struct PointerTraits
|
||||
{
|
||||
enum { value = false };
|
||||
typedef void type;
|
||||
};
|
||||
template <class U> struct PointerTraits<U*>
|
||||
{
|
||||
enum { value = true };
|
||||
typedef U type;
|
||||
};
|
||||
template <class U> struct PointerTraits<U*&>
|
||||
{
|
||||
enum { value = true };
|
||||
typedef U type;
|
||||
};
|
||||
|
||||
template <class U> struct UnConst
|
||||
{
|
||||
typedef U type;
|
||||
enum { value = 0 };
|
||||
};
|
||||
template <class U> struct UnConst<const U>
|
||||
{
|
||||
typedef U type;
|
||||
enum { value = 1 };
|
||||
};
|
||||
template <class U> struct UnConst<const U&>
|
||||
{
|
||||
typedef U& type;
|
||||
enum { value = 1 };
|
||||
};
|
||||
|
||||
template <class U> struct UnVolatile
|
||||
{
|
||||
typedef U type;
|
||||
enum { value = 0 };
|
||||
};
|
||||
template <class U> struct UnVolatile<volatile U>
|
||||
{
|
||||
typedef U type;
|
||||
enum { value = 1 };
|
||||
};
|
||||
template <class U> struct UnVolatile<volatile U&>
|
||||
{
|
||||
typedef U& type;
|
||||
enum { value = 1 };
|
||||
};
|
||||
} // namespace type_traits_detail
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_TYPE_TRAITS_DETAIL_HPP
|
||||
|
|
@ -0,0 +1,121 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_VEC_DISTANCE_DETAIL_HPP
|
||||
#define OPENCV_CUDA_VEC_DISTANCE_DETAIL_HPP
|
||||
|
||||
#include "../datamov_utils.hpp"
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
namespace vec_distance_detail
|
||||
{
|
||||
template <int THREAD_DIM, int N> struct UnrollVecDiffCached
|
||||
{
|
||||
template <typename Dist, typename T1, typename T2>
|
||||
static __device__ void calcCheck(const T1* vecCached, const T2* vecGlob, int len, Dist& dist, int ind)
|
||||
{
|
||||
if (ind < len)
|
||||
{
|
||||
T1 val1 = *vecCached++;
|
||||
|
||||
T2 val2;
|
||||
ForceGlob<T2>::Load(vecGlob, ind, val2);
|
||||
|
||||
dist.reduceIter(val1, val2);
|
||||
|
||||
UnrollVecDiffCached<THREAD_DIM, N - 1>::calcCheck(vecCached, vecGlob, len, dist, ind + THREAD_DIM);
|
||||
}
|
||||
}
|
||||
|
||||
template <typename Dist, typename T1, typename T2>
|
||||
static __device__ void calcWithoutCheck(const T1* vecCached, const T2* vecGlob, Dist& dist)
|
||||
{
|
||||
T1 val1 = *vecCached++;
|
||||
|
||||
T2 val2;
|
||||
ForceGlob<T2>::Load(vecGlob, 0, val2);
|
||||
vecGlob += THREAD_DIM;
|
||||
|
||||
dist.reduceIter(val1, val2);
|
||||
|
||||
UnrollVecDiffCached<THREAD_DIM, N - 1>::calcWithoutCheck(vecCached, vecGlob, dist);
|
||||
}
|
||||
};
|
||||
template <int THREAD_DIM> struct UnrollVecDiffCached<THREAD_DIM, 0>
|
||||
{
|
||||
template <typename Dist, typename T1, typename T2>
|
||||
static __device__ __forceinline__ void calcCheck(const T1*, const T2*, int, Dist&, int)
|
||||
{
|
||||
}
|
||||
|
||||
template <typename Dist, typename T1, typename T2>
|
||||
static __device__ __forceinline__ void calcWithoutCheck(const T1*, const T2*, Dist&)
|
||||
{
|
||||
}
|
||||
};
|
||||
|
||||
template <int THREAD_DIM, int MAX_LEN, bool LEN_EQ_MAX_LEN> struct VecDiffCachedCalculator;
|
||||
template <int THREAD_DIM, int MAX_LEN> struct VecDiffCachedCalculator<THREAD_DIM, MAX_LEN, false>
|
||||
{
|
||||
template <typename Dist, typename T1, typename T2>
|
||||
static __device__ __forceinline__ void calc(const T1* vecCached, const T2* vecGlob, int len, Dist& dist, int tid)
|
||||
{
|
||||
UnrollVecDiffCached<THREAD_DIM, MAX_LEN / THREAD_DIM>::calcCheck(vecCached, vecGlob, len, dist, tid);
|
||||
}
|
||||
};
|
||||
template <int THREAD_DIM, int MAX_LEN> struct VecDiffCachedCalculator<THREAD_DIM, MAX_LEN, true>
|
||||
{
|
||||
template <typename Dist, typename T1, typename T2>
|
||||
static __device__ __forceinline__ void calc(const T1* vecCached, const T2* vecGlob, int len, Dist& dist, int tid)
|
||||
{
|
||||
UnrollVecDiffCached<THREAD_DIM, MAX_LEN / THREAD_DIM>::calcWithoutCheck(vecCached, vecGlob + tid, dist);
|
||||
}
|
||||
};
|
||||
} // namespace vec_distance_detail
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_VEC_DISTANCE_DETAIL_HPP
|
||||
|
|
@ -0,0 +1,88 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_DYNAMIC_SMEM_HPP
|
||||
#define OPENCV_CUDA_DYNAMIC_SMEM_HPP
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template<class T> struct DynamicSharedMem
|
||||
{
|
||||
__device__ __forceinline__ operator T*()
|
||||
{
|
||||
extern __shared__ int __smem[];
|
||||
return (T*)__smem;
|
||||
}
|
||||
|
||||
__device__ __forceinline__ operator const T*() const
|
||||
{
|
||||
extern __shared__ int __smem[];
|
||||
return (T*)__smem;
|
||||
}
|
||||
};
|
||||
|
||||
// specialize for double to avoid unaligned memory access compile errors
|
||||
template<> struct DynamicSharedMem<double>
|
||||
{
|
||||
__device__ __forceinline__ operator double*()
|
||||
{
|
||||
extern __shared__ double __smem_d[];
|
||||
return (double*)__smem_d;
|
||||
}
|
||||
|
||||
__device__ __forceinline__ operator const double*() const
|
||||
{
|
||||
extern __shared__ double __smem_d[];
|
||||
return (double*)__smem_d;
|
||||
}
|
||||
};
|
||||
}}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_DYNAMIC_SMEM_HPP
|
||||
|
|
@ -0,0 +1,269 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_EMULATION_HPP_
|
||||
#define OPENCV_CUDA_EMULATION_HPP_
|
||||
|
||||
#include "common.hpp"
|
||||
#include "warp_reduce.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
struct Emulation
|
||||
{
|
||||
|
||||
static __device__ __forceinline__ int syncthreadsOr(int pred)
|
||||
{
|
||||
#if defined (__CUDA_ARCH__) && (__CUDA_ARCH__ < 200)
|
||||
// just campilation stab
|
||||
return 0;
|
||||
#else
|
||||
return __syncthreads_or(pred);
|
||||
#endif
|
||||
}
|
||||
|
||||
template<int CTA_SIZE>
|
||||
static __forceinline__ __device__ int Ballot(int predicate)
|
||||
{
|
||||
#if defined (__CUDA_ARCH__) && (__CUDA_ARCH__ >= 200)
|
||||
return __ballot(predicate);
|
||||
#else
|
||||
__shared__ volatile int cta_buffer[CTA_SIZE];
|
||||
|
||||
int tid = threadIdx.x;
|
||||
cta_buffer[tid] = predicate ? (1 << (tid & 31)) : 0;
|
||||
return warp_reduce(cta_buffer);
|
||||
#endif
|
||||
}
|
||||
|
||||
struct smem
|
||||
{
|
||||
enum { TAG_MASK = (1U << ( (sizeof(unsigned int) << 3) - 5U)) - 1U };
|
||||
|
||||
template<typename T>
|
||||
static __device__ __forceinline__ T atomicInc(T* address, T val)
|
||||
{
|
||||
#if defined (__CUDA_ARCH__) && (__CUDA_ARCH__ < 120)
|
||||
T count;
|
||||
unsigned int tag = threadIdx.x << ( (sizeof(unsigned int) << 3) - 5U);
|
||||
do
|
||||
{
|
||||
count = *address & TAG_MASK;
|
||||
count = tag | (count + 1);
|
||||
*address = count;
|
||||
} while (*address != count);
|
||||
|
||||
return (count & TAG_MASK) - 1;
|
||||
#else
|
||||
return ::atomicInc(address, val);
|
||||
#endif
|
||||
}
|
||||
|
||||
template<typename T>
|
||||
static __device__ __forceinline__ T atomicAdd(T* address, T val)
|
||||
{
|
||||
#if defined (__CUDA_ARCH__) && (__CUDA_ARCH__ < 120)
|
||||
T count;
|
||||
unsigned int tag = threadIdx.x << ( (sizeof(unsigned int) << 3) - 5U);
|
||||
do
|
||||
{
|
||||
count = *address & TAG_MASK;
|
||||
count = tag | (count + val);
|
||||
*address = count;
|
||||
} while (*address != count);
|
||||
|
||||
return (count & TAG_MASK) - val;
|
||||
#else
|
||||
return ::atomicAdd(address, val);
|
||||
#endif
|
||||
}
|
||||
|
||||
template<typename T>
|
||||
static __device__ __forceinline__ T atomicMin(T* address, T val)
|
||||
{
|
||||
#if defined (__CUDA_ARCH__) && (__CUDA_ARCH__ < 120)
|
||||
T count = ::min(*address, val);
|
||||
do
|
||||
{
|
||||
*address = count;
|
||||
} while (*address > count);
|
||||
|
||||
return count;
|
||||
#else
|
||||
return ::atomicMin(address, val);
|
||||
#endif
|
||||
}
|
||||
}; // struct cmem
|
||||
|
||||
struct glob
|
||||
{
|
||||
static __device__ __forceinline__ int atomicAdd(int* address, int val)
|
||||
{
|
||||
return ::atomicAdd(address, val);
|
||||
}
|
||||
static __device__ __forceinline__ unsigned int atomicAdd(unsigned int* address, unsigned int val)
|
||||
{
|
||||
return ::atomicAdd(address, val);
|
||||
}
|
||||
static __device__ __forceinline__ float atomicAdd(float* address, float val)
|
||||
{
|
||||
#if __CUDA_ARCH__ >= 200
|
||||
return ::atomicAdd(address, val);
|
||||
#else
|
||||
int* address_as_i = (int*) address;
|
||||
int old = *address_as_i, assumed;
|
||||
do {
|
||||
assumed = old;
|
||||
old = ::atomicCAS(address_as_i, assumed,
|
||||
__float_as_int(val + __int_as_float(assumed)));
|
||||
} while (assumed != old);
|
||||
return __int_as_float(old);
|
||||
#endif
|
||||
}
|
||||
static __device__ __forceinline__ double atomicAdd(double* address, double val)
|
||||
{
|
||||
#if __CUDA_ARCH__ >= 130
|
||||
unsigned long long int* address_as_ull = (unsigned long long int*) address;
|
||||
unsigned long long int old = *address_as_ull, assumed;
|
||||
do {
|
||||
assumed = old;
|
||||
old = ::atomicCAS(address_as_ull, assumed,
|
||||
__double_as_longlong(val + __longlong_as_double(assumed)));
|
||||
} while (assumed != old);
|
||||
return __longlong_as_double(old);
|
||||
#else
|
||||
CV_UNUSED(address);
|
||||
CV_UNUSED(val);
|
||||
return 0.0;
|
||||
#endif
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ int atomicMin(int* address, int val)
|
||||
{
|
||||
return ::atomicMin(address, val);
|
||||
}
|
||||
static __device__ __forceinline__ float atomicMin(float* address, float val)
|
||||
{
|
||||
#if __CUDA_ARCH__ >= 120
|
||||
int* address_as_i = (int*) address;
|
||||
int old = *address_as_i, assumed;
|
||||
do {
|
||||
assumed = old;
|
||||
old = ::atomicCAS(address_as_i, assumed,
|
||||
__float_as_int(::fminf(val, __int_as_float(assumed))));
|
||||
} while (assumed != old);
|
||||
return __int_as_float(old);
|
||||
#else
|
||||
CV_UNUSED(address);
|
||||
CV_UNUSED(val);
|
||||
return 0.0f;
|
||||
#endif
|
||||
}
|
||||
static __device__ __forceinline__ double atomicMin(double* address, double val)
|
||||
{
|
||||
#if __CUDA_ARCH__ >= 130
|
||||
unsigned long long int* address_as_ull = (unsigned long long int*) address;
|
||||
unsigned long long int old = *address_as_ull, assumed;
|
||||
do {
|
||||
assumed = old;
|
||||
old = ::atomicCAS(address_as_ull, assumed,
|
||||
__double_as_longlong(::fmin(val, __longlong_as_double(assumed))));
|
||||
} while (assumed != old);
|
||||
return __longlong_as_double(old);
|
||||
#else
|
||||
CV_UNUSED(address);
|
||||
CV_UNUSED(val);
|
||||
return 0.0;
|
||||
#endif
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ int atomicMax(int* address, int val)
|
||||
{
|
||||
return ::atomicMax(address, val);
|
||||
}
|
||||
static __device__ __forceinline__ float atomicMax(float* address, float val)
|
||||
{
|
||||
#if __CUDA_ARCH__ >= 120
|
||||
int* address_as_i = (int*) address;
|
||||
int old = *address_as_i, assumed;
|
||||
do {
|
||||
assumed = old;
|
||||
old = ::atomicCAS(address_as_i, assumed,
|
||||
__float_as_int(::fmaxf(val, __int_as_float(assumed))));
|
||||
} while (assumed != old);
|
||||
return __int_as_float(old);
|
||||
#else
|
||||
CV_UNUSED(address);
|
||||
CV_UNUSED(val);
|
||||
return 0.0f;
|
||||
#endif
|
||||
}
|
||||
static __device__ __forceinline__ double atomicMax(double* address, double val)
|
||||
{
|
||||
#if __CUDA_ARCH__ >= 130
|
||||
unsigned long long int* address_as_ull = (unsigned long long int*) address;
|
||||
unsigned long long int old = *address_as_ull, assumed;
|
||||
do {
|
||||
assumed = old;
|
||||
old = ::atomicCAS(address_as_ull, assumed,
|
||||
__double_as_longlong(::fmax(val, __longlong_as_double(assumed))));
|
||||
} while (assumed != old);
|
||||
return __longlong_as_double(old);
|
||||
#else
|
||||
CV_UNUSED(address);
|
||||
CV_UNUSED(val);
|
||||
return 0.0;
|
||||
#endif
|
||||
}
|
||||
};
|
||||
}; //struct Emulation
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif /* OPENCV_CUDA_EMULATION_HPP_ */
|
||||
|
|
@ -0,0 +1,286 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_FILTERS_HPP
|
||||
#define OPENCV_CUDA_FILTERS_HPP
|
||||
|
||||
#include "saturate_cast.hpp"
|
||||
#include "vec_traits.hpp"
|
||||
#include "vec_math.hpp"
|
||||
#include "type_traits.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template <typename Ptr2D> struct PointFilter
|
||||
{
|
||||
typedef typename Ptr2D::elem_type elem_type;
|
||||
typedef float index_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ PointFilter(const Ptr2D& src_, float fx = 0.f, float fy = 0.f)
|
||||
: src(src_)
|
||||
{
|
||||
CV_UNUSED(fx);
|
||||
CV_UNUSED(fy);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ elem_type operator ()(float y, float x) const
|
||||
{
|
||||
return src(__float2int_rz(y), __float2int_rz(x));
|
||||
}
|
||||
|
||||
Ptr2D src;
|
||||
};
|
||||
|
||||
template <typename Ptr2D> struct LinearFilter
|
||||
{
|
||||
typedef typename Ptr2D::elem_type elem_type;
|
||||
typedef float index_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ LinearFilter(const Ptr2D& src_, float fx = 0.f, float fy = 0.f)
|
||||
: src(src_)
|
||||
{
|
||||
CV_UNUSED(fx);
|
||||
CV_UNUSED(fy);
|
||||
}
|
||||
__device__ __forceinline__ elem_type operator ()(float y, float x) const
|
||||
{
|
||||
typedef typename TypeVec<float, VecTraits<elem_type>::cn>::vec_type work_type;
|
||||
|
||||
work_type out = VecTraits<work_type>::all(0);
|
||||
|
||||
const int x1 = __float2int_rd(x);
|
||||
const int y1 = __float2int_rd(y);
|
||||
const int x2 = x1 + 1;
|
||||
const int y2 = y1 + 1;
|
||||
|
||||
elem_type src_reg = src(y1, x1);
|
||||
out = out + src_reg * ((x2 - x) * (y2 - y));
|
||||
|
||||
src_reg = src(y1, x2);
|
||||
out = out + src_reg * ((x - x1) * (y2 - y));
|
||||
|
||||
src_reg = src(y2, x1);
|
||||
out = out + src_reg * ((x2 - x) * (y - y1));
|
||||
|
||||
src_reg = src(y2, x2);
|
||||
out = out + src_reg * ((x - x1) * (y - y1));
|
||||
|
||||
return saturate_cast<elem_type>(out);
|
||||
}
|
||||
|
||||
Ptr2D src;
|
||||
};
|
||||
|
||||
template <typename Ptr2D> struct CubicFilter
|
||||
{
|
||||
typedef typename Ptr2D::elem_type elem_type;
|
||||
typedef float index_type;
|
||||
typedef typename TypeVec<float, VecTraits<elem_type>::cn>::vec_type work_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ CubicFilter(const Ptr2D& src_, float fx = 0.f, float fy = 0.f)
|
||||
: src(src_)
|
||||
{
|
||||
CV_UNUSED(fx);
|
||||
CV_UNUSED(fy);
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ float bicubicCoeff(float x_)
|
||||
{
|
||||
float x = fabsf(x_);
|
||||
if (x <= 1.0f)
|
||||
{
|
||||
return x * x * (1.5f * x - 2.5f) + 1.0f;
|
||||
}
|
||||
else if (x < 2.0f)
|
||||
{
|
||||
return x * (x * (-0.5f * x + 2.5f) - 4.0f) + 2.0f;
|
||||
}
|
||||
else
|
||||
{
|
||||
return 0.0f;
|
||||
}
|
||||
}
|
||||
|
||||
__device__ elem_type operator ()(float y, float x) const
|
||||
{
|
||||
const float xmin = ::ceilf(x - 2.0f);
|
||||
const float xmax = ::floorf(x + 2.0f);
|
||||
|
||||
const float ymin = ::ceilf(y - 2.0f);
|
||||
const float ymax = ::floorf(y + 2.0f);
|
||||
|
||||
work_type sum = VecTraits<work_type>::all(0);
|
||||
float wsum = 0.0f;
|
||||
|
||||
for (float cy = ymin; cy <= ymax; cy += 1.0f)
|
||||
{
|
||||
for (float cx = xmin; cx <= xmax; cx += 1.0f)
|
||||
{
|
||||
const float w = bicubicCoeff(x - cx) * bicubicCoeff(y - cy);
|
||||
sum = sum + w * src(__float2int_rd(cy), __float2int_rd(cx));
|
||||
wsum += w;
|
||||
}
|
||||
}
|
||||
|
||||
work_type res = (!wsum)? VecTraits<work_type>::all(0) : sum / wsum;
|
||||
|
||||
return saturate_cast<elem_type>(res);
|
||||
}
|
||||
|
||||
Ptr2D src;
|
||||
};
|
||||
// for integer scaling
|
||||
template <typename Ptr2D> struct IntegerAreaFilter
|
||||
{
|
||||
typedef typename Ptr2D::elem_type elem_type;
|
||||
typedef float index_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ IntegerAreaFilter(const Ptr2D& src_, float scale_x_, float scale_y_)
|
||||
: src(src_), scale_x(scale_x_), scale_y(scale_y_), scale(1.f / (scale_x * scale_y)) {}
|
||||
|
||||
__device__ __forceinline__ elem_type operator ()(float y, float x) const
|
||||
{
|
||||
float fsx1 = x * scale_x;
|
||||
float fsx2 = fsx1 + scale_x;
|
||||
|
||||
int sx1 = __float2int_ru(fsx1);
|
||||
int sx2 = __float2int_rd(fsx2);
|
||||
|
||||
float fsy1 = y * scale_y;
|
||||
float fsy2 = fsy1 + scale_y;
|
||||
|
||||
int sy1 = __float2int_ru(fsy1);
|
||||
int sy2 = __float2int_rd(fsy2);
|
||||
|
||||
typedef typename TypeVec<float, VecTraits<elem_type>::cn>::vec_type work_type;
|
||||
work_type out = VecTraits<work_type>::all(0.f);
|
||||
|
||||
for(int dy = sy1; dy < sy2; ++dy)
|
||||
for(int dx = sx1; dx < sx2; ++dx)
|
||||
{
|
||||
out = out + src(dy, dx) * scale;
|
||||
}
|
||||
|
||||
return saturate_cast<elem_type>(out);
|
||||
}
|
||||
|
||||
Ptr2D src;
|
||||
float scale_x, scale_y ,scale;
|
||||
};
|
||||
|
||||
template <typename Ptr2D> struct AreaFilter
|
||||
{
|
||||
typedef typename Ptr2D::elem_type elem_type;
|
||||
typedef float index_type;
|
||||
|
||||
explicit __host__ __device__ __forceinline__ AreaFilter(const Ptr2D& src_, float scale_x_, float scale_y_)
|
||||
: src(src_), scale_x(scale_x_), scale_y(scale_y_){}
|
||||
|
||||
__device__ __forceinline__ elem_type operator ()(float y, float x) const
|
||||
{
|
||||
float fsx1 = x * scale_x;
|
||||
float fsx2 = fsx1 + scale_x;
|
||||
|
||||
int sx1 = __float2int_ru(fsx1);
|
||||
int sx2 = __float2int_rd(fsx2);
|
||||
|
||||
float fsy1 = y * scale_y;
|
||||
float fsy2 = fsy1 + scale_y;
|
||||
|
||||
int sy1 = __float2int_ru(fsy1);
|
||||
int sy2 = __float2int_rd(fsy2);
|
||||
|
||||
float scale = 1.f / (fminf(scale_x, src.width - fsx1) * fminf(scale_y, src.height - fsy1));
|
||||
|
||||
typedef typename TypeVec<float, VecTraits<elem_type>::cn>::vec_type work_type;
|
||||
work_type out = VecTraits<work_type>::all(0.f);
|
||||
|
||||
for (int dy = sy1; dy < sy2; ++dy)
|
||||
{
|
||||
for (int dx = sx1; dx < sx2; ++dx)
|
||||
out = out + src(dy, dx) * scale;
|
||||
|
||||
if (sx1 > fsx1)
|
||||
out = out + src(dy, (sx1 -1) ) * ((sx1 - fsx1) * scale);
|
||||
|
||||
if (sx2 < fsx2)
|
||||
out = out + src(dy, sx2) * ((fsx2 -sx2) * scale);
|
||||
}
|
||||
|
||||
if (sy1 > fsy1)
|
||||
for (int dx = sx1; dx < sx2; ++dx)
|
||||
out = out + src( (sy1 - 1) , dx) * ((sy1 -fsy1) * scale);
|
||||
|
||||
if (sy2 < fsy2)
|
||||
for (int dx = sx1; dx < sx2; ++dx)
|
||||
out = out + src(sy2, dx) * ((fsy2 -sy2) * scale);
|
||||
|
||||
if ((sy1 > fsy1) && (sx1 > fsx1))
|
||||
out = out + src( (sy1 - 1) , (sx1 - 1)) * ((sy1 -fsy1) * (sx1 -fsx1) * scale);
|
||||
|
||||
if ((sy1 > fsy1) && (sx2 < fsx2))
|
||||
out = out + src( (sy1 - 1) , sx2) * ((sy1 -fsy1) * (fsx2 -sx2) * scale);
|
||||
|
||||
if ((sy2 < fsy2) && (sx2 < fsx2))
|
||||
out = out + src(sy2, sx2) * ((fsy2 -sy2) * (fsx2 -sx2) * scale);
|
||||
|
||||
if ((sy2 < fsy2) && (sx1 > fsx1))
|
||||
out = out + src(sy2, (sx1 - 1)) * ((fsy2 -sy2) * (sx1 -fsx1) * scale);
|
||||
|
||||
return saturate_cast<elem_type>(out);
|
||||
}
|
||||
|
||||
Ptr2D src;
|
||||
float scale_x, scale_y;
|
||||
int width, haight;
|
||||
};
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_FILTERS_HPP
|
||||
|
|
@ -0,0 +1,79 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_DEVICE_FUNCATTRIB_HPP
|
||||
#define OPENCV_CUDA_DEVICE_FUNCATTRIB_HPP
|
||||
|
||||
#include <cstdio>
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template<class Func>
|
||||
void printFuncAttrib(Func& func)
|
||||
{
|
||||
|
||||
cudaFuncAttributes attrs;
|
||||
cudaFuncGetAttributes(&attrs, func);
|
||||
|
||||
printf("=== Function stats ===\n");
|
||||
printf("Name: \n");
|
||||
printf("sharedSizeBytes = %d\n", attrs.sharedSizeBytes);
|
||||
printf("constSizeBytes = %d\n", attrs.constSizeBytes);
|
||||
printf("localSizeBytes = %d\n", attrs.localSizeBytes);
|
||||
printf("maxThreadsPerBlock = %d\n", attrs.maxThreadsPerBlock);
|
||||
printf("numRegs = %d\n", attrs.numRegs);
|
||||
printf("ptxVersion = %d\n", attrs.ptxVersion);
|
||||
printf("binaryVersion = %d\n", attrs.binaryVersion);
|
||||
printf("\n");
|
||||
fflush(stdout);
|
||||
}
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif /* OPENCV_CUDA_DEVICE_FUNCATTRIB_HPP */
|
||||
|
|
@ -0,0 +1,811 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_FUNCTIONAL_HPP
|
||||
#define OPENCV_CUDA_FUNCTIONAL_HPP
|
||||
|
||||
#include <functional>
|
||||
#include "saturate_cast.hpp"
|
||||
#include "vec_traits.hpp"
|
||||
#include "type_traits.hpp"
|
||||
#include "device_functions.h"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
// Function Objects
|
||||
#ifdef CV_CXX11
|
||||
template<typename Argument, typename Result> struct unary_function
|
||||
{
|
||||
typedef Argument argument_type;
|
||||
typedef Result result_type;
|
||||
};
|
||||
template<typename Argument1, typename Argument2, typename Result> struct binary_function
|
||||
{
|
||||
typedef Argument1 first_argument_type;
|
||||
typedef Argument2 second_argument_type;
|
||||
typedef Result result_type;
|
||||
};
|
||||
#else
|
||||
template<typename Argument, typename Result> struct unary_function : public std::unary_function<Argument, Result> {};
|
||||
template<typename Argument1, typename Argument2, typename Result> struct binary_function : public std::binary_function<Argument1, Argument2, Result> {};
|
||||
#endif
|
||||
|
||||
// Arithmetic Operations
|
||||
template <typename T> struct plus : binary_function<T, T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a + b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ plus() {}
|
||||
__host__ __device__ __forceinline__ plus(const plus&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct minus : binary_function<T, T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a - b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ minus() {}
|
||||
__host__ __device__ __forceinline__ minus(const minus&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct multiplies : binary_function<T, T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a * b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ multiplies() {}
|
||||
__host__ __device__ __forceinline__ multiplies(const multiplies&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct divides : binary_function<T, T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a / b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ divides() {}
|
||||
__host__ __device__ __forceinline__ divides(const divides&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct modulus : binary_function<T, T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a % b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ modulus() {}
|
||||
__host__ __device__ __forceinline__ modulus(const modulus&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct negate : unary_function<T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType a) const
|
||||
{
|
||||
return -a;
|
||||
}
|
||||
__host__ __device__ __forceinline__ negate() {}
|
||||
__host__ __device__ __forceinline__ negate(const negate&) {}
|
||||
};
|
||||
|
||||
// Comparison Operations
|
||||
template <typename T> struct equal_to : binary_function<T, T, bool>
|
||||
{
|
||||
__device__ __forceinline__ bool operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a == b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ equal_to() {}
|
||||
__host__ __device__ __forceinline__ equal_to(const equal_to&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct not_equal_to : binary_function<T, T, bool>
|
||||
{
|
||||
__device__ __forceinline__ bool operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a != b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ not_equal_to() {}
|
||||
__host__ __device__ __forceinline__ not_equal_to(const not_equal_to&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct greater : binary_function<T, T, bool>
|
||||
{
|
||||
__device__ __forceinline__ bool operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a > b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ greater() {}
|
||||
__host__ __device__ __forceinline__ greater(const greater&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct less : binary_function<T, T, bool>
|
||||
{
|
||||
__device__ __forceinline__ bool operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a < b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ less() {}
|
||||
__host__ __device__ __forceinline__ less(const less&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct greater_equal : binary_function<T, T, bool>
|
||||
{
|
||||
__device__ __forceinline__ bool operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a >= b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ greater_equal() {}
|
||||
__host__ __device__ __forceinline__ greater_equal(const greater_equal&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct less_equal : binary_function<T, T, bool>
|
||||
{
|
||||
__device__ __forceinline__ bool operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a <= b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ less_equal() {}
|
||||
__host__ __device__ __forceinline__ less_equal(const less_equal&) {}
|
||||
};
|
||||
|
||||
// Logical Operations
|
||||
template <typename T> struct logical_and : binary_function<T, T, bool>
|
||||
{
|
||||
__device__ __forceinline__ bool operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a && b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ logical_and() {}
|
||||
__host__ __device__ __forceinline__ logical_and(const logical_and&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct logical_or : binary_function<T, T, bool>
|
||||
{
|
||||
__device__ __forceinline__ bool operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a || b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ logical_or() {}
|
||||
__host__ __device__ __forceinline__ logical_or(const logical_or&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct logical_not : unary_function<T, bool>
|
||||
{
|
||||
__device__ __forceinline__ bool operator ()(typename TypeTraits<T>::ParameterType a) const
|
||||
{
|
||||
return !a;
|
||||
}
|
||||
__host__ __device__ __forceinline__ logical_not() {}
|
||||
__host__ __device__ __forceinline__ logical_not(const logical_not&) {}
|
||||
};
|
||||
|
||||
// Bitwise Operations
|
||||
template <typename T> struct bit_and : binary_function<T, T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a & b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ bit_and() {}
|
||||
__host__ __device__ __forceinline__ bit_and(const bit_and&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct bit_or : binary_function<T, T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a | b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ bit_or() {}
|
||||
__host__ __device__ __forceinline__ bit_or(const bit_or&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct bit_xor : binary_function<T, T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType a,
|
||||
typename TypeTraits<T>::ParameterType b) const
|
||||
{
|
||||
return a ^ b;
|
||||
}
|
||||
__host__ __device__ __forceinline__ bit_xor() {}
|
||||
__host__ __device__ __forceinline__ bit_xor(const bit_xor&) {}
|
||||
};
|
||||
|
||||
template <typename T> struct bit_not : unary_function<T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType v) const
|
||||
{
|
||||
return ~v;
|
||||
}
|
||||
__host__ __device__ __forceinline__ bit_not() {}
|
||||
__host__ __device__ __forceinline__ bit_not(const bit_not&) {}
|
||||
};
|
||||
|
||||
// Generalized Identity Operations
|
||||
template <typename T> struct identity : unary_function<T, T>
|
||||
{
|
||||
__device__ __forceinline__ typename TypeTraits<T>::ParameterType operator()(typename TypeTraits<T>::ParameterType x) const
|
||||
{
|
||||
return x;
|
||||
}
|
||||
__host__ __device__ __forceinline__ identity() {}
|
||||
__host__ __device__ __forceinline__ identity(const identity&) {}
|
||||
};
|
||||
|
||||
template <typename T1, typename T2> struct project1st : binary_function<T1, T2, T1>
|
||||
{
|
||||
__device__ __forceinline__ typename TypeTraits<T1>::ParameterType operator()(typename TypeTraits<T1>::ParameterType lhs, typename TypeTraits<T2>::ParameterType rhs) const
|
||||
{
|
||||
return lhs;
|
||||
}
|
||||
__host__ __device__ __forceinline__ project1st() {}
|
||||
__host__ __device__ __forceinline__ project1st(const project1st&) {}
|
||||
};
|
||||
|
||||
template <typename T1, typename T2> struct project2nd : binary_function<T1, T2, T2>
|
||||
{
|
||||
__device__ __forceinline__ typename TypeTraits<T2>::ParameterType operator()(typename TypeTraits<T1>::ParameterType lhs, typename TypeTraits<T2>::ParameterType rhs) const
|
||||
{
|
||||
return rhs;
|
||||
}
|
||||
__host__ __device__ __forceinline__ project2nd() {}
|
||||
__host__ __device__ __forceinline__ project2nd(const project2nd&) {}
|
||||
};
|
||||
|
||||
// Min/Max Operations
|
||||
|
||||
#define OPENCV_CUDA_IMPLEMENT_MINMAX(name, type, op) \
|
||||
template <> struct name<type> : binary_function<type, type, type> \
|
||||
{ \
|
||||
__device__ __forceinline__ type operator()(type lhs, type rhs) const {return op(lhs, rhs);} \
|
||||
__host__ __device__ __forceinline__ name() {}\
|
||||
__host__ __device__ __forceinline__ name(const name&) {}\
|
||||
};
|
||||
|
||||
template <typename T> struct maximum : binary_function<T, T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator()(typename TypeTraits<T>::ParameterType lhs, typename TypeTraits<T>::ParameterType rhs) const
|
||||
{
|
||||
return max(lhs, rhs);
|
||||
}
|
||||
__host__ __device__ __forceinline__ maximum() {}
|
||||
__host__ __device__ __forceinline__ maximum(const maximum&) {}
|
||||
};
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(maximum, uchar, ::max)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(maximum, schar, ::max)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(maximum, char, ::max)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(maximum, ushort, ::max)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(maximum, short, ::max)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(maximum, int, ::max)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(maximum, uint, ::max)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(maximum, float, ::fmax)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(maximum, double, ::fmax)
|
||||
|
||||
template <typename T> struct minimum : binary_function<T, T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator()(typename TypeTraits<T>::ParameterType lhs, typename TypeTraits<T>::ParameterType rhs) const
|
||||
{
|
||||
return min(lhs, rhs);
|
||||
}
|
||||
__host__ __device__ __forceinline__ minimum() {}
|
||||
__host__ __device__ __forceinline__ minimum(const minimum&) {}
|
||||
};
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(minimum, uchar, ::min)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(minimum, schar, ::min)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(minimum, char, ::min)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(minimum, ushort, ::min)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(minimum, short, ::min)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(minimum, int, ::min)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(minimum, uint, ::min)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(minimum, float, ::fmin)
|
||||
OPENCV_CUDA_IMPLEMENT_MINMAX(minimum, double, ::fmin)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_MINMAX
|
||||
|
||||
// Math functions
|
||||
|
||||
template <typename T> struct abs_func : unary_function<T, T>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType x) const
|
||||
{
|
||||
return abs(x);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ abs_func() {}
|
||||
__host__ __device__ __forceinline__ abs_func(const abs_func&) {}
|
||||
};
|
||||
template <> struct abs_func<unsigned char> : unary_function<unsigned char, unsigned char>
|
||||
{
|
||||
__device__ __forceinline__ unsigned char operator ()(unsigned char x) const
|
||||
{
|
||||
return x;
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ abs_func() {}
|
||||
__host__ __device__ __forceinline__ abs_func(const abs_func&) {}
|
||||
};
|
||||
template <> struct abs_func<signed char> : unary_function<signed char, signed char>
|
||||
{
|
||||
__device__ __forceinline__ signed char operator ()(signed char x) const
|
||||
{
|
||||
return ::abs((int)x);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ abs_func() {}
|
||||
__host__ __device__ __forceinline__ abs_func(const abs_func&) {}
|
||||
};
|
||||
template <> struct abs_func<char> : unary_function<char, char>
|
||||
{
|
||||
__device__ __forceinline__ char operator ()(char x) const
|
||||
{
|
||||
return ::abs((int)x);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ abs_func() {}
|
||||
__host__ __device__ __forceinline__ abs_func(const abs_func&) {}
|
||||
};
|
||||
template <> struct abs_func<unsigned short> : unary_function<unsigned short, unsigned short>
|
||||
{
|
||||
__device__ __forceinline__ unsigned short operator ()(unsigned short x) const
|
||||
{
|
||||
return x;
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ abs_func() {}
|
||||
__host__ __device__ __forceinline__ abs_func(const abs_func&) {}
|
||||
};
|
||||
template <> struct abs_func<short> : unary_function<short, short>
|
||||
{
|
||||
__device__ __forceinline__ short operator ()(short x) const
|
||||
{
|
||||
return ::abs((int)x);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ abs_func() {}
|
||||
__host__ __device__ __forceinline__ abs_func(const abs_func&) {}
|
||||
};
|
||||
template <> struct abs_func<unsigned int> : unary_function<unsigned int, unsigned int>
|
||||
{
|
||||
__device__ __forceinline__ unsigned int operator ()(unsigned int x) const
|
||||
{
|
||||
return x;
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ abs_func() {}
|
||||
__host__ __device__ __forceinline__ abs_func(const abs_func&) {}
|
||||
};
|
||||
template <> struct abs_func<int> : unary_function<int, int>
|
||||
{
|
||||
__device__ __forceinline__ int operator ()(int x) const
|
||||
{
|
||||
return ::abs(x);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ abs_func() {}
|
||||
__host__ __device__ __forceinline__ abs_func(const abs_func&) {}
|
||||
};
|
||||
template <> struct abs_func<float> : unary_function<float, float>
|
||||
{
|
||||
__device__ __forceinline__ float operator ()(float x) const
|
||||
{
|
||||
return ::fabsf(x);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ abs_func() {}
|
||||
__host__ __device__ __forceinline__ abs_func(const abs_func&) {}
|
||||
};
|
||||
template <> struct abs_func<double> : unary_function<double, double>
|
||||
{
|
||||
__device__ __forceinline__ double operator ()(double x) const
|
||||
{
|
||||
return ::fabs(x);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ abs_func() {}
|
||||
__host__ __device__ __forceinline__ abs_func(const abs_func&) {}
|
||||
};
|
||||
|
||||
#define OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(name, func) \
|
||||
template <typename T> struct name ## _func : unary_function<T, float> \
|
||||
{ \
|
||||
__device__ __forceinline__ float operator ()(typename TypeTraits<T>::ParameterType v) const \
|
||||
{ \
|
||||
return func ## f(v); \
|
||||
} \
|
||||
__host__ __device__ __forceinline__ name ## _func() {} \
|
||||
__host__ __device__ __forceinline__ name ## _func(const name ## _func&) {} \
|
||||
}; \
|
||||
template <> struct name ## _func<double> : unary_function<double, double> \
|
||||
{ \
|
||||
__device__ __forceinline__ double operator ()(double v) const \
|
||||
{ \
|
||||
return func(v); \
|
||||
} \
|
||||
__host__ __device__ __forceinline__ name ## _func() {} \
|
||||
__host__ __device__ __forceinline__ name ## _func(const name ## _func&) {} \
|
||||
};
|
||||
|
||||
#define OPENCV_CUDA_IMPLEMENT_BIN_FUNCTOR(name, func) \
|
||||
template <typename T> struct name ## _func : binary_function<T, T, float> \
|
||||
{ \
|
||||
__device__ __forceinline__ float operator ()(typename TypeTraits<T>::ParameterType v1, typename TypeTraits<T>::ParameterType v2) const \
|
||||
{ \
|
||||
return func ## f(v1, v2); \
|
||||
} \
|
||||
__host__ __device__ __forceinline__ name ## _func() {} \
|
||||
__host__ __device__ __forceinline__ name ## _func(const name ## _func&) {} \
|
||||
}; \
|
||||
template <> struct name ## _func<double> : binary_function<double, double, double> \
|
||||
{ \
|
||||
__device__ __forceinline__ double operator ()(double v1, double v2) const \
|
||||
{ \
|
||||
return func(v1, v2); \
|
||||
} \
|
||||
__host__ __device__ __forceinline__ name ## _func() {} \
|
||||
__host__ __device__ __forceinline__ name ## _func(const name ## _func&) {} \
|
||||
};
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(sqrt, ::sqrt)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(exp, ::exp)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(exp2, ::exp2)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(exp10, ::exp10)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(log, ::log)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(log2, ::log2)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(log10, ::log10)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(sin, ::sin)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(cos, ::cos)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(tan, ::tan)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(asin, ::asin)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(acos, ::acos)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(atan, ::atan)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(sinh, ::sinh)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(cosh, ::cosh)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(tanh, ::tanh)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(asinh, ::asinh)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(acosh, ::acosh)
|
||||
OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR(atanh, ::atanh)
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_BIN_FUNCTOR(hypot, ::hypot)
|
||||
OPENCV_CUDA_IMPLEMENT_BIN_FUNCTOR(atan2, ::atan2)
|
||||
OPENCV_CUDA_IMPLEMENT_BIN_FUNCTOR(pow, ::pow)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR
|
||||
#undef OPENCV_CUDA_IMPLEMENT_UN_FUNCTOR_NO_DOUBLE
|
||||
#undef OPENCV_CUDA_IMPLEMENT_BIN_FUNCTOR
|
||||
|
||||
template<typename T> struct hypot_sqr_func : binary_function<T, T, float>
|
||||
{
|
||||
__device__ __forceinline__ T operator ()(typename TypeTraits<T>::ParameterType src1, typename TypeTraits<T>::ParameterType src2) const
|
||||
{
|
||||
return src1 * src1 + src2 * src2;
|
||||
}
|
||||
__host__ __device__ __forceinline__ hypot_sqr_func() {}
|
||||
__host__ __device__ __forceinline__ hypot_sqr_func(const hypot_sqr_func&) {}
|
||||
};
|
||||
|
||||
// Saturate Cast Functor
|
||||
template <typename T, typename D> struct saturate_cast_func : unary_function<T, D>
|
||||
{
|
||||
__device__ __forceinline__ D operator ()(typename TypeTraits<T>::ParameterType v) const
|
||||
{
|
||||
return saturate_cast<D>(v);
|
||||
}
|
||||
__host__ __device__ __forceinline__ saturate_cast_func() {}
|
||||
__host__ __device__ __forceinline__ saturate_cast_func(const saturate_cast_func&) {}
|
||||
};
|
||||
|
||||
// Threshold Functors
|
||||
template <typename T> struct thresh_binary_func : unary_function<T, T>
|
||||
{
|
||||
__host__ __device__ __forceinline__ thresh_binary_func(T thresh_, T maxVal_) : thresh(thresh_), maxVal(maxVal_) {}
|
||||
|
||||
__device__ __forceinline__ T operator()(typename TypeTraits<T>::ParameterType src) const
|
||||
{
|
||||
return (src > thresh) * maxVal;
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ thresh_binary_func() {}
|
||||
__host__ __device__ __forceinline__ thresh_binary_func(const thresh_binary_func& other)
|
||||
: thresh(other.thresh), maxVal(other.maxVal) {}
|
||||
|
||||
T thresh;
|
||||
T maxVal;
|
||||
};
|
||||
|
||||
template <typename T> struct thresh_binary_inv_func : unary_function<T, T>
|
||||
{
|
||||
__host__ __device__ __forceinline__ thresh_binary_inv_func(T thresh_, T maxVal_) : thresh(thresh_), maxVal(maxVal_) {}
|
||||
|
||||
__device__ __forceinline__ T operator()(typename TypeTraits<T>::ParameterType src) const
|
||||
{
|
||||
return (src <= thresh) * maxVal;
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ thresh_binary_inv_func() {}
|
||||
__host__ __device__ __forceinline__ thresh_binary_inv_func(const thresh_binary_inv_func& other)
|
||||
: thresh(other.thresh), maxVal(other.maxVal) {}
|
||||
|
||||
T thresh;
|
||||
T maxVal;
|
||||
};
|
||||
|
||||
template <typename T> struct thresh_trunc_func : unary_function<T, T>
|
||||
{
|
||||
explicit __host__ __device__ __forceinline__ thresh_trunc_func(T thresh_, T maxVal_ = 0) : thresh(thresh_) {CV_UNUSED(maxVal_);}
|
||||
|
||||
__device__ __forceinline__ T operator()(typename TypeTraits<T>::ParameterType src) const
|
||||
{
|
||||
return minimum<T>()(src, thresh);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ thresh_trunc_func() {}
|
||||
__host__ __device__ __forceinline__ thresh_trunc_func(const thresh_trunc_func& other)
|
||||
: thresh(other.thresh) {}
|
||||
|
||||
T thresh;
|
||||
};
|
||||
|
||||
template <typename T> struct thresh_to_zero_func : unary_function<T, T>
|
||||
{
|
||||
explicit __host__ __device__ __forceinline__ thresh_to_zero_func(T thresh_, T maxVal_ = 0) : thresh(thresh_) {CV_UNUSED(maxVal_);}
|
||||
|
||||
__device__ __forceinline__ T operator()(typename TypeTraits<T>::ParameterType src) const
|
||||
{
|
||||
return (src > thresh) * src;
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ thresh_to_zero_func() {}
|
||||
__host__ __device__ __forceinline__ thresh_to_zero_func(const thresh_to_zero_func& other)
|
||||
: thresh(other.thresh) {}
|
||||
|
||||
T thresh;
|
||||
};
|
||||
|
||||
template <typename T> struct thresh_to_zero_inv_func : unary_function<T, T>
|
||||
{
|
||||
explicit __host__ __device__ __forceinline__ thresh_to_zero_inv_func(T thresh_, T maxVal_ = 0) : thresh(thresh_) {CV_UNUSED(maxVal_);}
|
||||
|
||||
__device__ __forceinline__ T operator()(typename TypeTraits<T>::ParameterType src) const
|
||||
{
|
||||
return (src <= thresh) * src;
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ thresh_to_zero_inv_func() {}
|
||||
__host__ __device__ __forceinline__ thresh_to_zero_inv_func(const thresh_to_zero_inv_func& other)
|
||||
: thresh(other.thresh) {}
|
||||
|
||||
T thresh;
|
||||
};
|
||||
|
||||
// Function Object Adaptors
|
||||
template <typename Predicate> struct unary_negate : unary_function<typename Predicate::argument_type, bool>
|
||||
{
|
||||
explicit __host__ __device__ __forceinline__ unary_negate(const Predicate& p) : pred(p) {}
|
||||
|
||||
__device__ __forceinline__ bool operator()(typename TypeTraits<typename Predicate::argument_type>::ParameterType x) const
|
||||
{
|
||||
return !pred(x);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ unary_negate() {}
|
||||
__host__ __device__ __forceinline__ unary_negate(const unary_negate& other) : pred(other.pred) {}
|
||||
|
||||
Predicate pred;
|
||||
};
|
||||
|
||||
template <typename Predicate> __host__ __device__ __forceinline__ unary_negate<Predicate> not1(const Predicate& pred)
|
||||
{
|
||||
return unary_negate<Predicate>(pred);
|
||||
}
|
||||
|
||||
template <typename Predicate> struct binary_negate : binary_function<typename Predicate::first_argument_type, typename Predicate::second_argument_type, bool>
|
||||
{
|
||||
explicit __host__ __device__ __forceinline__ binary_negate(const Predicate& p) : pred(p) {}
|
||||
|
||||
__device__ __forceinline__ bool operator()(typename TypeTraits<typename Predicate::first_argument_type>::ParameterType x,
|
||||
typename TypeTraits<typename Predicate::second_argument_type>::ParameterType y) const
|
||||
{
|
||||
return !pred(x,y);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ binary_negate() {}
|
||||
__host__ __device__ __forceinline__ binary_negate(const binary_negate& other) : pred(other.pred) {}
|
||||
|
||||
Predicate pred;
|
||||
};
|
||||
|
||||
template <typename BinaryPredicate> __host__ __device__ __forceinline__ binary_negate<BinaryPredicate> not2(const BinaryPredicate& pred)
|
||||
{
|
||||
return binary_negate<BinaryPredicate>(pred);
|
||||
}
|
||||
|
||||
template <typename Op> struct binder1st : unary_function<typename Op::second_argument_type, typename Op::result_type>
|
||||
{
|
||||
__host__ __device__ __forceinline__ binder1st(const Op& op_, const typename Op::first_argument_type& arg1_) : op(op_), arg1(arg1_) {}
|
||||
|
||||
__device__ __forceinline__ typename Op::result_type operator ()(typename TypeTraits<typename Op::second_argument_type>::ParameterType a) const
|
||||
{
|
||||
return op(arg1, a);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ binder1st() {}
|
||||
__host__ __device__ __forceinline__ binder1st(const binder1st& other) : op(other.op), arg1(other.arg1) {}
|
||||
|
||||
Op op;
|
||||
typename Op::first_argument_type arg1;
|
||||
};
|
||||
|
||||
template <typename Op, typename T> __host__ __device__ __forceinline__ binder1st<Op> bind1st(const Op& op, const T& x)
|
||||
{
|
||||
return binder1st<Op>(op, typename Op::first_argument_type(x));
|
||||
}
|
||||
|
||||
template <typename Op> struct binder2nd : unary_function<typename Op::first_argument_type, typename Op::result_type>
|
||||
{
|
||||
__host__ __device__ __forceinline__ binder2nd(const Op& op_, const typename Op::second_argument_type& arg2_) : op(op_), arg2(arg2_) {}
|
||||
|
||||
__forceinline__ __device__ typename Op::result_type operator ()(typename TypeTraits<typename Op::first_argument_type>::ParameterType a) const
|
||||
{
|
||||
return op(a, arg2);
|
||||
}
|
||||
|
||||
__host__ __device__ __forceinline__ binder2nd() {}
|
||||
__host__ __device__ __forceinline__ binder2nd(const binder2nd& other) : op(other.op), arg2(other.arg2) {}
|
||||
|
||||
Op op;
|
||||
typename Op::second_argument_type arg2;
|
||||
};
|
||||
|
||||
template <typename Op, typename T> __host__ __device__ __forceinline__ binder2nd<Op> bind2nd(const Op& op, const T& x)
|
||||
{
|
||||
return binder2nd<Op>(op, typename Op::second_argument_type(x));
|
||||
}
|
||||
|
||||
// Functor Traits
|
||||
template <typename F> struct IsUnaryFunction
|
||||
{
|
||||
typedef char Yes;
|
||||
struct No {Yes a[2];};
|
||||
|
||||
template <typename T, typename D> static Yes check(unary_function<T, D>);
|
||||
static No check(...);
|
||||
|
||||
static F makeF();
|
||||
|
||||
enum { value = (sizeof(check(makeF())) == sizeof(Yes)) };
|
||||
};
|
||||
|
||||
template <typename F> struct IsBinaryFunction
|
||||
{
|
||||
typedef char Yes;
|
||||
struct No {Yes a[2];};
|
||||
|
||||
template <typename T1, typename T2, typename D> static Yes check(binary_function<T1, T2, D>);
|
||||
static No check(...);
|
||||
|
||||
static F makeF();
|
||||
|
||||
enum { value = (sizeof(check(makeF())) == sizeof(Yes)) };
|
||||
};
|
||||
|
||||
namespace functional_detail
|
||||
{
|
||||
template <size_t src_elem_size, size_t dst_elem_size> struct UnOpShift { enum { shift = 1 }; };
|
||||
template <size_t src_elem_size> struct UnOpShift<src_elem_size, 1> { enum { shift = 4 }; };
|
||||
template <size_t src_elem_size> struct UnOpShift<src_elem_size, 2> { enum { shift = 2 }; };
|
||||
|
||||
template <typename T, typename D> struct DefaultUnaryShift
|
||||
{
|
||||
enum { shift = UnOpShift<sizeof(T), sizeof(D)>::shift };
|
||||
};
|
||||
|
||||
template <size_t src_elem_size1, size_t src_elem_size2, size_t dst_elem_size> struct BinOpShift { enum { shift = 1 }; };
|
||||
template <size_t src_elem_size1, size_t src_elem_size2> struct BinOpShift<src_elem_size1, src_elem_size2, 1> { enum { shift = 4 }; };
|
||||
template <size_t src_elem_size1, size_t src_elem_size2> struct BinOpShift<src_elem_size1, src_elem_size2, 2> { enum { shift = 2 }; };
|
||||
|
||||
template <typename T1, typename T2, typename D> struct DefaultBinaryShift
|
||||
{
|
||||
enum { shift = BinOpShift<sizeof(T1), sizeof(T2), sizeof(D)>::shift };
|
||||
};
|
||||
|
||||
template <typename Func, bool unary = IsUnaryFunction<Func>::value> struct ShiftDispatcher;
|
||||
template <typename Func> struct ShiftDispatcher<Func, true>
|
||||
{
|
||||
enum { shift = DefaultUnaryShift<typename Func::argument_type, typename Func::result_type>::shift };
|
||||
};
|
||||
template <typename Func> struct ShiftDispatcher<Func, false>
|
||||
{
|
||||
enum { shift = DefaultBinaryShift<typename Func::first_argument_type, typename Func::second_argument_type, typename Func::result_type>::shift };
|
||||
};
|
||||
}
|
||||
|
||||
template <typename Func> struct DefaultTransformShift
|
||||
{
|
||||
enum { shift = functional_detail::ShiftDispatcher<Func>::shift };
|
||||
};
|
||||
|
||||
template <typename Func> struct DefaultTransformFunctorTraits
|
||||
{
|
||||
enum { simple_block_dim_x = 16 };
|
||||
enum { simple_block_dim_y = 16 };
|
||||
|
||||
enum { smart_block_dim_x = 16 };
|
||||
enum { smart_block_dim_y = 16 };
|
||||
enum { smart_shift = DefaultTransformShift<Func>::shift };
|
||||
};
|
||||
|
||||
template <typename Func> struct TransformFunctorTraits : DefaultTransformFunctorTraits<Func> {};
|
||||
|
||||
#define OPENCV_CUDA_TRANSFORM_FUNCTOR_TRAITS(type) \
|
||||
template <> struct TransformFunctorTraits< type > : DefaultTransformFunctorTraits< type >
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_FUNCTIONAL_HPP
|
||||
|
|
@ -0,0 +1,128 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_LIMITS_HPP
|
||||
#define OPENCV_CUDA_LIMITS_HPP
|
||||
|
||||
#include <limits.h>
|
||||
#include <float.h>
|
||||
#include "common.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template <class T> struct numeric_limits;
|
||||
|
||||
template <> struct numeric_limits<bool>
|
||||
{
|
||||
__device__ __forceinline__ static bool min() { return false; }
|
||||
__device__ __forceinline__ static bool max() { return true; }
|
||||
static const bool is_signed = false;
|
||||
};
|
||||
|
||||
template <> struct numeric_limits<signed char>
|
||||
{
|
||||
__device__ __forceinline__ static signed char min() { return SCHAR_MIN; }
|
||||
__device__ __forceinline__ static signed char max() { return SCHAR_MAX; }
|
||||
static const bool is_signed = true;
|
||||
};
|
||||
|
||||
template <> struct numeric_limits<unsigned char>
|
||||
{
|
||||
__device__ __forceinline__ static unsigned char min() { return 0; }
|
||||
__device__ __forceinline__ static unsigned char max() { return UCHAR_MAX; }
|
||||
static const bool is_signed = false;
|
||||
};
|
||||
|
||||
template <> struct numeric_limits<short>
|
||||
{
|
||||
__device__ __forceinline__ static short min() { return SHRT_MIN; }
|
||||
__device__ __forceinline__ static short max() { return SHRT_MAX; }
|
||||
static const bool is_signed = true;
|
||||
};
|
||||
|
||||
template <> struct numeric_limits<unsigned short>
|
||||
{
|
||||
__device__ __forceinline__ static unsigned short min() { return 0; }
|
||||
__device__ __forceinline__ static unsigned short max() { return USHRT_MAX; }
|
||||
static const bool is_signed = false;
|
||||
};
|
||||
|
||||
template <> struct numeric_limits<int>
|
||||
{
|
||||
__device__ __forceinline__ static int min() { return INT_MIN; }
|
||||
__device__ __forceinline__ static int max() { return INT_MAX; }
|
||||
static const bool is_signed = true;
|
||||
};
|
||||
|
||||
template <> struct numeric_limits<unsigned int>
|
||||
{
|
||||
__device__ __forceinline__ static unsigned int min() { return 0; }
|
||||
__device__ __forceinline__ static unsigned int max() { return UINT_MAX; }
|
||||
static const bool is_signed = false;
|
||||
};
|
||||
|
||||
template <> struct numeric_limits<float>
|
||||
{
|
||||
__device__ __forceinline__ static float min() { return FLT_MIN; }
|
||||
__device__ __forceinline__ static float max() { return FLT_MAX; }
|
||||
__device__ __forceinline__ static float epsilon() { return FLT_EPSILON; }
|
||||
static const bool is_signed = true;
|
||||
};
|
||||
|
||||
template <> struct numeric_limits<double>
|
||||
{
|
||||
__device__ __forceinline__ static double min() { return DBL_MIN; }
|
||||
__device__ __forceinline__ static double max() { return DBL_MAX; }
|
||||
__device__ __forceinline__ static double epsilon() { return DBL_EPSILON; }
|
||||
static const bool is_signed = true;
|
||||
};
|
||||
}}} // namespace cv { namespace cuda { namespace cudev {
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_LIMITS_HPP
|
||||
|
|
@ -0,0 +1,209 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_REDUCE_HPP
|
||||
#define OPENCV_CUDA_REDUCE_HPP
|
||||
|
||||
#ifndef THRUST_DEBUG // eliminate -Wundef warning
|
||||
#define THRUST_DEBUG 0
|
||||
#endif
|
||||
|
||||
#include <thrust/tuple.h>
|
||||
#include "detail/reduce.hpp"
|
||||
#include "detail/reduce_key_val.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template <int N, typename T, class Op>
|
||||
__device__ __forceinline__ void reduce(volatile T* smem, T& val, unsigned int tid, const Op& op)
|
||||
{
|
||||
reduce_detail::Dispatcher<N>::reductor::template reduce<volatile T*, T&, const Op&>(smem, val, tid, op);
|
||||
}
|
||||
template <int N,
|
||||
typename P0, typename P1, typename P2, typename P3, typename P4, typename P5, typename P6, typename P7, typename P8, typename P9,
|
||||
typename R0, typename R1, typename R2, typename R3, typename R4, typename R5, typename R6, typename R7, typename R8, typename R9,
|
||||
class Op0, class Op1, class Op2, class Op3, class Op4, class Op5, class Op6, class Op7, class Op8, class Op9>
|
||||
__device__ __forceinline__ void reduce(const thrust::tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9>& smem,
|
||||
const thrust::tuple<R0, R1, R2, R3, R4, R5, R6, R7, R8, R9>& val,
|
||||
unsigned int tid,
|
||||
const thrust::tuple<Op0, Op1, Op2, Op3, Op4, Op5, Op6, Op7, Op8, Op9>& op)
|
||||
{
|
||||
reduce_detail::Dispatcher<N>::reductor::template reduce<
|
||||
const thrust::tuple<P0, P1, P2, P3, P4, P5, P6, P7, P8, P9>&,
|
||||
const thrust::tuple<R0, R1, R2, R3, R4, R5, R6, R7, R8, R9>&,
|
||||
const thrust::tuple<Op0, Op1, Op2, Op3, Op4, Op5, Op6, Op7, Op8, Op9>&>(smem, val, tid, op);
|
||||
}
|
||||
|
||||
template <unsigned int N, typename K, typename V, class Cmp>
|
||||
__device__ __forceinline__ void reduceKeyVal(volatile K* skeys, K& key, volatile V* svals, V& val, unsigned int tid, const Cmp& cmp)
|
||||
{
|
||||
reduce_key_val_detail::Dispatcher<N>::reductor::template reduce<volatile K*, K&, volatile V*, V&, const Cmp&>(skeys, key, svals, val, tid, cmp);
|
||||
}
|
||||
template <unsigned int N,
|
||||
typename K,
|
||||
typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
||||
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9,
|
||||
class Cmp>
|
||||
__device__ __forceinline__ void reduceKeyVal(volatile K* skeys, K& key,
|
||||
const thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& svals,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
||||
unsigned int tid, const Cmp& cmp)
|
||||
{
|
||||
reduce_key_val_detail::Dispatcher<N>::reductor::template reduce<volatile K*, K&,
|
||||
const thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>&,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>&,
|
||||
const Cmp&>(skeys, key, svals, val, tid, cmp);
|
||||
}
|
||||
template <unsigned int N,
|
||||
typename KP0, typename KP1, typename KP2, typename KP3, typename KP4, typename KP5, typename KP6, typename KP7, typename KP8, typename KP9,
|
||||
typename KR0, typename KR1, typename KR2, typename KR3, typename KR4, typename KR5, typename KR6, typename KR7, typename KR8, typename KR9,
|
||||
typename VP0, typename VP1, typename VP2, typename VP3, typename VP4, typename VP5, typename VP6, typename VP7, typename VP8, typename VP9,
|
||||
typename VR0, typename VR1, typename VR2, typename VR3, typename VR4, typename VR5, typename VR6, typename VR7, typename VR8, typename VR9,
|
||||
class Cmp0, class Cmp1, class Cmp2, class Cmp3, class Cmp4, class Cmp5, class Cmp6, class Cmp7, class Cmp8, class Cmp9>
|
||||
__device__ __forceinline__ void reduceKeyVal(const thrust::tuple<KP0, KP1, KP2, KP3, KP4, KP5, KP6, KP7, KP8, KP9>& skeys,
|
||||
const thrust::tuple<KR0, KR1, KR2, KR3, KR4, KR5, KR6, KR7, KR8, KR9>& key,
|
||||
const thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>& svals,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>& val,
|
||||
unsigned int tid,
|
||||
const thrust::tuple<Cmp0, Cmp1, Cmp2, Cmp3, Cmp4, Cmp5, Cmp6, Cmp7, Cmp8, Cmp9>& cmp)
|
||||
{
|
||||
reduce_key_val_detail::Dispatcher<N>::reductor::template reduce<
|
||||
const thrust::tuple<KP0, KP1, KP2, KP3, KP4, KP5, KP6, KP7, KP8, KP9>&,
|
||||
const thrust::tuple<KR0, KR1, KR2, KR3, KR4, KR5, KR6, KR7, KR8, KR9>&,
|
||||
const thrust::tuple<VP0, VP1, VP2, VP3, VP4, VP5, VP6, VP7, VP8, VP9>&,
|
||||
const thrust::tuple<VR0, VR1, VR2, VR3, VR4, VR5, VR6, VR7, VR8, VR9>&,
|
||||
const thrust::tuple<Cmp0, Cmp1, Cmp2, Cmp3, Cmp4, Cmp5, Cmp6, Cmp7, Cmp8, Cmp9>&
|
||||
>(skeys, key, svals, val, tid, cmp);
|
||||
}
|
||||
|
||||
// smem_tuple
|
||||
|
||||
template <typename T0>
|
||||
__device__ __forceinline__
|
||||
thrust::tuple<volatile T0*>
|
||||
smem_tuple(T0* t0)
|
||||
{
|
||||
return thrust::make_tuple((volatile T0*) t0);
|
||||
}
|
||||
|
||||
template <typename T0, typename T1>
|
||||
__device__ __forceinline__
|
||||
thrust::tuple<volatile T0*, volatile T1*>
|
||||
smem_tuple(T0* t0, T1* t1)
|
||||
{
|
||||
return thrust::make_tuple((volatile T0*) t0, (volatile T1*) t1);
|
||||
}
|
||||
|
||||
template <typename T0, typename T1, typename T2>
|
||||
__device__ __forceinline__
|
||||
thrust::tuple<volatile T0*, volatile T1*, volatile T2*>
|
||||
smem_tuple(T0* t0, T1* t1, T2* t2)
|
||||
{
|
||||
return thrust::make_tuple((volatile T0*) t0, (volatile T1*) t1, (volatile T2*) t2);
|
||||
}
|
||||
|
||||
template <typename T0, typename T1, typename T2, typename T3>
|
||||
__device__ __forceinline__
|
||||
thrust::tuple<volatile T0*, volatile T1*, volatile T2*, volatile T3*>
|
||||
smem_tuple(T0* t0, T1* t1, T2* t2, T3* t3)
|
||||
{
|
||||
return thrust::make_tuple((volatile T0*) t0, (volatile T1*) t1, (volatile T2*) t2, (volatile T3*) t3);
|
||||
}
|
||||
|
||||
template <typename T0, typename T1, typename T2, typename T3, typename T4>
|
||||
__device__ __forceinline__
|
||||
thrust::tuple<volatile T0*, volatile T1*, volatile T2*, volatile T3*, volatile T4*>
|
||||
smem_tuple(T0* t0, T1* t1, T2* t2, T3* t3, T4* t4)
|
||||
{
|
||||
return thrust::make_tuple((volatile T0*) t0, (volatile T1*) t1, (volatile T2*) t2, (volatile T3*) t3, (volatile T4*) t4);
|
||||
}
|
||||
|
||||
template <typename T0, typename T1, typename T2, typename T3, typename T4, typename T5>
|
||||
__device__ __forceinline__
|
||||
thrust::tuple<volatile T0*, volatile T1*, volatile T2*, volatile T3*, volatile T4*, volatile T5*>
|
||||
smem_tuple(T0* t0, T1* t1, T2* t2, T3* t3, T4* t4, T5* t5)
|
||||
{
|
||||
return thrust::make_tuple((volatile T0*) t0, (volatile T1*) t1, (volatile T2*) t2, (volatile T3*) t3, (volatile T4*) t4, (volatile T5*) t5);
|
||||
}
|
||||
|
||||
template <typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6>
|
||||
__device__ __forceinline__
|
||||
thrust::tuple<volatile T0*, volatile T1*, volatile T2*, volatile T3*, volatile T4*, volatile T5*, volatile T6*>
|
||||
smem_tuple(T0* t0, T1* t1, T2* t2, T3* t3, T4* t4, T5* t5, T6* t6)
|
||||
{
|
||||
return thrust::make_tuple((volatile T0*) t0, (volatile T1*) t1, (volatile T2*) t2, (volatile T3*) t3, (volatile T4*) t4, (volatile T5*) t5, (volatile T6*) t6);
|
||||
}
|
||||
|
||||
template <typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7>
|
||||
__device__ __forceinline__
|
||||
thrust::tuple<volatile T0*, volatile T1*, volatile T2*, volatile T3*, volatile T4*, volatile T5*, volatile T6*, volatile T7*>
|
||||
smem_tuple(T0* t0, T1* t1, T2* t2, T3* t3, T4* t4, T5* t5, T6* t6, T7* t7)
|
||||
{
|
||||
return thrust::make_tuple((volatile T0*) t0, (volatile T1*) t1, (volatile T2*) t2, (volatile T3*) t3, (volatile T4*) t4, (volatile T5*) t5, (volatile T6*) t6, (volatile T7*) t7);
|
||||
}
|
||||
|
||||
template <typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8>
|
||||
__device__ __forceinline__
|
||||
thrust::tuple<volatile T0*, volatile T1*, volatile T2*, volatile T3*, volatile T4*, volatile T5*, volatile T6*, volatile T7*, volatile T8*>
|
||||
smem_tuple(T0* t0, T1* t1, T2* t2, T3* t3, T4* t4, T5* t5, T6* t6, T7* t7, T8* t8)
|
||||
{
|
||||
return thrust::make_tuple((volatile T0*) t0, (volatile T1*) t1, (volatile T2*) t2, (volatile T3*) t3, (volatile T4*) t4, (volatile T5*) t5, (volatile T6*) t6, (volatile T7*) t7, (volatile T8*) t8);
|
||||
}
|
||||
|
||||
template <typename T0, typename T1, typename T2, typename T3, typename T4, typename T5, typename T6, typename T7, typename T8, typename T9>
|
||||
__device__ __forceinline__
|
||||
thrust::tuple<volatile T0*, volatile T1*, volatile T2*, volatile T3*, volatile T4*, volatile T5*, volatile T6*, volatile T7*, volatile T8*, volatile T9*>
|
||||
smem_tuple(T0* t0, T1* t1, T2* t2, T3* t3, T4* t4, T5* t5, T6* t6, T7* t7, T8* t8, T9* t9)
|
||||
{
|
||||
return thrust::make_tuple((volatile T0*) t0, (volatile T1*) t1, (volatile T2*) t2, (volatile T3*) t3, (volatile T4*) t4, (volatile T5*) t5, (volatile T6*) t6, (volatile T7*) t7, (volatile T8*) t8, (volatile T9*) t9);
|
||||
}
|
||||
}}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_REDUCE_HPP
|
||||
|
|
@ -0,0 +1,292 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_SATURATE_CAST_HPP
|
||||
#define OPENCV_CUDA_SATURATE_CAST_HPP
|
||||
|
||||
#include "common.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(uchar v) { return _Tp(v); }
|
||||
template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(schar v) { return _Tp(v); }
|
||||
template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(ushort v) { return _Tp(v); }
|
||||
template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(short v) { return _Tp(v); }
|
||||
template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(uint v) { return _Tp(v); }
|
||||
template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(int v) { return _Tp(v); }
|
||||
template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(float v) { return _Tp(v); }
|
||||
template<typename _Tp> __device__ __forceinline__ _Tp saturate_cast(double v) { return _Tp(v); }
|
||||
|
||||
template<> __device__ __forceinline__ uchar saturate_cast<uchar>(schar v)
|
||||
{
|
||||
uint res = 0;
|
||||
int vi = v;
|
||||
asm("cvt.sat.u8.s8 %0, %1;" : "=r"(res) : "r"(vi));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ uchar saturate_cast<uchar>(short v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.sat.u8.s16 %0, %1;" : "=r"(res) : "h"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ uchar saturate_cast<uchar>(ushort v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.sat.u8.u16 %0, %1;" : "=r"(res) : "h"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ uchar saturate_cast<uchar>(int v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.sat.u8.s32 %0, %1;" : "=r"(res) : "r"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ uchar saturate_cast<uchar>(uint v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.sat.u8.u32 %0, %1;" : "=r"(res) : "r"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ uchar saturate_cast<uchar>(float v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.rni.sat.u8.f32 %0, %1;" : "=r"(res) : "f"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ uchar saturate_cast<uchar>(double v)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 130
|
||||
uint res = 0;
|
||||
asm("cvt.rni.sat.u8.f64 %0, %1;" : "=r"(res) : "d"(v));
|
||||
return res;
|
||||
#else
|
||||
return saturate_cast<uchar>((float)v);
|
||||
#endif
|
||||
}
|
||||
|
||||
template<> __device__ __forceinline__ schar saturate_cast<schar>(uchar v)
|
||||
{
|
||||
uint res = 0;
|
||||
uint vi = v;
|
||||
asm("cvt.sat.s8.u8 %0, %1;" : "=r"(res) : "r"(vi));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ schar saturate_cast<schar>(short v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.sat.s8.s16 %0, %1;" : "=r"(res) : "h"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ schar saturate_cast<schar>(ushort v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.sat.s8.u16 %0, %1;" : "=r"(res) : "h"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ schar saturate_cast<schar>(int v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.sat.s8.s32 %0, %1;" : "=r"(res) : "r"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ schar saturate_cast<schar>(uint v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.sat.s8.u32 %0, %1;" : "=r"(res) : "r"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ schar saturate_cast<schar>(float v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.rni.sat.s8.f32 %0, %1;" : "=r"(res) : "f"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ schar saturate_cast<schar>(double v)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 130
|
||||
uint res = 0;
|
||||
asm("cvt.rni.sat.s8.f64 %0, %1;" : "=r"(res) : "d"(v));
|
||||
return res;
|
||||
#else
|
||||
return saturate_cast<schar>((float)v);
|
||||
#endif
|
||||
}
|
||||
|
||||
template<> __device__ __forceinline__ ushort saturate_cast<ushort>(schar v)
|
||||
{
|
||||
ushort res = 0;
|
||||
int vi = v;
|
||||
asm("cvt.sat.u16.s8 %0, %1;" : "=h"(res) : "r"(vi));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ ushort saturate_cast<ushort>(short v)
|
||||
{
|
||||
ushort res = 0;
|
||||
asm("cvt.sat.u16.s16 %0, %1;" : "=h"(res) : "h"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ ushort saturate_cast<ushort>(int v)
|
||||
{
|
||||
ushort res = 0;
|
||||
asm("cvt.sat.u16.s32 %0, %1;" : "=h"(res) : "r"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ ushort saturate_cast<ushort>(uint v)
|
||||
{
|
||||
ushort res = 0;
|
||||
asm("cvt.sat.u16.u32 %0, %1;" : "=h"(res) : "r"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ ushort saturate_cast<ushort>(float v)
|
||||
{
|
||||
ushort res = 0;
|
||||
asm("cvt.rni.sat.u16.f32 %0, %1;" : "=h"(res) : "f"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ ushort saturate_cast<ushort>(double v)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 130
|
||||
ushort res = 0;
|
||||
asm("cvt.rni.sat.u16.f64 %0, %1;" : "=h"(res) : "d"(v));
|
||||
return res;
|
||||
#else
|
||||
return saturate_cast<ushort>((float)v);
|
||||
#endif
|
||||
}
|
||||
|
||||
template<> __device__ __forceinline__ short saturate_cast<short>(ushort v)
|
||||
{
|
||||
short res = 0;
|
||||
asm("cvt.sat.s16.u16 %0, %1;" : "=h"(res) : "h"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ short saturate_cast<short>(int v)
|
||||
{
|
||||
short res = 0;
|
||||
asm("cvt.sat.s16.s32 %0, %1;" : "=h"(res) : "r"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ short saturate_cast<short>(uint v)
|
||||
{
|
||||
short res = 0;
|
||||
asm("cvt.sat.s16.u32 %0, %1;" : "=h"(res) : "r"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ short saturate_cast<short>(float v)
|
||||
{
|
||||
short res = 0;
|
||||
asm("cvt.rni.sat.s16.f32 %0, %1;" : "=h"(res) : "f"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ short saturate_cast<short>(double v)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 130
|
||||
short res = 0;
|
||||
asm("cvt.rni.sat.s16.f64 %0, %1;" : "=h"(res) : "d"(v));
|
||||
return res;
|
||||
#else
|
||||
return saturate_cast<short>((float)v);
|
||||
#endif
|
||||
}
|
||||
|
||||
template<> __device__ __forceinline__ int saturate_cast<int>(uint v)
|
||||
{
|
||||
int res = 0;
|
||||
asm("cvt.sat.s32.u32 %0, %1;" : "=r"(res) : "r"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ int saturate_cast<int>(float v)
|
||||
{
|
||||
return __float2int_rn(v);
|
||||
}
|
||||
template<> __device__ __forceinline__ int saturate_cast<int>(double v)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 130
|
||||
return __double2int_rn(v);
|
||||
#else
|
||||
return saturate_cast<int>((float)v);
|
||||
#endif
|
||||
}
|
||||
|
||||
template<> __device__ __forceinline__ uint saturate_cast<uint>(schar v)
|
||||
{
|
||||
uint res = 0;
|
||||
int vi = v;
|
||||
asm("cvt.sat.u32.s8 %0, %1;" : "=r"(res) : "r"(vi));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ uint saturate_cast<uint>(short v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.sat.u32.s16 %0, %1;" : "=r"(res) : "h"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ uint saturate_cast<uint>(int v)
|
||||
{
|
||||
uint res = 0;
|
||||
asm("cvt.sat.u32.s32 %0, %1;" : "=r"(res) : "r"(v));
|
||||
return res;
|
||||
}
|
||||
template<> __device__ __forceinline__ uint saturate_cast<uint>(float v)
|
||||
{
|
||||
return __float2uint_rn(v);
|
||||
}
|
||||
template<> __device__ __forceinline__ uint saturate_cast<uint>(double v)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 130
|
||||
return __double2uint_rn(v);
|
||||
#else
|
||||
return saturate_cast<uint>((float)v);
|
||||
#endif
|
||||
}
|
||||
}}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif /* OPENCV_CUDA_SATURATE_CAST_HPP */
|
||||
|
|
@ -0,0 +1,258 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_SCAN_HPP
|
||||
#define OPENCV_CUDA_SCAN_HPP
|
||||
|
||||
#include "opencv2/core/cuda/common.hpp"
|
||||
#include "opencv2/core/cuda/utility.hpp"
|
||||
#include "opencv2/core/cuda/warp.hpp"
|
||||
#include "opencv2/core/cuda/warp_shuffle.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
enum ScanKind { EXCLUSIVE = 0, INCLUSIVE = 1 };
|
||||
|
||||
template <ScanKind Kind, typename T, typename F> struct WarpScan
|
||||
{
|
||||
__device__ __forceinline__ WarpScan() {}
|
||||
__device__ __forceinline__ WarpScan(const WarpScan& other) { CV_UNUSED(other); }
|
||||
|
||||
__device__ __forceinline__ T operator()( volatile T *ptr , const unsigned int idx)
|
||||
{
|
||||
const unsigned int lane = idx & 31;
|
||||
F op;
|
||||
|
||||
if ( lane >= 1) ptr [idx ] = op(ptr [idx - 1], ptr [idx]);
|
||||
if ( lane >= 2) ptr [idx ] = op(ptr [idx - 2], ptr [idx]);
|
||||
if ( lane >= 4) ptr [idx ] = op(ptr [idx - 4], ptr [idx]);
|
||||
if ( lane >= 8) ptr [idx ] = op(ptr [idx - 8], ptr [idx]);
|
||||
if ( lane >= 16) ptr [idx ] = op(ptr [idx - 16], ptr [idx]);
|
||||
|
||||
if( Kind == INCLUSIVE )
|
||||
return ptr [idx];
|
||||
else
|
||||
return (lane > 0) ? ptr [idx - 1] : 0;
|
||||
}
|
||||
|
||||
__device__ __forceinline__ unsigned int index(const unsigned int tid)
|
||||
{
|
||||
return tid;
|
||||
}
|
||||
|
||||
__device__ __forceinline__ void init(volatile T *ptr){}
|
||||
|
||||
static const int warp_offset = 0;
|
||||
|
||||
typedef WarpScan<INCLUSIVE, T, F> merge;
|
||||
};
|
||||
|
||||
template <ScanKind Kind , typename T, typename F> struct WarpScanNoComp
|
||||
{
|
||||
__device__ __forceinline__ WarpScanNoComp() {}
|
||||
__device__ __forceinline__ WarpScanNoComp(const WarpScanNoComp& other) { CV_UNUSED(other); }
|
||||
|
||||
__device__ __forceinline__ T operator()( volatile T *ptr , const unsigned int idx)
|
||||
{
|
||||
const unsigned int lane = threadIdx.x & 31;
|
||||
F op;
|
||||
|
||||
ptr [idx ] = op(ptr [idx - 1], ptr [idx]);
|
||||
ptr [idx ] = op(ptr [idx - 2], ptr [idx]);
|
||||
ptr [idx ] = op(ptr [idx - 4], ptr [idx]);
|
||||
ptr [idx ] = op(ptr [idx - 8], ptr [idx]);
|
||||
ptr [idx ] = op(ptr [idx - 16], ptr [idx]);
|
||||
|
||||
if( Kind == INCLUSIVE )
|
||||
return ptr [idx];
|
||||
else
|
||||
return (lane > 0) ? ptr [idx - 1] : 0;
|
||||
}
|
||||
|
||||
__device__ __forceinline__ unsigned int index(const unsigned int tid)
|
||||
{
|
||||
return (tid >> warp_log) * warp_smem_stride + 16 + (tid & warp_mask);
|
||||
}
|
||||
|
||||
__device__ __forceinline__ void init(volatile T *ptr)
|
||||
{
|
||||
ptr[threadIdx.x] = 0;
|
||||
}
|
||||
|
||||
static const int warp_smem_stride = 32 + 16 + 1;
|
||||
static const int warp_offset = 16;
|
||||
static const int warp_log = 5;
|
||||
static const int warp_mask = 31;
|
||||
|
||||
typedef WarpScanNoComp<INCLUSIVE, T, F> merge;
|
||||
};
|
||||
|
||||
template <ScanKind Kind , typename T, typename Sc, typename F> struct BlockScan
|
||||
{
|
||||
__device__ __forceinline__ BlockScan() {}
|
||||
__device__ __forceinline__ BlockScan(const BlockScan& other) { CV_UNUSED(other); }
|
||||
|
||||
__device__ __forceinline__ T operator()(volatile T *ptr)
|
||||
{
|
||||
const unsigned int tid = threadIdx.x;
|
||||
const unsigned int lane = tid & warp_mask;
|
||||
const unsigned int warp = tid >> warp_log;
|
||||
|
||||
Sc scan;
|
||||
typename Sc::merge merge_scan;
|
||||
const unsigned int idx = scan.index(tid);
|
||||
|
||||
T val = scan(ptr, idx);
|
||||
__syncthreads ();
|
||||
|
||||
if( warp == 0)
|
||||
scan.init(ptr);
|
||||
__syncthreads ();
|
||||
|
||||
if( lane == 31 )
|
||||
ptr [scan.warp_offset + warp ] = (Kind == INCLUSIVE) ? val : ptr [idx];
|
||||
__syncthreads ();
|
||||
|
||||
if( warp == 0 )
|
||||
merge_scan(ptr, idx);
|
||||
__syncthreads();
|
||||
|
||||
if ( warp > 0)
|
||||
val = ptr [scan.warp_offset + warp - 1] + val;
|
||||
__syncthreads ();
|
||||
|
||||
ptr[idx] = val;
|
||||
__syncthreads ();
|
||||
|
||||
return val ;
|
||||
}
|
||||
|
||||
static const int warp_log = 5;
|
||||
static const int warp_mask = 31;
|
||||
};
|
||||
|
||||
template <typename T>
|
||||
__device__ T warpScanInclusive(T idata, volatile T* s_Data, unsigned int tid)
|
||||
{
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
const unsigned int laneId = cv::cuda::device::Warp::laneId();
|
||||
|
||||
// scan on shuffl functions
|
||||
#pragma unroll
|
||||
for (int i = 1; i <= (OPENCV_CUDA_WARP_SIZE / 2); i *= 2)
|
||||
{
|
||||
const T n = cv::cuda::device::shfl_up(idata, i);
|
||||
if (laneId >= i)
|
||||
idata += n;
|
||||
}
|
||||
|
||||
return idata;
|
||||
#else
|
||||
unsigned int pos = 2 * tid - (tid & (OPENCV_CUDA_WARP_SIZE - 1));
|
||||
s_Data[pos] = 0;
|
||||
pos += OPENCV_CUDA_WARP_SIZE;
|
||||
s_Data[pos] = idata;
|
||||
|
||||
s_Data[pos] += s_Data[pos - 1];
|
||||
s_Data[pos] += s_Data[pos - 2];
|
||||
s_Data[pos] += s_Data[pos - 4];
|
||||
s_Data[pos] += s_Data[pos - 8];
|
||||
s_Data[pos] += s_Data[pos - 16];
|
||||
|
||||
return s_Data[pos];
|
||||
#endif
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
__device__ __forceinline__ T warpScanExclusive(T idata, volatile T* s_Data, unsigned int tid)
|
||||
{
|
||||
return warpScanInclusive(idata, s_Data, tid) - idata;
|
||||
}
|
||||
|
||||
template <int tiNumScanThreads, typename T>
|
||||
__device__ T blockScanInclusive(T idata, volatile T* s_Data, unsigned int tid)
|
||||
{
|
||||
if (tiNumScanThreads > OPENCV_CUDA_WARP_SIZE)
|
||||
{
|
||||
//Bottom-level inclusive warp scan
|
||||
T warpResult = warpScanInclusive(idata, s_Data, tid);
|
||||
|
||||
//Save top elements of each warp for exclusive warp scan
|
||||
//sync to wait for warp scans to complete (because s_Data is being overwritten)
|
||||
__syncthreads();
|
||||
if ((tid & (OPENCV_CUDA_WARP_SIZE - 1)) == (OPENCV_CUDA_WARP_SIZE - 1))
|
||||
{
|
||||
s_Data[tid >> OPENCV_CUDA_LOG_WARP_SIZE] = warpResult;
|
||||
}
|
||||
|
||||
//wait for warp scans to complete
|
||||
__syncthreads();
|
||||
|
||||
if (tid < (tiNumScanThreads / OPENCV_CUDA_WARP_SIZE) )
|
||||
{
|
||||
//grab top warp elements
|
||||
T val = s_Data[tid];
|
||||
//calculate exclusive scan and write back to shared memory
|
||||
s_Data[tid] = warpScanExclusive(val, s_Data, tid);
|
||||
}
|
||||
|
||||
//return updated warp scans with exclusive scan results
|
||||
__syncthreads();
|
||||
|
||||
return warpResult + s_Data[tid >> OPENCV_CUDA_LOG_WARP_SIZE];
|
||||
}
|
||||
else
|
||||
{
|
||||
return warpScanInclusive(idata, s_Data, tid);
|
||||
}
|
||||
}
|
||||
}}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_SCAN_HPP
|
||||
|
|
@ -0,0 +1,869 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
/*
|
||||
* Copyright (c) 2013 NVIDIA Corporation. All rights reserved.
|
||||
*
|
||||
* Redistribution and use in source and binary forms, with or without
|
||||
* modification, are permitted provided that the following conditions are met:
|
||||
*
|
||||
* Redistributions of source code must retain the above copyright notice,
|
||||
* this list of conditions and the following disclaimer.
|
||||
*
|
||||
* Redistributions in binary form must reproduce the above copyright notice,
|
||||
* this list of conditions and the following disclaimer in the documentation
|
||||
* and/or other materials provided with the distribution.
|
||||
*
|
||||
* Neither the name of NVIDIA Corporation nor the names of its contributors
|
||||
* may be used to endorse or promote products derived from this software
|
||||
* without specific prior written permission.
|
||||
*
|
||||
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
|
||||
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
||||
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
||||
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
|
||||
* LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
|
||||
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
|
||||
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
|
||||
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
|
||||
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
|
||||
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
|
||||
* POSSIBILITY OF SUCH DAMAGE.
|
||||
*/
|
||||
|
||||
#ifndef OPENCV_CUDA_SIMD_FUNCTIONS_HPP
|
||||
#define OPENCV_CUDA_SIMD_FUNCTIONS_HPP
|
||||
|
||||
#include "common.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
// 2
|
||||
|
||||
static __device__ __forceinline__ unsigned int vadd2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vadd2.u32.u32.u32.sat %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#elif __CUDA_ARCH__ >= 200
|
||||
asm("vadd.u32.u32.u32.sat %0.h0, %1.h0, %2.h0, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vadd.u32.u32.u32.sat %0.h1, %1.h1, %2.h1, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int s;
|
||||
s = a ^ b; // sum bits
|
||||
r = a + b; // actual sum
|
||||
s = s ^ r; // determine carry-ins for each bit position
|
||||
s = s & 0x00010000; // carry-in to high word (= carry-out from low word)
|
||||
r = r - s; // subtract out carry-out from low word
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsub2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vsub2.u32.u32.u32.sat %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#elif __CUDA_ARCH__ >= 200
|
||||
asm("vsub.u32.u32.u32.sat %0.h0, %1.h0, %2.h0, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vsub.u32.u32.u32.sat %0.h1, %1.h1, %2.h1, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int s;
|
||||
s = a ^ b; // sum bits
|
||||
r = a - b; // actual sum
|
||||
s = s ^ r; // determine carry-ins for each bit position
|
||||
s = s & 0x00010000; // borrow to high word
|
||||
r = r + s; // compensate for borrow from low word
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vabsdiff2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vabsdiff2.u32.u32.u32.sat %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#elif __CUDA_ARCH__ >= 200
|
||||
asm("vabsdiff.u32.u32.u32.sat %0.h0, %1.h0, %2.h0, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vabsdiff.u32.u32.u32.sat %0.h1, %1.h1, %2.h1, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int s, t, u, v;
|
||||
s = a & 0x0000ffff; // extract low halfword
|
||||
r = b & 0x0000ffff; // extract low halfword
|
||||
u = ::max(r, s); // maximum of low halfwords
|
||||
v = ::min(r, s); // minimum of low halfwords
|
||||
s = a & 0xffff0000; // extract high halfword
|
||||
r = b & 0xffff0000; // extract high halfword
|
||||
t = ::max(r, s); // maximum of high halfwords
|
||||
s = ::min(r, s); // minimum of high halfwords
|
||||
r = u | t; // maximum of both halfwords
|
||||
s = v | s; // minimum of both halfwords
|
||||
r = r - s; // |a - b| = max(a,b) - min(a,b);
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vavg2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, s;
|
||||
|
||||
// HAKMEM #23: a + b = 2 * (a & b) + (a ^ b) ==>
|
||||
// (a + b) / 2 = (a & b) + ((a ^ b) >> 1)
|
||||
s = a ^ b;
|
||||
r = a & b;
|
||||
s = s & 0xfffefffe; // ensure shift doesn't cross halfword boundaries
|
||||
s = s >> 1;
|
||||
s = r + s;
|
||||
|
||||
return s;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vavrg2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vavrg2.u32.u32.u32 %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
// HAKMEM #23: a + b = 2 * (a | b) - (a ^ b) ==>
|
||||
// (a + b + 1) / 2 = (a | b) - ((a ^ b) >> 1)
|
||||
unsigned int s;
|
||||
s = a ^ b;
|
||||
r = a | b;
|
||||
s = s & 0xfffefffe; // ensure shift doesn't cross half-word boundaries
|
||||
s = s >> 1;
|
||||
r = r - s;
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vseteq2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset2.u32.u32.eq %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
// inspired by Alan Mycroft's null-byte detection algorithm:
|
||||
// null_byte(x) = ((x - 0x01010101) & (~x & 0x80808080))
|
||||
unsigned int c;
|
||||
r = a ^ b; // 0x0000 if a == b
|
||||
c = r | 0x80008000; // set msbs, to catch carry out
|
||||
r = r ^ c; // extract msbs, msb = 1 if r < 0x8000
|
||||
c = c - 0x00010001; // msb = 0, if r was 0x0000 or 0x8000
|
||||
c = r & ~c; // msb = 1, if r was 0x0000
|
||||
r = c >> 15; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmpeq2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, c;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vseteq2(a, b);
|
||||
c = r << 16; // convert bool
|
||||
r = c - r; // into mask
|
||||
#else
|
||||
// inspired by Alan Mycroft's null-byte detection algorithm:
|
||||
// null_byte(x) = ((x - 0x01010101) & (~x & 0x80808080))
|
||||
r = a ^ b; // 0x0000 if a == b
|
||||
c = r | 0x80008000; // set msbs, to catch carry out
|
||||
r = r ^ c; // extract msbs, msb = 1 if r < 0x8000
|
||||
c = c - 0x00010001; // msb = 0, if r was 0x0000 or 0x8000
|
||||
c = r & ~c; // msb = 1, if r was 0x0000
|
||||
r = c >> 15; // convert
|
||||
r = c - r; // msbs to
|
||||
r = c | r; // mask
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsetge2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset2.u32.u32.ge %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int c;
|
||||
asm("not.b32 %0, %0;" : "+r"(b));
|
||||
c = vavrg2(a, b); // (a + ~b + 1) / 2 = (a - b) / 2
|
||||
c = c & 0x80008000; // msb = carry-outs
|
||||
r = c >> 15; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmpge2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, c;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vsetge2(a, b);
|
||||
c = r << 16; // convert bool
|
||||
r = c - r; // into mask
|
||||
#else
|
||||
asm("not.b32 %0, %0;" : "+r"(b));
|
||||
c = vavrg2(a, b); // (a + ~b + 1) / 2 = (a - b) / 2
|
||||
c = c & 0x80008000; // msb = carry-outs
|
||||
r = c >> 15; // convert
|
||||
r = c - r; // msbs to
|
||||
r = c | r; // mask
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsetgt2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset2.u32.u32.gt %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int c;
|
||||
asm("not.b32 %0, %0;" : "+r"(b));
|
||||
c = vavg2(a, b); // (a + ~b) / 2 = (a - b) / 2 [rounded down]
|
||||
c = c & 0x80008000; // msbs = carry-outs
|
||||
r = c >> 15; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmpgt2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, c;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vsetgt2(a, b);
|
||||
c = r << 16; // convert bool
|
||||
r = c - r; // into mask
|
||||
#else
|
||||
asm("not.b32 %0, %0;" : "+r"(b));
|
||||
c = vavg2(a, b); // (a + ~b) / 2 = (a - b) / 2 [rounded down]
|
||||
c = c & 0x80008000; // msbs = carry-outs
|
||||
r = c >> 15; // convert
|
||||
r = c - r; // msbs to
|
||||
r = c | r; // mask
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsetle2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset2.u32.u32.le %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int c;
|
||||
asm("not.b32 %0, %0;" : "+r"(a));
|
||||
c = vavrg2(a, b); // (b + ~a + 1) / 2 = (b - a) / 2
|
||||
c = c & 0x80008000; // msb = carry-outs
|
||||
r = c >> 15; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmple2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, c;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vsetle2(a, b);
|
||||
c = r << 16; // convert bool
|
||||
r = c - r; // into mask
|
||||
#else
|
||||
asm("not.b32 %0, %0;" : "+r"(a));
|
||||
c = vavrg2(a, b); // (b + ~a + 1) / 2 = (b - a) / 2
|
||||
c = c & 0x80008000; // msb = carry-outs
|
||||
r = c >> 15; // convert
|
||||
r = c - r; // msbs to
|
||||
r = c | r; // mask
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsetlt2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset2.u32.u32.lt %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int c;
|
||||
asm("not.b32 %0, %0;" : "+r"(a));
|
||||
c = vavg2(a, b); // (b + ~a) / 2 = (b - a) / 2 [rounded down]
|
||||
c = c & 0x80008000; // msb = carry-outs
|
||||
r = c >> 15; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmplt2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, c;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vsetlt2(a, b);
|
||||
c = r << 16; // convert bool
|
||||
r = c - r; // into mask
|
||||
#else
|
||||
asm("not.b32 %0, %0;" : "+r"(a));
|
||||
c = vavg2(a, b); // (b + ~a) / 2 = (b - a) / 2 [rounded down]
|
||||
c = c & 0x80008000; // msb = carry-outs
|
||||
r = c >> 15; // convert
|
||||
r = c - r; // msbs to
|
||||
r = c | r; // mask
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsetne2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm ("vset2.u32.u32.ne %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
// inspired by Alan Mycroft's null-byte detection algorithm:
|
||||
// null_byte(x) = ((x - 0x01010101) & (~x & 0x80808080))
|
||||
unsigned int c;
|
||||
r = a ^ b; // 0x0000 if a == b
|
||||
c = r | 0x80008000; // set msbs, to catch carry out
|
||||
c = c - 0x00010001; // msb = 0, if r was 0x0000 or 0x8000
|
||||
c = r | c; // msb = 1, if r was not 0x0000
|
||||
c = c & 0x80008000; // extract msbs
|
||||
r = c >> 15; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmpne2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, c;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vsetne2(a, b);
|
||||
c = r << 16; // convert bool
|
||||
r = c - r; // into mask
|
||||
#else
|
||||
// inspired by Alan Mycroft's null-byte detection algorithm:
|
||||
// null_byte(x) = ((x - 0x01010101) & (~x & 0x80808080))
|
||||
r = a ^ b; // 0x0000 if a == b
|
||||
c = r | 0x80008000; // set msbs, to catch carry out
|
||||
c = c - 0x00010001; // msb = 0, if r was 0x0000 or 0x8000
|
||||
c = r | c; // msb = 1, if r was not 0x0000
|
||||
c = c & 0x80008000; // extract msbs
|
||||
r = c >> 15; // convert
|
||||
r = c - r; // msbs to
|
||||
r = c | r; // mask
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vmax2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vmax2.u32.u32.u32 %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#elif __CUDA_ARCH__ >= 200
|
||||
asm("vmax.u32.u32.u32 %0.h0, %1.h0, %2.h0, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vmax.u32.u32.u32 %0.h1, %1.h1, %2.h1, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int s, t, u;
|
||||
r = a & 0x0000ffff; // extract low halfword
|
||||
s = b & 0x0000ffff; // extract low halfword
|
||||
t = ::max(r, s); // maximum of low halfwords
|
||||
r = a & 0xffff0000; // extract high halfword
|
||||
s = b & 0xffff0000; // extract high halfword
|
||||
u = ::max(r, s); // maximum of high halfwords
|
||||
r = t | u; // combine halfword maximums
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vmin2(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vmin2.u32.u32.u32 %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#elif __CUDA_ARCH__ >= 200
|
||||
asm("vmin.u32.u32.u32 %0.h0, %1.h0, %2.h0, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vmin.u32.u32.u32 %0.h1, %1.h1, %2.h1, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int s, t, u;
|
||||
r = a & 0x0000ffff; // extract low halfword
|
||||
s = b & 0x0000ffff; // extract low halfword
|
||||
t = ::min(r, s); // minimum of low halfwords
|
||||
r = a & 0xffff0000; // extract high halfword
|
||||
s = b & 0xffff0000; // extract high halfword
|
||||
u = ::min(r, s); // minimum of high halfwords
|
||||
r = t | u; // combine halfword minimums
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
// 4
|
||||
|
||||
static __device__ __forceinline__ unsigned int vadd4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vadd4.u32.u32.u32.sat %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#elif __CUDA_ARCH__ >= 200
|
||||
asm("vadd.u32.u32.u32.sat %0.b0, %1.b0, %2.b0, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vadd.u32.u32.u32.sat %0.b1, %1.b1, %2.b1, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vadd.u32.u32.u32.sat %0.b2, %1.b2, %2.b2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vadd.u32.u32.u32.sat %0.b3, %1.b3, %2.b3, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int s, t;
|
||||
s = a ^ b; // sum bits
|
||||
r = a & 0x7f7f7f7f; // clear msbs
|
||||
t = b & 0x7f7f7f7f; // clear msbs
|
||||
s = s & 0x80808080; // msb sum bits
|
||||
r = r + t; // add without msbs, record carry-out in msbs
|
||||
r = r ^ s; // sum of msb sum and carry-in bits, w/o carry-out
|
||||
#endif /* __CUDA_ARCH__ >= 300 */
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsub4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vsub4.u32.u32.u32.sat %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#elif __CUDA_ARCH__ >= 200
|
||||
asm("vsub.u32.u32.u32.sat %0.b0, %1.b0, %2.b0, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vsub.u32.u32.u32.sat %0.b1, %1.b1, %2.b1, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vsub.u32.u32.u32.sat %0.b2, %1.b2, %2.b2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vsub.u32.u32.u32.sat %0.b3, %1.b3, %2.b3, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int s, t;
|
||||
s = a ^ ~b; // inverted sum bits
|
||||
r = a | 0x80808080; // set msbs
|
||||
t = b & 0x7f7f7f7f; // clear msbs
|
||||
s = s & 0x80808080; // inverted msb sum bits
|
||||
r = r - t; // subtract w/o msbs, record inverted borrows in msb
|
||||
r = r ^ s; // combine inverted msb sum bits and borrows
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vavg4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, s;
|
||||
|
||||
// HAKMEM #23: a + b = 2 * (a & b) + (a ^ b) ==>
|
||||
// (a + b) / 2 = (a & b) + ((a ^ b) >> 1)
|
||||
s = a ^ b;
|
||||
r = a & b;
|
||||
s = s & 0xfefefefe; // ensure following shift doesn't cross byte boundaries
|
||||
s = s >> 1;
|
||||
s = r + s;
|
||||
|
||||
return s;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vavrg4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vavrg4.u32.u32.u32 %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
// HAKMEM #23: a + b = 2 * (a | b) - (a ^ b) ==>
|
||||
// (a + b + 1) / 2 = (a | b) - ((a ^ b) >> 1)
|
||||
unsigned int c;
|
||||
c = a ^ b;
|
||||
r = a | b;
|
||||
c = c & 0xfefefefe; // ensure following shift doesn't cross byte boundaries
|
||||
c = c >> 1;
|
||||
r = r - c;
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vseteq4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset4.u32.u32.eq %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
// inspired by Alan Mycroft's null-byte detection algorithm:
|
||||
// null_byte(x) = ((x - 0x01010101) & (~x & 0x80808080))
|
||||
unsigned int c;
|
||||
r = a ^ b; // 0x00 if a == b
|
||||
c = r | 0x80808080; // set msbs, to catch carry out
|
||||
r = r ^ c; // extract msbs, msb = 1 if r < 0x80
|
||||
c = c - 0x01010101; // msb = 0, if r was 0x00 or 0x80
|
||||
c = r & ~c; // msb = 1, if r was 0x00
|
||||
r = c >> 7; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmpeq4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, t;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vseteq4(a, b);
|
||||
t = r << 8; // convert bool
|
||||
r = t - r; // to mask
|
||||
#else
|
||||
// inspired by Alan Mycroft's null-byte detection algorithm:
|
||||
// null_byte(x) = ((x - 0x01010101) & (~x & 0x80808080))
|
||||
t = a ^ b; // 0x00 if a == b
|
||||
r = t | 0x80808080; // set msbs, to catch carry out
|
||||
t = t ^ r; // extract msbs, msb = 1 if t < 0x80
|
||||
r = r - 0x01010101; // msb = 0, if t was 0x00 or 0x80
|
||||
r = t & ~r; // msb = 1, if t was 0x00
|
||||
t = r >> 7; // build mask
|
||||
t = r - t; // from
|
||||
r = t | r; // msbs
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsetle4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset4.u32.u32.le %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int c;
|
||||
asm("not.b32 %0, %0;" : "+r"(a));
|
||||
c = vavrg4(a, b); // (b + ~a + 1) / 2 = (b - a) / 2
|
||||
c = c & 0x80808080; // msb = carry-outs
|
||||
r = c >> 7; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmple4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, c;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vsetle4(a, b);
|
||||
c = r << 8; // convert bool
|
||||
r = c - r; // to mask
|
||||
#else
|
||||
asm("not.b32 %0, %0;" : "+r"(a));
|
||||
c = vavrg4(a, b); // (b + ~a + 1) / 2 = (b - a) / 2
|
||||
c = c & 0x80808080; // msbs = carry-outs
|
||||
r = c >> 7; // convert
|
||||
r = c - r; // msbs to
|
||||
r = c | r; // mask
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsetlt4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset4.u32.u32.lt %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int c;
|
||||
asm("not.b32 %0, %0;" : "+r"(a));
|
||||
c = vavg4(a, b); // (b + ~a) / 2 = (b - a) / 2 [rounded down]
|
||||
c = c & 0x80808080; // msb = carry-outs
|
||||
r = c >> 7; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmplt4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, c;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vsetlt4(a, b);
|
||||
c = r << 8; // convert bool
|
||||
r = c - r; // to mask
|
||||
#else
|
||||
asm("not.b32 %0, %0;" : "+r"(a));
|
||||
c = vavg4(a, b); // (b + ~a) / 2 = (b - a) / 2 [rounded down]
|
||||
c = c & 0x80808080; // msbs = carry-outs
|
||||
r = c >> 7; // convert
|
||||
r = c - r; // msbs to
|
||||
r = c | r; // mask
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsetge4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset4.u32.u32.ge %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int c;
|
||||
asm("not.b32 %0, %0;" : "+r"(b));
|
||||
c = vavrg4(a, b); // (a + ~b + 1) / 2 = (a - b) / 2
|
||||
c = c & 0x80808080; // msb = carry-outs
|
||||
r = c >> 7; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmpge4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, s;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vsetge4(a, b);
|
||||
s = r << 8; // convert bool
|
||||
r = s - r; // to mask
|
||||
#else
|
||||
asm ("not.b32 %0,%0;" : "+r"(b));
|
||||
r = vavrg4 (a, b); // (a + ~b + 1) / 2 = (a - b) / 2
|
||||
r = r & 0x80808080; // msb = carry-outs
|
||||
s = r >> 7; // build mask
|
||||
s = r - s; // from
|
||||
r = s | r; // msbs
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsetgt4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset4.u32.u32.gt %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int c;
|
||||
asm("not.b32 %0, %0;" : "+r"(b));
|
||||
c = vavg4(a, b); // (a + ~b) / 2 = (a - b) / 2 [rounded down]
|
||||
c = c & 0x80808080; // msb = carry-outs
|
||||
r = c >> 7; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmpgt4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, c;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vsetgt4(a, b);
|
||||
c = r << 8; // convert bool
|
||||
r = c - r; // to mask
|
||||
#else
|
||||
asm("not.b32 %0, %0;" : "+r"(b));
|
||||
c = vavg4(a, b); // (a + ~b) / 2 = (a - b) / 2 [rounded down]
|
||||
c = c & 0x80808080; // msb = carry-outs
|
||||
r = c >> 7; // convert
|
||||
r = c - r; // msbs to
|
||||
r = c | r; // mask
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vsetne4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vset4.u32.u32.ne %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
// inspired by Alan Mycroft's null-byte detection algorithm:
|
||||
// null_byte(x) = ((x - 0x01010101) & (~x & 0x80808080))
|
||||
unsigned int c;
|
||||
r = a ^ b; // 0x00 if a == b
|
||||
c = r | 0x80808080; // set msbs, to catch carry out
|
||||
c = c - 0x01010101; // msb = 0, if r was 0x00 or 0x80
|
||||
c = r | c; // msb = 1, if r was not 0x00
|
||||
c = c & 0x80808080; // extract msbs
|
||||
r = c >> 7; // convert to bool
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vcmpne4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r, c;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
r = vsetne4(a, b);
|
||||
c = r << 8; // convert bool
|
||||
r = c - r; // to mask
|
||||
#else
|
||||
// inspired by Alan Mycroft's null-byte detection algorithm:
|
||||
// null_byte(x) = ((x - 0x01010101) & (~x & 0x80808080))
|
||||
r = a ^ b; // 0x00 if a == b
|
||||
c = r | 0x80808080; // set msbs, to catch carry out
|
||||
c = c - 0x01010101; // msb = 0, if r was 0x00 or 0x80
|
||||
c = r | c; // msb = 1, if r was not 0x00
|
||||
c = c & 0x80808080; // extract msbs
|
||||
r = c >> 7; // convert
|
||||
r = c - r; // msbs to
|
||||
r = c | r; // mask
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vabsdiff4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vabsdiff4.u32.u32.u32.sat %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#elif __CUDA_ARCH__ >= 200
|
||||
asm("vabsdiff.u32.u32.u32.sat %0.b0, %1.b0, %2.b0, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vabsdiff.u32.u32.u32.sat %0.b1, %1.b1, %2.b1, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vabsdiff.u32.u32.u32.sat %0.b2, %1.b2, %2.b2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vabsdiff.u32.u32.u32.sat %0.b3, %1.b3, %2.b3, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int s;
|
||||
s = vcmpge4(a, b); // mask = 0xff if a >= b
|
||||
r = a ^ b; //
|
||||
s = (r & s) ^ b; // select a when a >= b, else select b => max(a,b)
|
||||
r = s ^ r; // select a when b >= a, else select b => min(a,b)
|
||||
r = s - r; // |a - b| = max(a,b) - min(a,b);
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vmax4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vmax4.u32.u32.u32 %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#elif __CUDA_ARCH__ >= 200
|
||||
asm("vmax.u32.u32.u32 %0.b0, %1.b0, %2.b0, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vmax.u32.u32.u32 %0.b1, %1.b1, %2.b1, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vmax.u32.u32.u32 %0.b2, %1.b2, %2.b2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vmax.u32.u32.u32 %0.b3, %1.b3, %2.b3, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int s;
|
||||
s = vcmpge4(a, b); // mask = 0xff if a >= b
|
||||
r = a & s; // select a when b >= a
|
||||
s = b & ~s; // select b when b < a
|
||||
r = r | s; // combine byte selections
|
||||
#endif
|
||||
|
||||
return r; // byte-wise unsigned maximum
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ unsigned int vmin4(unsigned int a, unsigned int b)
|
||||
{
|
||||
unsigned int r = 0;
|
||||
|
||||
#if __CUDA_ARCH__ >= 300
|
||||
asm("vmin4.u32.u32.u32 %0, %1, %2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#elif __CUDA_ARCH__ >= 200
|
||||
asm("vmin.u32.u32.u32 %0.b0, %1.b0, %2.b0, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vmin.u32.u32.u32 %0.b1, %1.b1, %2.b1, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vmin.u32.u32.u32 %0.b2, %1.b2, %2.b2, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
asm("vmin.u32.u32.u32 %0.b3, %1.b3, %2.b3, %3;" : "=r"(r) : "r"(a), "r"(b), "r"(r));
|
||||
#else
|
||||
unsigned int s;
|
||||
s = vcmpge4(b, a); // mask = 0xff if a >= b
|
||||
r = a & s; // select a when b >= a
|
||||
s = b & ~s; // select b when b < a
|
||||
r = r | s; // combine byte selections
|
||||
#endif
|
||||
|
||||
return r;
|
||||
}
|
||||
}}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_SIMD_FUNCTIONS_HPP
|
||||
|
|
@ -0,0 +1,75 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_TRANSFORM_HPP
|
||||
#define OPENCV_CUDA_TRANSFORM_HPP
|
||||
|
||||
#include "common.hpp"
|
||||
#include "utility.hpp"
|
||||
#include "detail/transform_detail.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template <typename T, typename D, typename UnOp, typename Mask>
|
||||
static inline void transform(PtrStepSz<T> src, PtrStepSz<D> dst, UnOp op, const Mask& mask, cudaStream_t stream)
|
||||
{
|
||||
typedef TransformFunctorTraits<UnOp> ft;
|
||||
transform_detail::TransformDispatcher<VecTraits<T>::cn == 1 && VecTraits<D>::cn == 1 && ft::smart_shift != 1>::call(src, dst, op, mask, stream);
|
||||
}
|
||||
|
||||
template <typename T1, typename T2, typename D, typename BinOp, typename Mask>
|
||||
static inline void transform(PtrStepSz<T1> src1, PtrStepSz<T2> src2, PtrStepSz<D> dst, BinOp op, const Mask& mask, cudaStream_t stream)
|
||||
{
|
||||
typedef TransformFunctorTraits<BinOp> ft;
|
||||
transform_detail::TransformDispatcher<VecTraits<T1>::cn == 1 && VecTraits<T2>::cn == 1 && VecTraits<D>::cn == 1 && ft::smart_shift != 1>::call(src1, src2, dst, op, mask, stream);
|
||||
}
|
||||
}}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_TRANSFORM_HPP
|
||||
|
|
@ -0,0 +1,90 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_TYPE_TRAITS_HPP
|
||||
#define OPENCV_CUDA_TYPE_TRAITS_HPP
|
||||
|
||||
#include "detail/type_traits_detail.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template <typename T> struct IsSimpleParameter
|
||||
{
|
||||
enum {value = type_traits_detail::IsIntegral<T>::value || type_traits_detail::IsFloat<T>::value ||
|
||||
type_traits_detail::PointerTraits<typename type_traits_detail::ReferenceTraits<T>::type>::value};
|
||||
};
|
||||
|
||||
template <typename T> struct TypeTraits
|
||||
{
|
||||
typedef typename type_traits_detail::UnConst<T>::type NonConstType;
|
||||
typedef typename type_traits_detail::UnVolatile<T>::type NonVolatileType;
|
||||
typedef typename type_traits_detail::UnVolatile<typename type_traits_detail::UnConst<T>::type>::type UnqualifiedType;
|
||||
typedef typename type_traits_detail::PointerTraits<UnqualifiedType>::type PointeeType;
|
||||
typedef typename type_traits_detail::ReferenceTraits<T>::type ReferredType;
|
||||
|
||||
enum { isConst = type_traits_detail::UnConst<T>::value };
|
||||
enum { isVolatile = type_traits_detail::UnVolatile<T>::value };
|
||||
|
||||
enum { isReference = type_traits_detail::ReferenceTraits<UnqualifiedType>::value };
|
||||
enum { isPointer = type_traits_detail::PointerTraits<typename type_traits_detail::ReferenceTraits<UnqualifiedType>::type>::value };
|
||||
|
||||
enum { isUnsignedInt = type_traits_detail::IsUnsignedIntegral<UnqualifiedType>::value };
|
||||
enum { isSignedInt = type_traits_detail::IsSignedIntergral<UnqualifiedType>::value };
|
||||
enum { isIntegral = type_traits_detail::IsIntegral<UnqualifiedType>::value };
|
||||
enum { isFloat = type_traits_detail::IsFloat<UnqualifiedType>::value };
|
||||
enum { isArith = isIntegral || isFloat };
|
||||
enum { isVec = type_traits_detail::IsVec<UnqualifiedType>::value };
|
||||
|
||||
typedef typename type_traits_detail::Select<IsSimpleParameter<UnqualifiedType>::value,
|
||||
T, typename type_traits_detail::AddParameterType<T>::type>::type ParameterType;
|
||||
};
|
||||
}}}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_TYPE_TRAITS_HPP
|
||||
|
|
@ -0,0 +1,230 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_UTILITY_HPP
|
||||
#define OPENCV_CUDA_UTILITY_HPP
|
||||
|
||||
#include "saturate_cast.hpp"
|
||||
#include "datamov_utils.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
struct CV_EXPORTS ThrustAllocator
|
||||
{
|
||||
typedef uchar value_type;
|
||||
virtual ~ThrustAllocator();
|
||||
virtual __device__ __host__ uchar* allocate(size_t numBytes) = 0;
|
||||
virtual __device__ __host__ void deallocate(uchar* ptr, size_t numBytes) = 0;
|
||||
static ThrustAllocator& getAllocator();
|
||||
static void setAllocator(ThrustAllocator* allocator);
|
||||
};
|
||||
#define OPENCV_CUDA_LOG_WARP_SIZE (5)
|
||||
#define OPENCV_CUDA_WARP_SIZE (1 << OPENCV_CUDA_LOG_WARP_SIZE)
|
||||
#define OPENCV_CUDA_LOG_MEM_BANKS ((__CUDA_ARCH__ >= 200) ? 5 : 4) // 32 banks on fermi, 16 on tesla
|
||||
#define OPENCV_CUDA_MEM_BANKS (1 << OPENCV_CUDA_LOG_MEM_BANKS)
|
||||
|
||||
///////////////////////////////////////////////////////////////////////////////
|
||||
// swap
|
||||
|
||||
template <typename T> void __device__ __host__ __forceinline__ swap(T& a, T& b)
|
||||
{
|
||||
const T temp = a;
|
||||
a = b;
|
||||
b = temp;
|
||||
}
|
||||
|
||||
///////////////////////////////////////////////////////////////////////////////
|
||||
// Mask Reader
|
||||
|
||||
struct SingleMask
|
||||
{
|
||||
explicit __host__ __device__ __forceinline__ SingleMask(PtrStepb mask_) : mask(mask_) {}
|
||||
__host__ __device__ __forceinline__ SingleMask(const SingleMask& mask_): mask(mask_.mask){}
|
||||
|
||||
__device__ __forceinline__ bool operator()(int y, int x) const
|
||||
{
|
||||
return mask.ptr(y)[x] != 0;
|
||||
}
|
||||
|
||||
PtrStepb mask;
|
||||
};
|
||||
|
||||
struct SingleMaskChannels
|
||||
{
|
||||
__host__ __device__ __forceinline__ SingleMaskChannels(PtrStepb mask_, int channels_)
|
||||
: mask(mask_), channels(channels_) {}
|
||||
__host__ __device__ __forceinline__ SingleMaskChannels(const SingleMaskChannels& mask_)
|
||||
:mask(mask_.mask), channels(mask_.channels){}
|
||||
|
||||
__device__ __forceinline__ bool operator()(int y, int x) const
|
||||
{
|
||||
return mask.ptr(y)[x / channels] != 0;
|
||||
}
|
||||
|
||||
PtrStepb mask;
|
||||
int channels;
|
||||
};
|
||||
|
||||
struct MaskCollection
|
||||
{
|
||||
explicit __host__ __device__ __forceinline__ MaskCollection(PtrStepb* maskCollection_)
|
||||
: maskCollection(maskCollection_) {}
|
||||
|
||||
__device__ __forceinline__ MaskCollection(const MaskCollection& masks_)
|
||||
: maskCollection(masks_.maskCollection), curMask(masks_.curMask){}
|
||||
|
||||
__device__ __forceinline__ void next()
|
||||
{
|
||||
curMask = *maskCollection++;
|
||||
}
|
||||
__device__ __forceinline__ void setMask(int z)
|
||||
{
|
||||
curMask = maskCollection[z];
|
||||
}
|
||||
|
||||
__device__ __forceinline__ bool operator()(int y, int x) const
|
||||
{
|
||||
uchar val;
|
||||
return curMask.data == 0 || (ForceGlob<uchar>::Load(curMask.ptr(y), x, val), (val != 0));
|
||||
}
|
||||
|
||||
const PtrStepb* maskCollection;
|
||||
PtrStepb curMask;
|
||||
};
|
||||
|
||||
struct WithOutMask
|
||||
{
|
||||
__host__ __device__ __forceinline__ WithOutMask(){}
|
||||
__host__ __device__ __forceinline__ WithOutMask(const WithOutMask&){}
|
||||
|
||||
__device__ __forceinline__ void next() const
|
||||
{
|
||||
}
|
||||
__device__ __forceinline__ void setMask(int) const
|
||||
{
|
||||
}
|
||||
|
||||
__device__ __forceinline__ bool operator()(int, int) const
|
||||
{
|
||||
return true;
|
||||
}
|
||||
|
||||
__device__ __forceinline__ bool operator()(int, int, int) const
|
||||
{
|
||||
return true;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ bool check(int, int)
|
||||
{
|
||||
return true;
|
||||
}
|
||||
|
||||
static __device__ __forceinline__ bool check(int, int, int)
|
||||
{
|
||||
return true;
|
||||
}
|
||||
};
|
||||
|
||||
///////////////////////////////////////////////////////////////////////////////
|
||||
// Solve linear system
|
||||
|
||||
// solve 2x2 linear system Ax=b
|
||||
template <typename T> __device__ __forceinline__ bool solve2x2(const T A[2][2], const T b[2], T x[2])
|
||||
{
|
||||
T det = A[0][0] * A[1][1] - A[1][0] * A[0][1];
|
||||
|
||||
if (det != 0)
|
||||
{
|
||||
double invdet = 1.0 / det;
|
||||
|
||||
x[0] = saturate_cast<T>(invdet * (b[0] * A[1][1] - b[1] * A[0][1]));
|
||||
|
||||
x[1] = saturate_cast<T>(invdet * (A[0][0] * b[1] - A[1][0] * b[0]));
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
|
||||
// solve 3x3 linear system Ax=b
|
||||
template <typename T> __device__ __forceinline__ bool solve3x3(const T A[3][3], const T b[3], T x[3])
|
||||
{
|
||||
T det = A[0][0] * (A[1][1] * A[2][2] - A[1][2] * A[2][1])
|
||||
- A[0][1] * (A[1][0] * A[2][2] - A[1][2] * A[2][0])
|
||||
+ A[0][2] * (A[1][0] * A[2][1] - A[1][1] * A[2][0]);
|
||||
|
||||
if (det != 0)
|
||||
{
|
||||
double invdet = 1.0 / det;
|
||||
|
||||
x[0] = saturate_cast<T>(invdet *
|
||||
(b[0] * (A[1][1] * A[2][2] - A[1][2] * A[2][1]) -
|
||||
A[0][1] * (b[1] * A[2][2] - A[1][2] * b[2] ) +
|
||||
A[0][2] * (b[1] * A[2][1] - A[1][1] * b[2] )));
|
||||
|
||||
x[1] = saturate_cast<T>(invdet *
|
||||
(A[0][0] * (b[1] * A[2][2] - A[1][2] * b[2] ) -
|
||||
b[0] * (A[1][0] * A[2][2] - A[1][2] * A[2][0]) +
|
||||
A[0][2] * (A[1][0] * b[2] - b[1] * A[2][0])));
|
||||
|
||||
x[2] = saturate_cast<T>(invdet *
|
||||
(A[0][0] * (A[1][1] * b[2] - b[1] * A[2][1]) -
|
||||
A[0][1] * (A[1][0] * b[2] - b[1] * A[2][0]) +
|
||||
b[0] * (A[1][0] * A[2][1] - A[1][1] * A[2][0])));
|
||||
|
||||
return true;
|
||||
}
|
||||
|
||||
return false;
|
||||
}
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_UTILITY_HPP
|
||||
|
|
@ -0,0 +1,232 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_VEC_DISTANCE_HPP
|
||||
#define OPENCV_CUDA_VEC_DISTANCE_HPP
|
||||
|
||||
#include "reduce.hpp"
|
||||
#include "functional.hpp"
|
||||
#include "detail/vec_distance_detail.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template <typename T> struct L1Dist
|
||||
{
|
||||
typedef int value_type;
|
||||
typedef int result_type;
|
||||
|
||||
__device__ __forceinline__ L1Dist() : mySum(0) {}
|
||||
|
||||
__device__ __forceinline__ void reduceIter(int val1, int val2)
|
||||
{
|
||||
mySum = __sad(val1, val2, mySum);
|
||||
}
|
||||
|
||||
template <int THREAD_DIM> __device__ __forceinline__ void reduceAll(int* smem, int tid)
|
||||
{
|
||||
reduce<THREAD_DIM>(smem, mySum, tid, plus<int>());
|
||||
}
|
||||
|
||||
__device__ __forceinline__ operator int() const
|
||||
{
|
||||
return mySum;
|
||||
}
|
||||
|
||||
int mySum;
|
||||
};
|
||||
template <> struct L1Dist<float>
|
||||
{
|
||||
typedef float value_type;
|
||||
typedef float result_type;
|
||||
|
||||
__device__ __forceinline__ L1Dist() : mySum(0.0f) {}
|
||||
|
||||
__device__ __forceinline__ void reduceIter(float val1, float val2)
|
||||
{
|
||||
mySum += ::fabs(val1 - val2);
|
||||
}
|
||||
|
||||
template <int THREAD_DIM> __device__ __forceinline__ void reduceAll(float* smem, int tid)
|
||||
{
|
||||
reduce<THREAD_DIM>(smem, mySum, tid, plus<float>());
|
||||
}
|
||||
|
||||
__device__ __forceinline__ operator float() const
|
||||
{
|
||||
return mySum;
|
||||
}
|
||||
|
||||
float mySum;
|
||||
};
|
||||
|
||||
struct L2Dist
|
||||
{
|
||||
typedef float value_type;
|
||||
typedef float result_type;
|
||||
|
||||
__device__ __forceinline__ L2Dist() : mySum(0.0f) {}
|
||||
|
||||
__device__ __forceinline__ void reduceIter(float val1, float val2)
|
||||
{
|
||||
float reg = val1 - val2;
|
||||
mySum += reg * reg;
|
||||
}
|
||||
|
||||
template <int THREAD_DIM> __device__ __forceinline__ void reduceAll(float* smem, int tid)
|
||||
{
|
||||
reduce<THREAD_DIM>(smem, mySum, tid, plus<float>());
|
||||
}
|
||||
|
||||
__device__ __forceinline__ operator float() const
|
||||
{
|
||||
return sqrtf(mySum);
|
||||
}
|
||||
|
||||
float mySum;
|
||||
};
|
||||
|
||||
struct HammingDist
|
||||
{
|
||||
typedef int value_type;
|
||||
typedef int result_type;
|
||||
|
||||
__device__ __forceinline__ HammingDist() : mySum(0) {}
|
||||
|
||||
__device__ __forceinline__ void reduceIter(int val1, int val2)
|
||||
{
|
||||
mySum += __popc(val1 ^ val2);
|
||||
}
|
||||
|
||||
template <int THREAD_DIM> __device__ __forceinline__ void reduceAll(int* smem, int tid)
|
||||
{
|
||||
reduce<THREAD_DIM>(smem, mySum, tid, plus<int>());
|
||||
}
|
||||
|
||||
__device__ __forceinline__ operator int() const
|
||||
{
|
||||
return mySum;
|
||||
}
|
||||
|
||||
int mySum;
|
||||
};
|
||||
|
||||
// calc distance between two vectors in global memory
|
||||
template <int THREAD_DIM, typename Dist, typename T1, typename T2>
|
||||
__device__ void calcVecDiffGlobal(const T1* vec1, const T2* vec2, int len, Dist& dist, typename Dist::result_type* smem, int tid)
|
||||
{
|
||||
for (int i = tid; i < len; i += THREAD_DIM)
|
||||
{
|
||||
T1 val1;
|
||||
ForceGlob<T1>::Load(vec1, i, val1);
|
||||
|
||||
T2 val2;
|
||||
ForceGlob<T2>::Load(vec2, i, val2);
|
||||
|
||||
dist.reduceIter(val1, val2);
|
||||
}
|
||||
|
||||
dist.reduceAll<THREAD_DIM>(smem, tid);
|
||||
}
|
||||
|
||||
// calc distance between two vectors, first vector is cached in register or shared memory, second vector is in global memory
|
||||
template <int THREAD_DIM, int MAX_LEN, bool LEN_EQ_MAX_LEN, typename Dist, typename T1, typename T2>
|
||||
__device__ __forceinline__ void calcVecDiffCached(const T1* vecCached, const T2* vecGlob, int len, Dist& dist, typename Dist::result_type* smem, int tid)
|
||||
{
|
||||
vec_distance_detail::VecDiffCachedCalculator<THREAD_DIM, MAX_LEN, LEN_EQ_MAX_LEN>::calc(vecCached, vecGlob, len, dist, tid);
|
||||
|
||||
dist.reduceAll<THREAD_DIM>(smem, tid);
|
||||
}
|
||||
|
||||
// calc distance between two vectors in global memory
|
||||
template <int THREAD_DIM, typename T1> struct VecDiffGlobal
|
||||
{
|
||||
explicit __device__ __forceinline__ VecDiffGlobal(const T1* vec1_, int = 0, void* = 0, int = 0, int = 0)
|
||||
{
|
||||
vec1 = vec1_;
|
||||
}
|
||||
|
||||
template <typename T2, typename Dist>
|
||||
__device__ __forceinline__ void calc(const T2* vec2, int len, Dist& dist, typename Dist::result_type* smem, int tid) const
|
||||
{
|
||||
calcVecDiffGlobal<THREAD_DIM>(vec1, vec2, len, dist, smem, tid);
|
||||
}
|
||||
|
||||
const T1* vec1;
|
||||
};
|
||||
|
||||
// calc distance between two vectors, first vector is cached in register memory, second vector is in global memory
|
||||
template <int THREAD_DIM, int MAX_LEN, bool LEN_EQ_MAX_LEN, typename U> struct VecDiffCachedRegister
|
||||
{
|
||||
template <typename T1> __device__ __forceinline__ VecDiffCachedRegister(const T1* vec1, int len, U* smem, int glob_tid, int tid)
|
||||
{
|
||||
if (glob_tid < len)
|
||||
smem[glob_tid] = vec1[glob_tid];
|
||||
__syncthreads();
|
||||
|
||||
U* vec1ValsPtr = vec1Vals;
|
||||
|
||||
#pragma unroll
|
||||
for (int i = tid; i < MAX_LEN; i += THREAD_DIM)
|
||||
*vec1ValsPtr++ = smem[i];
|
||||
|
||||
__syncthreads();
|
||||
}
|
||||
|
||||
template <typename T2, typename Dist>
|
||||
__device__ __forceinline__ void calc(const T2* vec2, int len, Dist& dist, typename Dist::result_type* smem, int tid) const
|
||||
{
|
||||
calcVecDiffCached<THREAD_DIM, MAX_LEN, LEN_EQ_MAX_LEN>(vec1Vals, vec2, len, dist, smem, tid);
|
||||
}
|
||||
|
||||
U vec1Vals[MAX_LEN / THREAD_DIM];
|
||||
};
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_VEC_DISTANCE_HPP
|
||||
|
|
@ -0,0 +1,930 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_VECMATH_HPP
|
||||
#define OPENCV_CUDA_VECMATH_HPP
|
||||
|
||||
#include "vec_traits.hpp"
|
||||
#include "saturate_cast.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
|
||||
// saturate_cast
|
||||
|
||||
namespace vec_math_detail
|
||||
{
|
||||
template <int cn, typename VecD> struct SatCastHelper;
|
||||
template <typename VecD> struct SatCastHelper<1, VecD>
|
||||
{
|
||||
template <typename VecS> static __device__ __forceinline__ VecD cast(const VecS& v)
|
||||
{
|
||||
typedef typename VecTraits<VecD>::elem_type D;
|
||||
return VecTraits<VecD>::make(saturate_cast<D>(v.x));
|
||||
}
|
||||
};
|
||||
template <typename VecD> struct SatCastHelper<2, VecD>
|
||||
{
|
||||
template <typename VecS> static __device__ __forceinline__ VecD cast(const VecS& v)
|
||||
{
|
||||
typedef typename VecTraits<VecD>::elem_type D;
|
||||
return VecTraits<VecD>::make(saturate_cast<D>(v.x), saturate_cast<D>(v.y));
|
||||
}
|
||||
};
|
||||
template <typename VecD> struct SatCastHelper<3, VecD>
|
||||
{
|
||||
template <typename VecS> static __device__ __forceinline__ VecD cast(const VecS& v)
|
||||
{
|
||||
typedef typename VecTraits<VecD>::elem_type D;
|
||||
return VecTraits<VecD>::make(saturate_cast<D>(v.x), saturate_cast<D>(v.y), saturate_cast<D>(v.z));
|
||||
}
|
||||
};
|
||||
template <typename VecD> struct SatCastHelper<4, VecD>
|
||||
{
|
||||
template <typename VecS> static __device__ __forceinline__ VecD cast(const VecS& v)
|
||||
{
|
||||
typedef typename VecTraits<VecD>::elem_type D;
|
||||
return VecTraits<VecD>::make(saturate_cast<D>(v.x), saturate_cast<D>(v.y), saturate_cast<D>(v.z), saturate_cast<D>(v.w));
|
||||
}
|
||||
};
|
||||
|
||||
template <typename VecD, typename VecS> static __device__ __forceinline__ VecD saturate_cast_helper(const VecS& v)
|
||||
{
|
||||
return SatCastHelper<VecTraits<VecD>::cn, VecD>::cast(v);
|
||||
}
|
||||
}
|
||||
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const uchar1& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const char1& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const ushort1& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const short1& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const uint1& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const int1& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const float1& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const double1& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const uchar2& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const char2& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const ushort2& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const short2& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const uint2& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const int2& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const float2& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const double2& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const uchar3& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const char3& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const ushort3& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const short3& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const uint3& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const int3& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const float3& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const double3& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const uchar4& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const char4& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const ushort4& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const short4& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const uint4& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const int4& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const float4& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
template<typename T> static __device__ __forceinline__ T saturate_cast(const double4& v) {return vec_math_detail::saturate_cast_helper<T>(v);}
|
||||
|
||||
// unary operators
|
||||
|
||||
#define CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(op, input_type, output_type) \
|
||||
__device__ __forceinline__ output_type ## 1 operator op(const input_type ## 1 & a) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 1>::make(op (a.x)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 2 operator op(const input_type ## 2 & a) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 2>::make(op (a.x), op (a.y)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 3 operator op(const input_type ## 3 & a) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 3>::make(op (a.x), op (a.y), op (a.z)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 4 operator op(const input_type ## 4 & a) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 4>::make(op (a.x), op (a.y), op (a.z), op (a.w)); \
|
||||
}
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(-, char, char)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(-, short, short)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(-, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(-, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(-, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(!, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(!, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(!, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(!, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(!, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(!, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(!, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(!, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(~, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(~, char, char)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(~, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(~, short, short)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(~, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_OP(~, uint, uint)
|
||||
|
||||
#undef CV_CUDEV_IMPLEMENT_VEC_UNARY_OP
|
||||
|
||||
// unary functions
|
||||
|
||||
#define CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(func_name, func, input_type, output_type) \
|
||||
__device__ __forceinline__ output_type ## 1 func_name(const input_type ## 1 & a) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 1>::make(func (a.x)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 2 func_name(const input_type ## 2 & a) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 2>::make(func (a.x), func (a.y)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 3 func_name(const input_type ## 3 & a) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 3>::make(func (a.x), func (a.y), func (a.z)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 4 func_name(const input_type ## 4 & a) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 4>::make(func (a.x), func (a.y), func (a.z), func (a.w)); \
|
||||
}
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(abs, /*::abs*/, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(abs, ::abs, char, char)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(abs, /*::abs*/, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(abs, ::abs, short, short)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(abs, ::abs, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(abs, /*::abs*/, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(abs, ::fabsf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(abs, ::fabs, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sqrt, ::sqrtf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sqrt, ::sqrtf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sqrt, ::sqrtf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sqrt, ::sqrtf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sqrt, ::sqrtf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sqrt, ::sqrtf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sqrt, ::sqrtf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sqrt, ::sqrt, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp, ::expf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp, ::expf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp, ::expf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp, ::expf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp, ::expf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp, ::expf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp, ::expf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp, ::exp, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp2, ::exp2f, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp2, ::exp2f, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp2, ::exp2f, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp2, ::exp2f, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp2, ::exp2f, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp2, ::exp2f, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp2, ::exp2f, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp2, ::exp2, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp10, ::exp10f, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp10, ::exp10f, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp10, ::exp10f, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp10, ::exp10f, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp10, ::exp10f, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp10, ::exp10f, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp10, ::exp10f, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(exp10, ::exp10, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log, ::logf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log, ::logf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log, ::logf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log, ::logf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log, ::logf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log, ::logf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log, ::logf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log, ::log, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log2, ::log2f, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log2, ::log2f, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log2, ::log2f, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log2, ::log2f, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log2, ::log2f, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log2, ::log2f, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log2, ::log2f, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log2, ::log2, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log10, ::log10f, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log10, ::log10f, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log10, ::log10f, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log10, ::log10f, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log10, ::log10f, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log10, ::log10f, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log10, ::log10f, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(log10, ::log10, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sin, ::sinf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sin, ::sinf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sin, ::sinf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sin, ::sinf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sin, ::sinf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sin, ::sinf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sin, ::sinf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sin, ::sin, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cos, ::cosf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cos, ::cosf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cos, ::cosf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cos, ::cosf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cos, ::cosf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cos, ::cosf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cos, ::cosf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cos, ::cos, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tan, ::tanf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tan, ::tanf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tan, ::tanf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tan, ::tanf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tan, ::tanf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tan, ::tanf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tan, ::tanf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tan, ::tan, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asin, ::asinf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asin, ::asinf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asin, ::asinf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asin, ::asinf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asin, ::asinf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asin, ::asinf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asin, ::asinf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asin, ::asin, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acos, ::acosf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acos, ::acosf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acos, ::acosf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acos, ::acosf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acos, ::acosf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acos, ::acosf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acos, ::acosf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acos, ::acos, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atan, ::atanf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atan, ::atanf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atan, ::atanf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atan, ::atanf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atan, ::atanf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atan, ::atanf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atan, ::atanf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atan, ::atan, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sinh, ::sinhf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sinh, ::sinhf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sinh, ::sinhf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sinh, ::sinhf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sinh, ::sinhf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sinh, ::sinhf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sinh, ::sinhf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(sinh, ::sinh, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cosh, ::coshf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cosh, ::coshf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cosh, ::coshf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cosh, ::coshf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cosh, ::coshf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cosh, ::coshf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cosh, ::coshf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(cosh, ::cosh, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tanh, ::tanhf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tanh, ::tanhf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tanh, ::tanhf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tanh, ::tanhf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tanh, ::tanhf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tanh, ::tanhf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tanh, ::tanhf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(tanh, ::tanh, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asinh, ::asinhf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asinh, ::asinhf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asinh, ::asinhf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asinh, ::asinhf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asinh, ::asinhf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asinh, ::asinhf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asinh, ::asinhf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(asinh, ::asinh, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acosh, ::acoshf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acosh, ::acoshf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acosh, ::acoshf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acosh, ::acoshf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acosh, ::acoshf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acosh, ::acoshf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acosh, ::acoshf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(acosh, ::acosh, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atanh, ::atanhf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atanh, ::atanhf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atanh, ::atanhf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atanh, ::atanhf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atanh, ::atanhf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atanh, ::atanhf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atanh, ::atanhf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC(atanh, ::atanh, double, double)
|
||||
|
||||
#undef CV_CUDEV_IMPLEMENT_VEC_UNARY_FUNC
|
||||
|
||||
// binary operators (vec & vec)
|
||||
|
||||
#define CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(op, input_type, output_type) \
|
||||
__device__ __forceinline__ output_type ## 1 operator op(const input_type ## 1 & a, const input_type ## 1 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 1>::make(a.x op b.x); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 2 operator op(const input_type ## 2 & a, const input_type ## 2 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 2>::make(a.x op b.x, a.y op b.y); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 3 operator op(const input_type ## 3 & a, const input_type ## 3 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 3>::make(a.x op b.x, a.y op b.y, a.z op b.z); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 4 operator op(const input_type ## 4 & a, const input_type ## 4 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 4>::make(a.x op b.x, a.y op b.y, a.z op b.z, a.w op b.w); \
|
||||
}
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(+, uchar, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(+, char, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(+, ushort, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(+, short, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(+, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(+, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(+, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(+, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(-, uchar, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(-, char, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(-, ushort, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(-, short, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(-, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(-, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(-, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(-, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(*, uchar, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(*, char, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(*, ushort, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(*, short, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(*, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(*, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(*, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(*, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(/, uchar, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(/, char, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(/, ushort, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(/, short, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(/, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(/, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(/, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(/, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(==, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(==, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(==, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(==, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(==, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(==, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(==, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(==, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(!=, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(!=, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(!=, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(!=, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(!=, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(!=, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(!=, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(!=, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>=, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>=, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>=, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>=, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>=, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>=, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>=, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(>=, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<=, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<=, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<=, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<=, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<=, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<=, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<=, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(<=, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&&, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&&, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&&, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&&, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&&, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&&, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&&, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&&, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(||, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(||, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(||, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(||, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(||, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(||, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(||, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(||, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&, char, char)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&, short, short)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(&, uint, uint)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(|, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(|, char, char)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(|, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(|, short, short)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(|, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(|, uint, uint)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(^, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(^, char, char)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(^, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(^, short, short)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(^, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_OP(^, uint, uint)
|
||||
|
||||
#undef CV_CUDEV_IMPLEMENT_VEC_BINARY_OP
|
||||
|
||||
// binary operators (vec & scalar)
|
||||
|
||||
#define CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(op, input_type, scalar_type, output_type) \
|
||||
__device__ __forceinline__ output_type ## 1 operator op(const input_type ## 1 & a, scalar_type s) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 1>::make(a.x op s); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 1 operator op(scalar_type s, const input_type ## 1 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 1>::make(s op b.x); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 2 operator op(const input_type ## 2 & a, scalar_type s) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 2>::make(a.x op s, a.y op s); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 2 operator op(scalar_type s, const input_type ## 2 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 2>::make(s op b.x, s op b.y); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 3 operator op(const input_type ## 3 & a, scalar_type s) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 3>::make(a.x op s, a.y op s, a.z op s); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 3 operator op(scalar_type s, const input_type ## 3 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 3>::make(s op b.x, s op b.y, s op b.z); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 4 operator op(const input_type ## 4 & a, scalar_type s) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 4>::make(a.x op s, a.y op s, a.z op s, a.w op s); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 4 operator op(scalar_type s, const input_type ## 4 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 4>::make(s op b.x, s op b.y, s op b.z, s op b.w); \
|
||||
}
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, uchar, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, uchar, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, uchar, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, char, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, char, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, char, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, ushort, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, ushort, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, ushort, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, short, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, short, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, short, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, int, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, int, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, int, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, uint, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, uint, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, uint, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, float, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, float, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(+, double, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, uchar, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, uchar, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, uchar, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, char, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, char, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, char, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, ushort, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, ushort, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, ushort, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, short, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, short, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, short, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, int, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, int, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, int, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, uint, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, uint, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, uint, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, float, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, float, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(-, double, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, uchar, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, uchar, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, uchar, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, char, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, char, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, char, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, ushort, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, ushort, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, ushort, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, short, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, short, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, short, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, int, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, int, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, int, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, uint, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, uint, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, uint, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, float, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, float, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(*, double, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, uchar, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, uchar, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, uchar, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, char, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, char, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, char, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, ushort, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, ushort, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, ushort, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, short, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, short, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, short, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, int, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, int, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, int, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, uint, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, uint, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, uint, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, float, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, float, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(/, double, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(==, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(==, char, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(==, ushort, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(==, short, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(==, int, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(==, uint, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(==, float, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(==, double, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(!=, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(!=, char, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(!=, ushort, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(!=, short, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(!=, int, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(!=, uint, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(!=, float, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(!=, double, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>, char, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>, ushort, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>, short, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>, int, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>, uint, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>, float, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>, double, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<, char, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<, ushort, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<, short, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<, int, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<, uint, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<, float, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<, double, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>=, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>=, char, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>=, ushort, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>=, short, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>=, int, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>=, uint, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>=, float, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(>=, double, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<=, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<=, char, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<=, ushort, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<=, short, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<=, int, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<=, uint, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<=, float, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(<=, double, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&&, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&&, char, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&&, ushort, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&&, short, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&&, int, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&&, uint, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&&, float, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&&, double, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(||, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(||, char, char, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(||, ushort, ushort, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(||, short, short, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(||, int, int, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(||, uint, uint, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(||, float, float, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(||, double, double, uchar)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&, char, char, char)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&, ushort, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&, short, short, short)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&, int, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(&, uint, uint, uint)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(|, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(|, char, char, char)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(|, ushort, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(|, short, short, short)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(|, int, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(|, uint, uint, uint)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(^, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(^, char, char, char)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(^, ushort, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(^, short, short, short)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(^, int, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP(^, uint, uint, uint)
|
||||
|
||||
#undef CV_CUDEV_IMPLEMENT_SCALAR_BINARY_OP
|
||||
|
||||
// binary function (vec & vec)
|
||||
|
||||
#define CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(func_name, func, input_type, output_type) \
|
||||
__device__ __forceinline__ output_type ## 1 func_name(const input_type ## 1 & a, const input_type ## 1 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 1>::make(func (a.x, b.x)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 2 func_name(const input_type ## 2 & a, const input_type ## 2 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 2>::make(func (a.x, b.x), func (a.y, b.y)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 3 func_name(const input_type ## 3 & a, const input_type ## 3 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 3>::make(func (a.x, b.x), func (a.y, b.y), func (a.z, b.z)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 4 func_name(const input_type ## 4 & a, const input_type ## 4 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 4>::make(func (a.x, b.x), func (a.y, b.y), func (a.z, b.z), func (a.w, b.w)); \
|
||||
}
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(max, ::max, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(max, ::max, char, char)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(max, ::max, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(max, ::max, short, short)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(max, ::max, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(max, ::max, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(max, ::fmaxf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(max, ::fmax, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(min, ::min, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(min, ::min, char, char)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(min, ::min, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(min, ::min, short, short)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(min, ::min, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(min, ::min, int, int)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(min, ::fminf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(min, ::fmin, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(hypot, ::hypotf, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(hypot, ::hypotf, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(hypot, ::hypotf, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(hypot, ::hypotf, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(hypot, ::hypotf, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(hypot, ::hypotf, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(hypot, ::hypotf, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(hypot, ::hypot, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(atan2, ::atan2f, uchar, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(atan2, ::atan2f, char, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(atan2, ::atan2f, ushort, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(atan2, ::atan2f, short, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(atan2, ::atan2f, uint, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(atan2, ::atan2f, int, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(atan2, ::atan2f, float, float)
|
||||
CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC(atan2, ::atan2, double, double)
|
||||
|
||||
#undef CV_CUDEV_IMPLEMENT_VEC_BINARY_FUNC
|
||||
|
||||
// binary function (vec & scalar)
|
||||
|
||||
#define CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(func_name, func, input_type, scalar_type, output_type) \
|
||||
__device__ __forceinline__ output_type ## 1 func_name(const input_type ## 1 & a, scalar_type s) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 1>::make(func ((output_type) a.x, (output_type) s)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 1 func_name(scalar_type s, const input_type ## 1 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 1>::make(func ((output_type) s, (output_type) b.x)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 2 func_name(const input_type ## 2 & a, scalar_type s) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 2>::make(func ((output_type) a.x, (output_type) s), func ((output_type) a.y, (output_type) s)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 2 func_name(scalar_type s, const input_type ## 2 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 2>::make(func ((output_type) s, (output_type) b.x), func ((output_type) s, (output_type) b.y)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 3 func_name(const input_type ## 3 & a, scalar_type s) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 3>::make(func ((output_type) a.x, (output_type) s), func ((output_type) a.y, (output_type) s), func ((output_type) a.z, (output_type) s)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 3 func_name(scalar_type s, const input_type ## 3 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 3>::make(func ((output_type) s, (output_type) b.x), func ((output_type) s, (output_type) b.y), func ((output_type) s, (output_type) b.z)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 4 func_name(const input_type ## 4 & a, scalar_type s) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 4>::make(func ((output_type) a.x, (output_type) s), func ((output_type) a.y, (output_type) s), func ((output_type) a.z, (output_type) s), func ((output_type) a.w, (output_type) s)); \
|
||||
} \
|
||||
__device__ __forceinline__ output_type ## 4 func_name(scalar_type s, const input_type ## 4 & b) \
|
||||
{ \
|
||||
return VecTraits<output_type ## 4>::make(func ((output_type) s, (output_type) b.x), func ((output_type) s, (output_type) b.y), func ((output_type) s, (output_type) b.z), func ((output_type) s, (output_type) b.w)); \
|
||||
}
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::max, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmaxf, uchar, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmax, uchar, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::max, char, char, char)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmaxf, char, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmax, char, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::max, ushort, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmaxf, ushort, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmax, ushort, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::max, short, short, short)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmaxf, short, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmax, short, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::max, uint, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmaxf, uint, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmax, uint, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::max, int, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmaxf, int, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmax, int, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmaxf, float, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmax, float, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(max, ::fmax, double, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::min, uchar, uchar, uchar)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fminf, uchar, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fmin, uchar, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::min, char, char, char)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fminf, char, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fmin, char, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::min, ushort, ushort, ushort)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fminf, ushort, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fmin, ushort, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::min, short, short, short)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fminf, short, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fmin, short, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::min, uint, uint, uint)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fminf, uint, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fmin, uint, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::min, int, int, int)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fminf, int, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fmin, int, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fminf, float, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fmin, float, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(min, ::fmin, double, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypotf, uchar, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypot, uchar, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypotf, char, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypot, char, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypotf, ushort, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypot, ushort, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypotf, short, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypot, short, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypotf, uint, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypot, uint, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypotf, int, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypot, int, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypotf, float, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypot, float, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(hypot, ::hypot, double, double, double)
|
||||
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2f, uchar, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2, uchar, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2f, char, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2, char, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2f, ushort, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2, ushort, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2f, short, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2, short, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2f, uint, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2, uint, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2f, int, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2, int, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2f, float, float, float)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2, float, double, double)
|
||||
CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC(atan2, ::atan2, double, double, double)
|
||||
|
||||
#undef CV_CUDEV_IMPLEMENT_SCALAR_BINARY_FUNC
|
||||
|
||||
}}} // namespace cv { namespace cuda { namespace device
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_VECMATH_HPP
|
||||
|
|
@ -0,0 +1,288 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_VEC_TRAITS_HPP
|
||||
#define OPENCV_CUDA_VEC_TRAITS_HPP
|
||||
|
||||
#include "common.hpp"
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template<typename T, int N> struct TypeVec;
|
||||
|
||||
struct __align__(8) uchar8
|
||||
{
|
||||
uchar a0, a1, a2, a3, a4, a5, a6, a7;
|
||||
};
|
||||
static __host__ __device__ __forceinline__ uchar8 make_uchar8(uchar a0, uchar a1, uchar a2, uchar a3, uchar a4, uchar a5, uchar a6, uchar a7)
|
||||
{
|
||||
uchar8 val = {a0, a1, a2, a3, a4, a5, a6, a7};
|
||||
return val;
|
||||
}
|
||||
struct __align__(8) char8
|
||||
{
|
||||
schar a0, a1, a2, a3, a4, a5, a6, a7;
|
||||
};
|
||||
static __host__ __device__ __forceinline__ char8 make_char8(schar a0, schar a1, schar a2, schar a3, schar a4, schar a5, schar a6, schar a7)
|
||||
{
|
||||
char8 val = {a0, a1, a2, a3, a4, a5, a6, a7};
|
||||
return val;
|
||||
}
|
||||
struct __align__(16) ushort8
|
||||
{
|
||||
ushort a0, a1, a2, a3, a4, a5, a6, a7;
|
||||
};
|
||||
static __host__ __device__ __forceinline__ ushort8 make_ushort8(ushort a0, ushort a1, ushort a2, ushort a3, ushort a4, ushort a5, ushort a6, ushort a7)
|
||||
{
|
||||
ushort8 val = {a0, a1, a2, a3, a4, a5, a6, a7};
|
||||
return val;
|
||||
}
|
||||
struct __align__(16) short8
|
||||
{
|
||||
short a0, a1, a2, a3, a4, a5, a6, a7;
|
||||
};
|
||||
static __host__ __device__ __forceinline__ short8 make_short8(short a0, short a1, short a2, short a3, short a4, short a5, short a6, short a7)
|
||||
{
|
||||
short8 val = {a0, a1, a2, a3, a4, a5, a6, a7};
|
||||
return val;
|
||||
}
|
||||
struct __align__(32) uint8
|
||||
{
|
||||
uint a0, a1, a2, a3, a4, a5, a6, a7;
|
||||
};
|
||||
static __host__ __device__ __forceinline__ uint8 make_uint8(uint a0, uint a1, uint a2, uint a3, uint a4, uint a5, uint a6, uint a7)
|
||||
{
|
||||
uint8 val = {a0, a1, a2, a3, a4, a5, a6, a7};
|
||||
return val;
|
||||
}
|
||||
struct __align__(32) int8
|
||||
{
|
||||
int a0, a1, a2, a3, a4, a5, a6, a7;
|
||||
};
|
||||
static __host__ __device__ __forceinline__ int8 make_int8(int a0, int a1, int a2, int a3, int a4, int a5, int a6, int a7)
|
||||
{
|
||||
int8 val = {a0, a1, a2, a3, a4, a5, a6, a7};
|
||||
return val;
|
||||
}
|
||||
struct __align__(32) float8
|
||||
{
|
||||
float a0, a1, a2, a3, a4, a5, a6, a7;
|
||||
};
|
||||
static __host__ __device__ __forceinline__ float8 make_float8(float a0, float a1, float a2, float a3, float a4, float a5, float a6, float a7)
|
||||
{
|
||||
float8 val = {a0, a1, a2, a3, a4, a5, a6, a7};
|
||||
return val;
|
||||
}
|
||||
struct double8
|
||||
{
|
||||
double a0, a1, a2, a3, a4, a5, a6, a7;
|
||||
};
|
||||
static __host__ __device__ __forceinline__ double8 make_double8(double a0, double a1, double a2, double a3, double a4, double a5, double a6, double a7)
|
||||
{
|
||||
double8 val = {a0, a1, a2, a3, a4, a5, a6, a7};
|
||||
return val;
|
||||
}
|
||||
|
||||
#define OPENCV_CUDA_IMPLEMENT_TYPE_VEC(type) \
|
||||
template<> struct TypeVec<type, 1> { typedef type vec_type; }; \
|
||||
template<> struct TypeVec<type ## 1, 1> { typedef type ## 1 vec_type; }; \
|
||||
template<> struct TypeVec<type, 2> { typedef type ## 2 vec_type; }; \
|
||||
template<> struct TypeVec<type ## 2, 2> { typedef type ## 2 vec_type; }; \
|
||||
template<> struct TypeVec<type, 3> { typedef type ## 3 vec_type; }; \
|
||||
template<> struct TypeVec<type ## 3, 3> { typedef type ## 3 vec_type; }; \
|
||||
template<> struct TypeVec<type, 4> { typedef type ## 4 vec_type; }; \
|
||||
template<> struct TypeVec<type ## 4, 4> { typedef type ## 4 vec_type; }; \
|
||||
template<> struct TypeVec<type, 8> { typedef type ## 8 vec_type; }; \
|
||||
template<> struct TypeVec<type ## 8, 8> { typedef type ## 8 vec_type; };
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_TYPE_VEC(uchar)
|
||||
OPENCV_CUDA_IMPLEMENT_TYPE_VEC(char)
|
||||
OPENCV_CUDA_IMPLEMENT_TYPE_VEC(ushort)
|
||||
OPENCV_CUDA_IMPLEMENT_TYPE_VEC(short)
|
||||
OPENCV_CUDA_IMPLEMENT_TYPE_VEC(int)
|
||||
OPENCV_CUDA_IMPLEMENT_TYPE_VEC(uint)
|
||||
OPENCV_CUDA_IMPLEMENT_TYPE_VEC(float)
|
||||
OPENCV_CUDA_IMPLEMENT_TYPE_VEC(double)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_TYPE_VEC
|
||||
|
||||
template<> struct TypeVec<schar, 1> { typedef schar vec_type; };
|
||||
template<> struct TypeVec<schar, 2> { typedef char2 vec_type; };
|
||||
template<> struct TypeVec<schar, 3> { typedef char3 vec_type; };
|
||||
template<> struct TypeVec<schar, 4> { typedef char4 vec_type; };
|
||||
template<> struct TypeVec<schar, 8> { typedef char8 vec_type; };
|
||||
|
||||
template<> struct TypeVec<bool, 1> { typedef uchar vec_type; };
|
||||
template<> struct TypeVec<bool, 2> { typedef uchar2 vec_type; };
|
||||
template<> struct TypeVec<bool, 3> { typedef uchar3 vec_type; };
|
||||
template<> struct TypeVec<bool, 4> { typedef uchar4 vec_type; };
|
||||
template<> struct TypeVec<bool, 8> { typedef uchar8 vec_type; };
|
||||
|
||||
template<typename T> struct VecTraits;
|
||||
|
||||
#define OPENCV_CUDA_IMPLEMENT_VEC_TRAITS(type) \
|
||||
template<> struct VecTraits<type> \
|
||||
{ \
|
||||
typedef type elem_type; \
|
||||
enum {cn=1}; \
|
||||
static __device__ __host__ __forceinline__ type all(type v) {return v;} \
|
||||
static __device__ __host__ __forceinline__ type make(type x) {return x;} \
|
||||
static __device__ __host__ __forceinline__ type make(const type* v) {return *v;} \
|
||||
}; \
|
||||
template<> struct VecTraits<type ## 1> \
|
||||
{ \
|
||||
typedef type elem_type; \
|
||||
enum {cn=1}; \
|
||||
static __device__ __host__ __forceinline__ type ## 1 all(type v) {return make_ ## type ## 1(v);} \
|
||||
static __device__ __host__ __forceinline__ type ## 1 make(type x) {return make_ ## type ## 1(x);} \
|
||||
static __device__ __host__ __forceinline__ type ## 1 make(const type* v) {return make_ ## type ## 1(*v);} \
|
||||
}; \
|
||||
template<> struct VecTraits<type ## 2> \
|
||||
{ \
|
||||
typedef type elem_type; \
|
||||
enum {cn=2}; \
|
||||
static __device__ __host__ __forceinline__ type ## 2 all(type v) {return make_ ## type ## 2(v, v);} \
|
||||
static __device__ __host__ __forceinline__ type ## 2 make(type x, type y) {return make_ ## type ## 2(x, y);} \
|
||||
static __device__ __host__ __forceinline__ type ## 2 make(const type* v) {return make_ ## type ## 2(v[0], v[1]);} \
|
||||
}; \
|
||||
template<> struct VecTraits<type ## 3> \
|
||||
{ \
|
||||
typedef type elem_type; \
|
||||
enum {cn=3}; \
|
||||
static __device__ __host__ __forceinline__ type ## 3 all(type v) {return make_ ## type ## 3(v, v, v);} \
|
||||
static __device__ __host__ __forceinline__ type ## 3 make(type x, type y, type z) {return make_ ## type ## 3(x, y, z);} \
|
||||
static __device__ __host__ __forceinline__ type ## 3 make(const type* v) {return make_ ## type ## 3(v[0], v[1], v[2]);} \
|
||||
}; \
|
||||
template<> struct VecTraits<type ## 4> \
|
||||
{ \
|
||||
typedef type elem_type; \
|
||||
enum {cn=4}; \
|
||||
static __device__ __host__ __forceinline__ type ## 4 all(type v) {return make_ ## type ## 4(v, v, v, v);} \
|
||||
static __device__ __host__ __forceinline__ type ## 4 make(type x, type y, type z, type w) {return make_ ## type ## 4(x, y, z, w);} \
|
||||
static __device__ __host__ __forceinline__ type ## 4 make(const type* v) {return make_ ## type ## 4(v[0], v[1], v[2], v[3]);} \
|
||||
}; \
|
||||
template<> struct VecTraits<type ## 8> \
|
||||
{ \
|
||||
typedef type elem_type; \
|
||||
enum {cn=8}; \
|
||||
static __device__ __host__ __forceinline__ type ## 8 all(type v) {return make_ ## type ## 8(v, v, v, v, v, v, v, v);} \
|
||||
static __device__ __host__ __forceinline__ type ## 8 make(type a0, type a1, type a2, type a3, type a4, type a5, type a6, type a7) {return make_ ## type ## 8(a0, a1, a2, a3, a4, a5, a6, a7);} \
|
||||
static __device__ __host__ __forceinline__ type ## 8 make(const type* v) {return make_ ## type ## 8(v[0], v[1], v[2], v[3], v[4], v[5], v[6], v[7]);} \
|
||||
};
|
||||
|
||||
OPENCV_CUDA_IMPLEMENT_VEC_TRAITS(uchar)
|
||||
OPENCV_CUDA_IMPLEMENT_VEC_TRAITS(ushort)
|
||||
OPENCV_CUDA_IMPLEMENT_VEC_TRAITS(short)
|
||||
OPENCV_CUDA_IMPLEMENT_VEC_TRAITS(int)
|
||||
OPENCV_CUDA_IMPLEMENT_VEC_TRAITS(uint)
|
||||
OPENCV_CUDA_IMPLEMENT_VEC_TRAITS(float)
|
||||
OPENCV_CUDA_IMPLEMENT_VEC_TRAITS(double)
|
||||
|
||||
#undef OPENCV_CUDA_IMPLEMENT_VEC_TRAITS
|
||||
|
||||
template<> struct VecTraits<char>
|
||||
{
|
||||
typedef char elem_type;
|
||||
enum {cn=1};
|
||||
static __device__ __host__ __forceinline__ char all(char v) {return v;}
|
||||
static __device__ __host__ __forceinline__ char make(char x) {return x;}
|
||||
static __device__ __host__ __forceinline__ char make(const char* x) {return *x;}
|
||||
};
|
||||
template<> struct VecTraits<schar>
|
||||
{
|
||||
typedef schar elem_type;
|
||||
enum {cn=1};
|
||||
static __device__ __host__ __forceinline__ schar all(schar v) {return v;}
|
||||
static __device__ __host__ __forceinline__ schar make(schar x) {return x;}
|
||||
static __device__ __host__ __forceinline__ schar make(const schar* x) {return *x;}
|
||||
};
|
||||
template<> struct VecTraits<char1>
|
||||
{
|
||||
typedef schar elem_type;
|
||||
enum {cn=1};
|
||||
static __device__ __host__ __forceinline__ char1 all(schar v) {return make_char1(v);}
|
||||
static __device__ __host__ __forceinline__ char1 make(schar x) {return make_char1(x);}
|
||||
static __device__ __host__ __forceinline__ char1 make(const schar* v) {return make_char1(v[0]);}
|
||||
};
|
||||
template<> struct VecTraits<char2>
|
||||
{
|
||||
typedef schar elem_type;
|
||||
enum {cn=2};
|
||||
static __device__ __host__ __forceinline__ char2 all(schar v) {return make_char2(v, v);}
|
||||
static __device__ __host__ __forceinline__ char2 make(schar x, schar y) {return make_char2(x, y);}
|
||||
static __device__ __host__ __forceinline__ char2 make(const schar* v) {return make_char2(v[0], v[1]);}
|
||||
};
|
||||
template<> struct VecTraits<char3>
|
||||
{
|
||||
typedef schar elem_type;
|
||||
enum {cn=3};
|
||||
static __device__ __host__ __forceinline__ char3 all(schar v) {return make_char3(v, v, v);}
|
||||
static __device__ __host__ __forceinline__ char3 make(schar x, schar y, schar z) {return make_char3(x, y, z);}
|
||||
static __device__ __host__ __forceinline__ char3 make(const schar* v) {return make_char3(v[0], v[1], v[2]);}
|
||||
};
|
||||
template<> struct VecTraits<char4>
|
||||
{
|
||||
typedef schar elem_type;
|
||||
enum {cn=4};
|
||||
static __device__ __host__ __forceinline__ char4 all(schar v) {return make_char4(v, v, v, v);}
|
||||
static __device__ __host__ __forceinline__ char4 make(schar x, schar y, schar z, schar w) {return make_char4(x, y, z, w);}
|
||||
static __device__ __host__ __forceinline__ char4 make(const schar* v) {return make_char4(v[0], v[1], v[2], v[3]);}
|
||||
};
|
||||
template<> struct VecTraits<char8>
|
||||
{
|
||||
typedef schar elem_type;
|
||||
enum {cn=8};
|
||||
static __device__ __host__ __forceinline__ char8 all(schar v) {return make_char8(v, v, v, v, v, v, v, v);}
|
||||
static __device__ __host__ __forceinline__ char8 make(schar a0, schar a1, schar a2, schar a3, schar a4, schar a5, schar a6, schar a7) {return make_char8(a0, a1, a2, a3, a4, a5, a6, a7);}
|
||||
static __device__ __host__ __forceinline__ char8 make(const schar* v) {return make_char8(v[0], v[1], v[2], v[3], v[4], v[5], v[6], v[7]);}
|
||||
};
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_VEC_TRAITS_HPP
|
||||
|
|
@ -0,0 +1,139 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_DEVICE_WARP_HPP
|
||||
#define OPENCV_CUDA_DEVICE_WARP_HPP
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
struct Warp
|
||||
{
|
||||
enum
|
||||
{
|
||||
LOG_WARP_SIZE = 5,
|
||||
WARP_SIZE = 1 << LOG_WARP_SIZE,
|
||||
STRIDE = WARP_SIZE
|
||||
};
|
||||
|
||||
/** \brief Returns the warp lane ID of the calling thread. */
|
||||
static __device__ __forceinline__ unsigned int laneId()
|
||||
{
|
||||
unsigned int ret;
|
||||
asm("mov.u32 %0, %%laneid;" : "=r"(ret) );
|
||||
return ret;
|
||||
}
|
||||
|
||||
template<typename It, typename T>
|
||||
static __device__ __forceinline__ void fill(It beg, It end, const T& value)
|
||||
{
|
||||
for(It t = beg + laneId(); t < end; t += STRIDE)
|
||||
*t = value;
|
||||
}
|
||||
|
||||
template<typename InIt, typename OutIt>
|
||||
static __device__ __forceinline__ OutIt copy(InIt beg, InIt end, OutIt out)
|
||||
{
|
||||
for(InIt t = beg + laneId(); t < end; t += STRIDE, out += STRIDE)
|
||||
*out = *t;
|
||||
return out;
|
||||
}
|
||||
|
||||
template<typename InIt, typename OutIt, class UnOp>
|
||||
static __device__ __forceinline__ OutIt transform(InIt beg, InIt end, OutIt out, UnOp op)
|
||||
{
|
||||
for(InIt t = beg + laneId(); t < end; t += STRIDE, out += STRIDE)
|
||||
*out = op(*t);
|
||||
return out;
|
||||
}
|
||||
|
||||
template<typename InIt1, typename InIt2, typename OutIt, class BinOp>
|
||||
static __device__ __forceinline__ OutIt transform(InIt1 beg1, InIt1 end1, InIt2 beg2, OutIt out, BinOp op)
|
||||
{
|
||||
unsigned int lane = laneId();
|
||||
|
||||
InIt1 t1 = beg1 + lane;
|
||||
InIt2 t2 = beg2 + lane;
|
||||
for(; t1 < end1; t1 += STRIDE, t2 += STRIDE, out += STRIDE)
|
||||
*out = op(*t1, *t2);
|
||||
return out;
|
||||
}
|
||||
|
||||
template <class T, class BinOp>
|
||||
static __device__ __forceinline__ T reduce(volatile T *ptr, BinOp op)
|
||||
{
|
||||
const unsigned int lane = laneId();
|
||||
|
||||
if (lane < 16)
|
||||
{
|
||||
T partial = ptr[lane];
|
||||
|
||||
ptr[lane] = partial = op(partial, ptr[lane + 16]);
|
||||
ptr[lane] = partial = op(partial, ptr[lane + 8]);
|
||||
ptr[lane] = partial = op(partial, ptr[lane + 4]);
|
||||
ptr[lane] = partial = op(partial, ptr[lane + 2]);
|
||||
ptr[lane] = partial = op(partial, ptr[lane + 1]);
|
||||
}
|
||||
|
||||
return *ptr;
|
||||
}
|
||||
|
||||
template<typename OutIt, typename T>
|
||||
static __device__ __forceinline__ void yota(OutIt beg, OutIt end, T value)
|
||||
{
|
||||
unsigned int lane = laneId();
|
||||
value += lane;
|
||||
|
||||
for(OutIt t = beg + lane; t < end; t += STRIDE, value += STRIDE)
|
||||
*t = value;
|
||||
}
|
||||
};
|
||||
}}} // namespace cv { namespace cuda { namespace cudev
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif /* OPENCV_CUDA_DEVICE_WARP_HPP */
|
||||
|
|
@ -0,0 +1,76 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_WARP_REDUCE_HPP__
|
||||
#define OPENCV_CUDA_WARP_REDUCE_HPP__
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
template <class T>
|
||||
__device__ __forceinline__ T warp_reduce(volatile T *ptr , const unsigned int tid = threadIdx.x)
|
||||
{
|
||||
const unsigned int lane = tid & 31; // index of thread in warp (0..31)
|
||||
|
||||
if (lane < 16)
|
||||
{
|
||||
T partial = ptr[tid];
|
||||
|
||||
ptr[tid] = partial = partial + ptr[tid + 16];
|
||||
ptr[tid] = partial = partial + ptr[tid + 8];
|
||||
ptr[tid] = partial = partial + ptr[tid + 4];
|
||||
ptr[tid] = partial = partial + ptr[tid + 2];
|
||||
ptr[tid] = partial = partial + ptr[tid + 1];
|
||||
}
|
||||
|
||||
return ptr[tid - lane];
|
||||
}
|
||||
}}} // namespace cv { namespace cuda { namespace cudev {
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif /* OPENCV_CUDA_WARP_REDUCE_HPP__ */
|
||||
|
|
@ -0,0 +1,162 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CUDA_WARP_SHUFFLE_HPP
|
||||
#define OPENCV_CUDA_WARP_SHUFFLE_HPP
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv { namespace cuda { namespace device
|
||||
{
|
||||
#if __CUDACC_VER_MAJOR__ >= 9
|
||||
# define __shfl(x, y, z) __shfl_sync(0xFFFFFFFFU, x, y, z)
|
||||
# define __shfl_up(x, y, z) __shfl_up_sync(0xFFFFFFFFU, x, y, z)
|
||||
# define __shfl_down(x, y, z) __shfl_down_sync(0xFFFFFFFFU, x, y, z)
|
||||
#endif
|
||||
template <typename T>
|
||||
__device__ __forceinline__ T shfl(T val, int srcLane, int width = warpSize)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
return __shfl(val, srcLane, width);
|
||||
#else
|
||||
return T();
|
||||
#endif
|
||||
}
|
||||
__device__ __forceinline__ unsigned int shfl(unsigned int val, int srcLane, int width = warpSize)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
return (unsigned int) __shfl((int) val, srcLane, width);
|
||||
#else
|
||||
return 0;
|
||||
#endif
|
||||
}
|
||||
__device__ __forceinline__ double shfl(double val, int srcLane, int width = warpSize)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
int lo = __double2loint(val);
|
||||
int hi = __double2hiint(val);
|
||||
|
||||
lo = __shfl(lo, srcLane, width);
|
||||
hi = __shfl(hi, srcLane, width);
|
||||
|
||||
return __hiloint2double(hi, lo);
|
||||
#else
|
||||
return 0.0;
|
||||
#endif
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
__device__ __forceinline__ T shfl_down(T val, unsigned int delta, int width = warpSize)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
return __shfl_down(val, delta, width);
|
||||
#else
|
||||
return T();
|
||||
#endif
|
||||
}
|
||||
__device__ __forceinline__ unsigned int shfl_down(unsigned int val, unsigned int delta, int width = warpSize)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
return (unsigned int) __shfl_down((int) val, delta, width);
|
||||
#else
|
||||
return 0;
|
||||
#endif
|
||||
}
|
||||
__device__ __forceinline__ double shfl_down(double val, unsigned int delta, int width = warpSize)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
int lo = __double2loint(val);
|
||||
int hi = __double2hiint(val);
|
||||
|
||||
lo = __shfl_down(lo, delta, width);
|
||||
hi = __shfl_down(hi, delta, width);
|
||||
|
||||
return __hiloint2double(hi, lo);
|
||||
#else
|
||||
return 0.0;
|
||||
#endif
|
||||
}
|
||||
|
||||
template <typename T>
|
||||
__device__ __forceinline__ T shfl_up(T val, unsigned int delta, int width = warpSize)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
return __shfl_up(val, delta, width);
|
||||
#else
|
||||
return T();
|
||||
#endif
|
||||
}
|
||||
__device__ __forceinline__ unsigned int shfl_up(unsigned int val, unsigned int delta, int width = warpSize)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
return (unsigned int) __shfl_up((int) val, delta, width);
|
||||
#else
|
||||
return 0;
|
||||
#endif
|
||||
}
|
||||
__device__ __forceinline__ double shfl_up(double val, unsigned int delta, int width = warpSize)
|
||||
{
|
||||
#if defined __CUDA_ARCH__ && __CUDA_ARCH__ >= 300
|
||||
int lo = __double2loint(val);
|
||||
int hi = __double2hiint(val);
|
||||
|
||||
lo = __shfl_up(lo, delta, width);
|
||||
hi = __shfl_up(hi, delta, width);
|
||||
|
||||
return __hiloint2double(hi, lo);
|
||||
#else
|
||||
return 0.0;
|
||||
#endif
|
||||
}
|
||||
}}}
|
||||
|
||||
# undef __shfl
|
||||
# undef __shfl_up
|
||||
# undef __shfl_down
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CUDA_WARP_SHUFFLE_HPP
|
||||
|
|
@ -0,0 +1,86 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CORE_CUDA_STREAM_ACCESSOR_HPP
|
||||
#define OPENCV_CORE_CUDA_STREAM_ACCESSOR_HPP
|
||||
|
||||
#ifndef __cplusplus
|
||||
# error cuda_stream_accessor.hpp header must be compiled as C++
|
||||
#endif
|
||||
|
||||
/** @file cuda_stream_accessor.hpp
|
||||
* This is only header file that depends on CUDA Runtime API. All other headers are independent.
|
||||
*/
|
||||
|
||||
#include <cuda_runtime.h>
|
||||
#include "opencv2/core/cuda.hpp"
|
||||
|
||||
namespace cv
|
||||
{
|
||||
namespace cuda
|
||||
{
|
||||
|
||||
//! @addtogroup cudacore_struct
|
||||
//! @{
|
||||
|
||||
/** @brief Class that enables getting cudaStream_t from cuda::Stream
|
||||
*/
|
||||
struct StreamAccessor
|
||||
{
|
||||
CV_EXPORTS static cudaStream_t getStream(const Stream& stream);
|
||||
CV_EXPORTS static Stream wrapStream(cudaStream_t stream);
|
||||
};
|
||||
|
||||
/** @brief Class that enables getting cudaEvent_t from cuda::Event
|
||||
*/
|
||||
struct EventAccessor
|
||||
{
|
||||
CV_EXPORTS static cudaEvent_t getEvent(const Event& event);
|
||||
CV_EXPORTS static Event wrapEvent(cudaEvent_t event);
|
||||
};
|
||||
|
||||
//! @}
|
||||
|
||||
}
|
||||
}
|
||||
|
||||
#endif /* OPENCV_CORE_CUDA_STREAM_ACCESSOR_HPP */
|
||||
|
|
@ -0,0 +1,142 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CORE_CUDA_TYPES_HPP
|
||||
#define OPENCV_CORE_CUDA_TYPES_HPP
|
||||
|
||||
#ifndef __cplusplus
|
||||
# error cuda_types.hpp header must be compiled as C++
|
||||
#endif
|
||||
|
||||
#if defined(__OPENCV_BUILD) && defined(__clang__)
|
||||
#pragma clang diagnostic ignored "-Winconsistent-missing-override"
|
||||
#endif
|
||||
#if defined(__OPENCV_BUILD) && defined(__GNUC__) && __GNUC__ >= 5
|
||||
#pragma GCC diagnostic ignored "-Wsuggest-override"
|
||||
#endif
|
||||
|
||||
/** @file
|
||||
* @deprecated Use @ref cudev instead.
|
||||
*/
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
#ifdef __CUDACC__
|
||||
#define __CV_CUDA_HOST_DEVICE__ __host__ __device__ __forceinline__
|
||||
#else
|
||||
#define __CV_CUDA_HOST_DEVICE__
|
||||
#endif
|
||||
|
||||
namespace cv
|
||||
{
|
||||
namespace cuda
|
||||
{
|
||||
|
||||
// Simple lightweight structures that encapsulates information about an image on device.
|
||||
// It is intended to pass to nvcc-compiled code. GpuMat depends on headers that nvcc can't compile
|
||||
|
||||
template <typename T> struct DevPtr
|
||||
{
|
||||
typedef T elem_type;
|
||||
typedef int index_type;
|
||||
|
||||
enum { elem_size = sizeof(elem_type) };
|
||||
|
||||
T* data;
|
||||
|
||||
__CV_CUDA_HOST_DEVICE__ DevPtr() : data(0) {}
|
||||
__CV_CUDA_HOST_DEVICE__ DevPtr(T* data_) : data(data_) {}
|
||||
|
||||
__CV_CUDA_HOST_DEVICE__ size_t elemSize() const { return elem_size; }
|
||||
__CV_CUDA_HOST_DEVICE__ operator T*() { return data; }
|
||||
__CV_CUDA_HOST_DEVICE__ operator const T*() const { return data; }
|
||||
};
|
||||
|
||||
template <typename T> struct PtrSz : public DevPtr<T>
|
||||
{
|
||||
__CV_CUDA_HOST_DEVICE__ PtrSz() : size(0) {}
|
||||
__CV_CUDA_HOST_DEVICE__ PtrSz(T* data_, size_t size_) : DevPtr<T>(data_), size(size_) {}
|
||||
|
||||
size_t size;
|
||||
};
|
||||
|
||||
template <typename T> struct PtrStep : public DevPtr<T>
|
||||
{
|
||||
__CV_CUDA_HOST_DEVICE__ PtrStep() : step(0) {}
|
||||
__CV_CUDA_HOST_DEVICE__ PtrStep(T* data_, size_t step_) : DevPtr<T>(data_), step(step_) {}
|
||||
|
||||
size_t step;
|
||||
|
||||
__CV_CUDA_HOST_DEVICE__ T* ptr(int y = 0) { return ( T*)( ( char*)DevPtr<T>::data + y * step); }
|
||||
__CV_CUDA_HOST_DEVICE__ const T* ptr(int y = 0) const { return (const T*)( (const char*)DevPtr<T>::data + y * step); }
|
||||
|
||||
__CV_CUDA_HOST_DEVICE__ T& operator ()(int y, int x) { return ptr(y)[x]; }
|
||||
__CV_CUDA_HOST_DEVICE__ const T& operator ()(int y, int x) const { return ptr(y)[x]; }
|
||||
};
|
||||
|
||||
template <typename T> struct PtrStepSz : public PtrStep<T>
|
||||
{
|
||||
__CV_CUDA_HOST_DEVICE__ PtrStepSz() : cols(0), rows(0) {}
|
||||
__CV_CUDA_HOST_DEVICE__ PtrStepSz(int rows_, int cols_, T* data_, size_t step_)
|
||||
: PtrStep<T>(data_, step_), cols(cols_), rows(rows_) {}
|
||||
|
||||
template <typename U>
|
||||
explicit PtrStepSz(const PtrStepSz<U>& d) : PtrStep<T>((T*)d.data, d.step), cols(d.cols), rows(d.rows){}
|
||||
|
||||
int cols;
|
||||
int rows;
|
||||
};
|
||||
|
||||
typedef PtrStepSz<unsigned char> PtrStepSzb;
|
||||
typedef PtrStepSz<float> PtrStepSzf;
|
||||
typedef PtrStepSz<int> PtrStepSzi;
|
||||
|
||||
typedef PtrStep<unsigned char> PtrStepb;
|
||||
typedef PtrStep<float> PtrStepf;
|
||||
typedef PtrStep<int> PtrStepi;
|
||||
|
||||
}
|
||||
}
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif /* OPENCV_CORE_CUDA_TYPES_HPP */
|
||||
|
|
@ -0,0 +1,247 @@
|
|||
// This file is part of OpenCV project.
|
||||
// It is subject to the license terms in the LICENSE file found in the top-level directory
|
||||
// of this distribution and at http://opencv.org/license.html.
|
||||
|
||||
#if defined __OPENCV_BUILD \
|
||||
|
||||
#include "cv_cpu_config.h"
|
||||
#include "cv_cpu_helper.h"
|
||||
|
||||
#ifdef CV_CPU_DISPATCH_MODE
|
||||
#define CV_CPU_OPTIMIZATION_NAMESPACE __CV_CAT(opt_, CV_CPU_DISPATCH_MODE)
|
||||
#define CV_CPU_OPTIMIZATION_NAMESPACE_BEGIN namespace __CV_CAT(opt_, CV_CPU_DISPATCH_MODE) {
|
||||
#define CV_CPU_OPTIMIZATION_NAMESPACE_END }
|
||||
#else
|
||||
#define CV_CPU_OPTIMIZATION_NAMESPACE cpu_baseline
|
||||
#define CV_CPU_OPTIMIZATION_NAMESPACE_BEGIN namespace cpu_baseline {
|
||||
#define CV_CPU_OPTIMIZATION_NAMESPACE_END }
|
||||
#endif
|
||||
|
||||
|
||||
#define __CV_CPU_DISPATCH_CHAIN_END(fn, args, mode, ...) /* done */
|
||||
#define __CV_CPU_DISPATCH(fn, args, mode, ...) __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
#define __CV_CPU_DISPATCH_EXPAND(fn, args, ...) __CV_EXPAND(__CV_CPU_DISPATCH(fn, args, __VA_ARGS__))
|
||||
#define CV_CPU_DISPATCH(fn, args, ...) __CV_CPU_DISPATCH_EXPAND(fn, args, __VA_ARGS__, END) // expand macros
|
||||
|
||||
|
||||
#if defined CV_ENABLE_INTRINSICS \
|
||||
&& !defined CV_DISABLE_OPTIMIZATION \
|
||||
&& !defined __CUDACC__ /* do not include SSE/AVX/NEON headers for NVCC compiler */ \
|
||||
|
||||
#ifdef CV_CPU_COMPILE_SSE2
|
||||
# include <emmintrin.h>
|
||||
# define CV_MMX 1
|
||||
# define CV_SSE 1
|
||||
# define CV_SSE2 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_SSE3
|
||||
# include <pmmintrin.h>
|
||||
# define CV_SSE3 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_SSSE3
|
||||
# include <tmmintrin.h>
|
||||
# define CV_SSSE3 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_SSE4_1
|
||||
# include <smmintrin.h>
|
||||
# define CV_SSE4_1 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_SSE4_2
|
||||
# include <nmmintrin.h>
|
||||
# define CV_SSE4_2 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_POPCNT
|
||||
# ifdef _MSC_VER
|
||||
# include <nmmintrin.h>
|
||||
# if defined(_M_X64)
|
||||
# define CV_POPCNT_U64 _mm_popcnt_u64
|
||||
# endif
|
||||
# define CV_POPCNT_U32 _mm_popcnt_u32
|
||||
# else
|
||||
# include <popcntintrin.h>
|
||||
# if defined(__x86_64__)
|
||||
# define CV_POPCNT_U64 __builtin_popcountll
|
||||
# endif
|
||||
# define CV_POPCNT_U32 __builtin_popcount
|
||||
# endif
|
||||
# define CV_POPCNT 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_AVX
|
||||
# include <immintrin.h>
|
||||
# define CV_AVX 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_FP16
|
||||
# if defined(__arm__) || defined(__aarch64__) || defined(_M_ARM)
|
||||
# include <arm_neon.h>
|
||||
# else
|
||||
# include <immintrin.h>
|
||||
# endif
|
||||
# define CV_FP16 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_AVX2
|
||||
# include <immintrin.h>
|
||||
# define CV_AVX2 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_AVX_512F
|
||||
# include <immintrin.h>
|
||||
# define CV_AVX_512F 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_AVX512_SKX
|
||||
# include <immintrin.h>
|
||||
# define CV_AVX512_SKX 1
|
||||
#endif
|
||||
#ifdef CV_CPU_COMPILE_FMA3
|
||||
# define CV_FMA3 1
|
||||
#endif
|
||||
|
||||
#if defined _WIN32 && defined(_M_ARM)
|
||||
# include <Intrin.h>
|
||||
# include <arm_neon.h>
|
||||
# define CV_NEON 1
|
||||
#elif defined(__ARM_NEON__) || (defined (__ARM_NEON) && defined(__aarch64__))
|
||||
# include <arm_neon.h>
|
||||
# define CV_NEON 1
|
||||
#endif
|
||||
|
||||
#if defined(__ARM_NEON__) || defined(__aarch64__)
|
||||
# include <arm_neon.h>
|
||||
#endif
|
||||
|
||||
#ifdef CV_CPU_COMPILE_VSX
|
||||
# include <altivec.h>
|
||||
# undef vector
|
||||
# undef pixel
|
||||
# undef bool
|
||||
# define CV_VSX 1
|
||||
#endif
|
||||
|
||||
#ifdef CV_CPU_COMPILE_VSX3
|
||||
# define CV_VSX3 1
|
||||
#endif
|
||||
|
||||
#endif // CV_ENABLE_INTRINSICS && !CV_DISABLE_OPTIMIZATION && !__CUDACC__
|
||||
|
||||
#if defined CV_CPU_COMPILE_AVX && !defined CV_CPU_BASELINE_COMPILE_AVX
|
||||
struct VZeroUpperGuard {
|
||||
#ifdef __GNUC__
|
||||
__attribute__((always_inline))
|
||||
#endif
|
||||
inline ~VZeroUpperGuard() { _mm256_zeroupper(); }
|
||||
};
|
||||
#define __CV_AVX_GUARD VZeroUpperGuard __vzeroupper_guard; CV_UNUSED(__vzeroupper_guard);
|
||||
#endif
|
||||
|
||||
#ifdef __CV_AVX_GUARD
|
||||
#define CV_AVX_GUARD __CV_AVX_GUARD
|
||||
#else
|
||||
#define CV_AVX_GUARD
|
||||
#endif
|
||||
|
||||
#endif // __OPENCV_BUILD
|
||||
|
||||
|
||||
|
||||
#if !defined __OPENCV_BUILD /* Compatibility code */ \
|
||||
&& !defined __CUDACC__ /* do not include SSE/AVX/NEON headers for NVCC compiler */
|
||||
#if defined __SSE2__ || defined _M_X64 || (defined _M_IX86_FP && _M_IX86_FP >= 2)
|
||||
# include <emmintrin.h>
|
||||
# define CV_MMX 1
|
||||
# define CV_SSE 1
|
||||
# define CV_SSE2 1
|
||||
#elif defined _WIN32 && defined(_M_ARM)
|
||||
# include <Intrin.h>
|
||||
# include <arm_neon.h>
|
||||
# define CV_NEON 1
|
||||
#elif defined(__ARM_NEON__) || (defined (__ARM_NEON) && defined(__aarch64__))
|
||||
# include <arm_neon.h>
|
||||
# define CV_NEON 1
|
||||
#elif defined(__VSX__) && defined(__PPC64__) && defined(__LITTLE_ENDIAN__)
|
||||
# include <altivec.h>
|
||||
# undef vector
|
||||
# undef pixel
|
||||
# undef bool
|
||||
# define CV_VSX 1
|
||||
#endif
|
||||
|
||||
#endif // !__OPENCV_BUILD && !__CUDACC (Compatibility code)
|
||||
|
||||
|
||||
|
||||
#ifndef CV_MMX
|
||||
# define CV_MMX 0
|
||||
#endif
|
||||
#ifndef CV_SSE
|
||||
# define CV_SSE 0
|
||||
#endif
|
||||
#ifndef CV_SSE2
|
||||
# define CV_SSE2 0
|
||||
#endif
|
||||
#ifndef CV_SSE3
|
||||
# define CV_SSE3 0
|
||||
#endif
|
||||
#ifndef CV_SSSE3
|
||||
# define CV_SSSE3 0
|
||||
#endif
|
||||
#ifndef CV_SSE4_1
|
||||
# define CV_SSE4_1 0
|
||||
#endif
|
||||
#ifndef CV_SSE4_2
|
||||
# define CV_SSE4_2 0
|
||||
#endif
|
||||
#ifndef CV_POPCNT
|
||||
# define CV_POPCNT 0
|
||||
#endif
|
||||
#ifndef CV_AVX
|
||||
# define CV_AVX 0
|
||||
#endif
|
||||
#ifndef CV_FP16
|
||||
# define CV_FP16 0
|
||||
#endif
|
||||
#ifndef CV_AVX2
|
||||
# define CV_AVX2 0
|
||||
#endif
|
||||
#ifndef CV_FMA3
|
||||
# define CV_FMA3 0
|
||||
#endif
|
||||
#ifndef CV_AVX_512F
|
||||
# define CV_AVX_512F 0
|
||||
#endif
|
||||
#ifndef CV_AVX_512BW
|
||||
# define CV_AVX_512BW 0
|
||||
#endif
|
||||
#ifndef CV_AVX_512CD
|
||||
# define CV_AVX_512CD 0
|
||||
#endif
|
||||
#ifndef CV_AVX_512DQ
|
||||
# define CV_AVX_512DQ 0
|
||||
#endif
|
||||
#ifndef CV_AVX_512ER
|
||||
# define CV_AVX_512ER 0
|
||||
#endif
|
||||
#ifndef CV_AVX_512IFMA512
|
||||
# define CV_AVX_512IFMA512 0
|
||||
#endif
|
||||
#ifndef CV_AVX_512PF
|
||||
# define CV_AVX_512PF 0
|
||||
#endif
|
||||
#ifndef CV_AVX_512VBMI
|
||||
# define CV_AVX_512VBMI 0
|
||||
#endif
|
||||
#ifndef CV_AVX_512VL
|
||||
# define CV_AVX_512VL 0
|
||||
#endif
|
||||
#ifndef CV_AVX512_SKX
|
||||
# define CV_AVX512_SKX 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_NEON
|
||||
# define CV_NEON 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_VSX
|
||||
# define CV_VSX 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_VSX3
|
||||
# define CV_VSX3 0
|
||||
#endif
|
||||
|
|
@ -0,0 +1,340 @@
|
|||
// AUTOGENERATED, DO NOT EDIT
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_SSE
|
||||
# define CV_TRY_SSE 1
|
||||
# define CV_CPU_FORCE_SSE 1
|
||||
# define CV_CPU_HAS_SUPPORT_SSE 1
|
||||
# define CV_CPU_CALL_SSE(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_SSE_(fn, args) return (opt_SSE::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_SSE
|
||||
# define CV_TRY_SSE 1
|
||||
# define CV_CPU_FORCE_SSE 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSE (cv::checkHardwareSupport(CV_CPU_SSE))
|
||||
# define CV_CPU_CALL_SSE(fn, args) if (CV_CPU_HAS_SUPPORT_SSE) return (opt_SSE::fn args)
|
||||
# define CV_CPU_CALL_SSE_(fn, args) if (CV_CPU_HAS_SUPPORT_SSE) return (opt_SSE::fn args)
|
||||
#else
|
||||
# define CV_TRY_SSE 0
|
||||
# define CV_CPU_FORCE_SSE 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSE 0
|
||||
# define CV_CPU_CALL_SSE(fn, args)
|
||||
# define CV_CPU_CALL_SSE_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_SSE(fn, args, mode, ...) CV_CPU_CALL_SSE(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_SSE2
|
||||
# define CV_TRY_SSE2 1
|
||||
# define CV_CPU_FORCE_SSE2 1
|
||||
# define CV_CPU_HAS_SUPPORT_SSE2 1
|
||||
# define CV_CPU_CALL_SSE2(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_SSE2_(fn, args) return (opt_SSE2::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_SSE2
|
||||
# define CV_TRY_SSE2 1
|
||||
# define CV_CPU_FORCE_SSE2 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSE2 (cv::checkHardwareSupport(CV_CPU_SSE2))
|
||||
# define CV_CPU_CALL_SSE2(fn, args) if (CV_CPU_HAS_SUPPORT_SSE2) return (opt_SSE2::fn args)
|
||||
# define CV_CPU_CALL_SSE2_(fn, args) if (CV_CPU_HAS_SUPPORT_SSE2) return (opt_SSE2::fn args)
|
||||
#else
|
||||
# define CV_TRY_SSE2 0
|
||||
# define CV_CPU_FORCE_SSE2 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSE2 0
|
||||
# define CV_CPU_CALL_SSE2(fn, args)
|
||||
# define CV_CPU_CALL_SSE2_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_SSE2(fn, args, mode, ...) CV_CPU_CALL_SSE2(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_SSE3
|
||||
# define CV_TRY_SSE3 1
|
||||
# define CV_CPU_FORCE_SSE3 1
|
||||
# define CV_CPU_HAS_SUPPORT_SSE3 1
|
||||
# define CV_CPU_CALL_SSE3(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_SSE3_(fn, args) return (opt_SSE3::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_SSE3
|
||||
# define CV_TRY_SSE3 1
|
||||
# define CV_CPU_FORCE_SSE3 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSE3 (cv::checkHardwareSupport(CV_CPU_SSE3))
|
||||
# define CV_CPU_CALL_SSE3(fn, args) if (CV_CPU_HAS_SUPPORT_SSE3) return (opt_SSE3::fn args)
|
||||
# define CV_CPU_CALL_SSE3_(fn, args) if (CV_CPU_HAS_SUPPORT_SSE3) return (opt_SSE3::fn args)
|
||||
#else
|
||||
# define CV_TRY_SSE3 0
|
||||
# define CV_CPU_FORCE_SSE3 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSE3 0
|
||||
# define CV_CPU_CALL_SSE3(fn, args)
|
||||
# define CV_CPU_CALL_SSE3_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_SSE3(fn, args, mode, ...) CV_CPU_CALL_SSE3(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_SSSE3
|
||||
# define CV_TRY_SSSE3 1
|
||||
# define CV_CPU_FORCE_SSSE3 1
|
||||
# define CV_CPU_HAS_SUPPORT_SSSE3 1
|
||||
# define CV_CPU_CALL_SSSE3(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_SSSE3_(fn, args) return (opt_SSSE3::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_SSSE3
|
||||
# define CV_TRY_SSSE3 1
|
||||
# define CV_CPU_FORCE_SSSE3 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSSE3 (cv::checkHardwareSupport(CV_CPU_SSSE3))
|
||||
# define CV_CPU_CALL_SSSE3(fn, args) if (CV_CPU_HAS_SUPPORT_SSSE3) return (opt_SSSE3::fn args)
|
||||
# define CV_CPU_CALL_SSSE3_(fn, args) if (CV_CPU_HAS_SUPPORT_SSSE3) return (opt_SSSE3::fn args)
|
||||
#else
|
||||
# define CV_TRY_SSSE3 0
|
||||
# define CV_CPU_FORCE_SSSE3 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSSE3 0
|
||||
# define CV_CPU_CALL_SSSE3(fn, args)
|
||||
# define CV_CPU_CALL_SSSE3_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_SSSE3(fn, args, mode, ...) CV_CPU_CALL_SSSE3(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_SSE4_1
|
||||
# define CV_TRY_SSE4_1 1
|
||||
# define CV_CPU_FORCE_SSE4_1 1
|
||||
# define CV_CPU_HAS_SUPPORT_SSE4_1 1
|
||||
# define CV_CPU_CALL_SSE4_1(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_SSE4_1_(fn, args) return (opt_SSE4_1::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_SSE4_1
|
||||
# define CV_TRY_SSE4_1 1
|
||||
# define CV_CPU_FORCE_SSE4_1 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSE4_1 (cv::checkHardwareSupport(CV_CPU_SSE4_1))
|
||||
# define CV_CPU_CALL_SSE4_1(fn, args) if (CV_CPU_HAS_SUPPORT_SSE4_1) return (opt_SSE4_1::fn args)
|
||||
# define CV_CPU_CALL_SSE4_1_(fn, args) if (CV_CPU_HAS_SUPPORT_SSE4_1) return (opt_SSE4_1::fn args)
|
||||
#else
|
||||
# define CV_TRY_SSE4_1 0
|
||||
# define CV_CPU_FORCE_SSE4_1 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSE4_1 0
|
||||
# define CV_CPU_CALL_SSE4_1(fn, args)
|
||||
# define CV_CPU_CALL_SSE4_1_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_SSE4_1(fn, args, mode, ...) CV_CPU_CALL_SSE4_1(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_SSE4_2
|
||||
# define CV_TRY_SSE4_2 1
|
||||
# define CV_CPU_FORCE_SSE4_2 1
|
||||
# define CV_CPU_HAS_SUPPORT_SSE4_2 1
|
||||
# define CV_CPU_CALL_SSE4_2(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_SSE4_2_(fn, args) return (opt_SSE4_2::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_SSE4_2
|
||||
# define CV_TRY_SSE4_2 1
|
||||
# define CV_CPU_FORCE_SSE4_2 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSE4_2 (cv::checkHardwareSupport(CV_CPU_SSE4_2))
|
||||
# define CV_CPU_CALL_SSE4_2(fn, args) if (CV_CPU_HAS_SUPPORT_SSE4_2) return (opt_SSE4_2::fn args)
|
||||
# define CV_CPU_CALL_SSE4_2_(fn, args) if (CV_CPU_HAS_SUPPORT_SSE4_2) return (opt_SSE4_2::fn args)
|
||||
#else
|
||||
# define CV_TRY_SSE4_2 0
|
||||
# define CV_CPU_FORCE_SSE4_2 0
|
||||
# define CV_CPU_HAS_SUPPORT_SSE4_2 0
|
||||
# define CV_CPU_CALL_SSE4_2(fn, args)
|
||||
# define CV_CPU_CALL_SSE4_2_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_SSE4_2(fn, args, mode, ...) CV_CPU_CALL_SSE4_2(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_POPCNT
|
||||
# define CV_TRY_POPCNT 1
|
||||
# define CV_CPU_FORCE_POPCNT 1
|
||||
# define CV_CPU_HAS_SUPPORT_POPCNT 1
|
||||
# define CV_CPU_CALL_POPCNT(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_POPCNT_(fn, args) return (opt_POPCNT::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_POPCNT
|
||||
# define CV_TRY_POPCNT 1
|
||||
# define CV_CPU_FORCE_POPCNT 0
|
||||
# define CV_CPU_HAS_SUPPORT_POPCNT (cv::checkHardwareSupport(CV_CPU_POPCNT))
|
||||
# define CV_CPU_CALL_POPCNT(fn, args) if (CV_CPU_HAS_SUPPORT_POPCNT) return (opt_POPCNT::fn args)
|
||||
# define CV_CPU_CALL_POPCNT_(fn, args) if (CV_CPU_HAS_SUPPORT_POPCNT) return (opt_POPCNT::fn args)
|
||||
#else
|
||||
# define CV_TRY_POPCNT 0
|
||||
# define CV_CPU_FORCE_POPCNT 0
|
||||
# define CV_CPU_HAS_SUPPORT_POPCNT 0
|
||||
# define CV_CPU_CALL_POPCNT(fn, args)
|
||||
# define CV_CPU_CALL_POPCNT_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_POPCNT(fn, args, mode, ...) CV_CPU_CALL_POPCNT(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_AVX
|
||||
# define CV_TRY_AVX 1
|
||||
# define CV_CPU_FORCE_AVX 1
|
||||
# define CV_CPU_HAS_SUPPORT_AVX 1
|
||||
# define CV_CPU_CALL_AVX(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_AVX_(fn, args) return (opt_AVX::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_AVX
|
||||
# define CV_TRY_AVX 1
|
||||
# define CV_CPU_FORCE_AVX 0
|
||||
# define CV_CPU_HAS_SUPPORT_AVX (cv::checkHardwareSupport(CV_CPU_AVX))
|
||||
# define CV_CPU_CALL_AVX(fn, args) if (CV_CPU_HAS_SUPPORT_AVX) return (opt_AVX::fn args)
|
||||
# define CV_CPU_CALL_AVX_(fn, args) if (CV_CPU_HAS_SUPPORT_AVX) return (opt_AVX::fn args)
|
||||
#else
|
||||
# define CV_TRY_AVX 0
|
||||
# define CV_CPU_FORCE_AVX 0
|
||||
# define CV_CPU_HAS_SUPPORT_AVX 0
|
||||
# define CV_CPU_CALL_AVX(fn, args)
|
||||
# define CV_CPU_CALL_AVX_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_AVX(fn, args, mode, ...) CV_CPU_CALL_AVX(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_FP16
|
||||
# define CV_TRY_FP16 1
|
||||
# define CV_CPU_FORCE_FP16 1
|
||||
# define CV_CPU_HAS_SUPPORT_FP16 1
|
||||
# define CV_CPU_CALL_FP16(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_FP16_(fn, args) return (opt_FP16::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_FP16
|
||||
# define CV_TRY_FP16 1
|
||||
# define CV_CPU_FORCE_FP16 0
|
||||
# define CV_CPU_HAS_SUPPORT_FP16 (cv::checkHardwareSupport(CV_CPU_FP16))
|
||||
# define CV_CPU_CALL_FP16(fn, args) if (CV_CPU_HAS_SUPPORT_FP16) return (opt_FP16::fn args)
|
||||
# define CV_CPU_CALL_FP16_(fn, args) if (CV_CPU_HAS_SUPPORT_FP16) return (opt_FP16::fn args)
|
||||
#else
|
||||
# define CV_TRY_FP16 0
|
||||
# define CV_CPU_FORCE_FP16 0
|
||||
# define CV_CPU_HAS_SUPPORT_FP16 0
|
||||
# define CV_CPU_CALL_FP16(fn, args)
|
||||
# define CV_CPU_CALL_FP16_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_FP16(fn, args, mode, ...) CV_CPU_CALL_FP16(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_AVX2
|
||||
# define CV_TRY_AVX2 1
|
||||
# define CV_CPU_FORCE_AVX2 1
|
||||
# define CV_CPU_HAS_SUPPORT_AVX2 1
|
||||
# define CV_CPU_CALL_AVX2(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_AVX2_(fn, args) return (opt_AVX2::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_AVX2
|
||||
# define CV_TRY_AVX2 1
|
||||
# define CV_CPU_FORCE_AVX2 0
|
||||
# define CV_CPU_HAS_SUPPORT_AVX2 (cv::checkHardwareSupport(CV_CPU_AVX2))
|
||||
# define CV_CPU_CALL_AVX2(fn, args) if (CV_CPU_HAS_SUPPORT_AVX2) return (opt_AVX2::fn args)
|
||||
# define CV_CPU_CALL_AVX2_(fn, args) if (CV_CPU_HAS_SUPPORT_AVX2) return (opt_AVX2::fn args)
|
||||
#else
|
||||
# define CV_TRY_AVX2 0
|
||||
# define CV_CPU_FORCE_AVX2 0
|
||||
# define CV_CPU_HAS_SUPPORT_AVX2 0
|
||||
# define CV_CPU_CALL_AVX2(fn, args)
|
||||
# define CV_CPU_CALL_AVX2_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_AVX2(fn, args, mode, ...) CV_CPU_CALL_AVX2(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_FMA3
|
||||
# define CV_TRY_FMA3 1
|
||||
# define CV_CPU_FORCE_FMA3 1
|
||||
# define CV_CPU_HAS_SUPPORT_FMA3 1
|
||||
# define CV_CPU_CALL_FMA3(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_FMA3_(fn, args) return (opt_FMA3::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_FMA3
|
||||
# define CV_TRY_FMA3 1
|
||||
# define CV_CPU_FORCE_FMA3 0
|
||||
# define CV_CPU_HAS_SUPPORT_FMA3 (cv::checkHardwareSupport(CV_CPU_FMA3))
|
||||
# define CV_CPU_CALL_FMA3(fn, args) if (CV_CPU_HAS_SUPPORT_FMA3) return (opt_FMA3::fn args)
|
||||
# define CV_CPU_CALL_FMA3_(fn, args) if (CV_CPU_HAS_SUPPORT_FMA3) return (opt_FMA3::fn args)
|
||||
#else
|
||||
# define CV_TRY_FMA3 0
|
||||
# define CV_CPU_FORCE_FMA3 0
|
||||
# define CV_CPU_HAS_SUPPORT_FMA3 0
|
||||
# define CV_CPU_CALL_FMA3(fn, args)
|
||||
# define CV_CPU_CALL_FMA3_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_FMA3(fn, args, mode, ...) CV_CPU_CALL_FMA3(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_AVX_512F
|
||||
# define CV_TRY_AVX_512F 1
|
||||
# define CV_CPU_FORCE_AVX_512F 1
|
||||
# define CV_CPU_HAS_SUPPORT_AVX_512F 1
|
||||
# define CV_CPU_CALL_AVX_512F(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_AVX_512F_(fn, args) return (opt_AVX_512F::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_AVX_512F
|
||||
# define CV_TRY_AVX_512F 1
|
||||
# define CV_CPU_FORCE_AVX_512F 0
|
||||
# define CV_CPU_HAS_SUPPORT_AVX_512F (cv::checkHardwareSupport(CV_CPU_AVX_512F))
|
||||
# define CV_CPU_CALL_AVX_512F(fn, args) if (CV_CPU_HAS_SUPPORT_AVX_512F) return (opt_AVX_512F::fn args)
|
||||
# define CV_CPU_CALL_AVX_512F_(fn, args) if (CV_CPU_HAS_SUPPORT_AVX_512F) return (opt_AVX_512F::fn args)
|
||||
#else
|
||||
# define CV_TRY_AVX_512F 0
|
||||
# define CV_CPU_FORCE_AVX_512F 0
|
||||
# define CV_CPU_HAS_SUPPORT_AVX_512F 0
|
||||
# define CV_CPU_CALL_AVX_512F(fn, args)
|
||||
# define CV_CPU_CALL_AVX_512F_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_AVX_512F(fn, args, mode, ...) CV_CPU_CALL_AVX_512F(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_AVX512_SKX
|
||||
# define CV_TRY_AVX512_SKX 1
|
||||
# define CV_CPU_FORCE_AVX512_SKX 1
|
||||
# define CV_CPU_HAS_SUPPORT_AVX512_SKX 1
|
||||
# define CV_CPU_CALL_AVX512_SKX(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_AVX512_SKX_(fn, args) return (opt_AVX512_SKX::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_AVX512_SKX
|
||||
# define CV_TRY_AVX512_SKX 1
|
||||
# define CV_CPU_FORCE_AVX512_SKX 0
|
||||
# define CV_CPU_HAS_SUPPORT_AVX512_SKX (cv::checkHardwareSupport(CV_CPU_AVX512_SKX))
|
||||
# define CV_CPU_CALL_AVX512_SKX(fn, args) if (CV_CPU_HAS_SUPPORT_AVX512_SKX) return (opt_AVX512_SKX::fn args)
|
||||
# define CV_CPU_CALL_AVX512_SKX_(fn, args) if (CV_CPU_HAS_SUPPORT_AVX512_SKX) return (opt_AVX512_SKX::fn args)
|
||||
#else
|
||||
# define CV_TRY_AVX512_SKX 0
|
||||
# define CV_CPU_FORCE_AVX512_SKX 0
|
||||
# define CV_CPU_HAS_SUPPORT_AVX512_SKX 0
|
||||
# define CV_CPU_CALL_AVX512_SKX(fn, args)
|
||||
# define CV_CPU_CALL_AVX512_SKX_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_AVX512_SKX(fn, args, mode, ...) CV_CPU_CALL_AVX512_SKX(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_NEON
|
||||
# define CV_TRY_NEON 1
|
||||
# define CV_CPU_FORCE_NEON 1
|
||||
# define CV_CPU_HAS_SUPPORT_NEON 1
|
||||
# define CV_CPU_CALL_NEON(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_NEON_(fn, args) return (opt_NEON::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_NEON
|
||||
# define CV_TRY_NEON 1
|
||||
# define CV_CPU_FORCE_NEON 0
|
||||
# define CV_CPU_HAS_SUPPORT_NEON (cv::checkHardwareSupport(CV_CPU_NEON))
|
||||
# define CV_CPU_CALL_NEON(fn, args) if (CV_CPU_HAS_SUPPORT_NEON) return (opt_NEON::fn args)
|
||||
# define CV_CPU_CALL_NEON_(fn, args) if (CV_CPU_HAS_SUPPORT_NEON) return (opt_NEON::fn args)
|
||||
#else
|
||||
# define CV_TRY_NEON 0
|
||||
# define CV_CPU_FORCE_NEON 0
|
||||
# define CV_CPU_HAS_SUPPORT_NEON 0
|
||||
# define CV_CPU_CALL_NEON(fn, args)
|
||||
# define CV_CPU_CALL_NEON_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_NEON(fn, args, mode, ...) CV_CPU_CALL_NEON(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_VSX
|
||||
# define CV_TRY_VSX 1
|
||||
# define CV_CPU_FORCE_VSX 1
|
||||
# define CV_CPU_HAS_SUPPORT_VSX 1
|
||||
# define CV_CPU_CALL_VSX(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_VSX_(fn, args) return (opt_VSX::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_VSX
|
||||
# define CV_TRY_VSX 1
|
||||
# define CV_CPU_FORCE_VSX 0
|
||||
# define CV_CPU_HAS_SUPPORT_VSX (cv::checkHardwareSupport(CV_CPU_VSX))
|
||||
# define CV_CPU_CALL_VSX(fn, args) if (CV_CPU_HAS_SUPPORT_VSX) return (opt_VSX::fn args)
|
||||
# define CV_CPU_CALL_VSX_(fn, args) if (CV_CPU_HAS_SUPPORT_VSX) return (opt_VSX::fn args)
|
||||
#else
|
||||
# define CV_TRY_VSX 0
|
||||
# define CV_CPU_FORCE_VSX 0
|
||||
# define CV_CPU_HAS_SUPPORT_VSX 0
|
||||
# define CV_CPU_CALL_VSX(fn, args)
|
||||
# define CV_CPU_CALL_VSX_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_VSX(fn, args, mode, ...) CV_CPU_CALL_VSX(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#if !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_COMPILE_VSX3
|
||||
# define CV_TRY_VSX3 1
|
||||
# define CV_CPU_FORCE_VSX3 1
|
||||
# define CV_CPU_HAS_SUPPORT_VSX3 1
|
||||
# define CV_CPU_CALL_VSX3(fn, args) return (cpu_baseline::fn args)
|
||||
# define CV_CPU_CALL_VSX3_(fn, args) return (opt_VSX3::fn args)
|
||||
#elif !defined CV_DISABLE_OPTIMIZATION && defined CV_ENABLE_INTRINSICS && defined CV_CPU_DISPATCH_COMPILE_VSX3
|
||||
# define CV_TRY_VSX3 1
|
||||
# define CV_CPU_FORCE_VSX3 0
|
||||
# define CV_CPU_HAS_SUPPORT_VSX3 (cv::checkHardwareSupport(CV_CPU_VSX3))
|
||||
# define CV_CPU_CALL_VSX3(fn, args) if (CV_CPU_HAS_SUPPORT_VSX3) return (opt_VSX3::fn args)
|
||||
# define CV_CPU_CALL_VSX3_(fn, args) if (CV_CPU_HAS_SUPPORT_VSX3) return (opt_VSX3::fn args)
|
||||
#else
|
||||
# define CV_TRY_VSX3 0
|
||||
# define CV_CPU_FORCE_VSX3 0
|
||||
# define CV_CPU_HAS_SUPPORT_VSX3 0
|
||||
# define CV_CPU_CALL_VSX3(fn, args)
|
||||
# define CV_CPU_CALL_VSX3_(fn, args)
|
||||
#endif
|
||||
#define __CV_CPU_DISPATCH_CHAIN_VSX3(fn, args, mode, ...) CV_CPU_CALL_VSX3(fn, args); __CV_EXPAND(__CV_CPU_DISPATCH_CHAIN_ ## mode(fn, args, __VA_ARGS__))
|
||||
|
||||
#define CV_CPU_CALL_BASELINE(fn, args) return (cpu_baseline::fn args)
|
||||
#define __CV_CPU_DISPATCH_CHAIN_BASELINE(fn, args, mode, ...) CV_CPU_CALL_BASELINE(fn, args) /* last in sequence */
|
||||
|
|
@ -0,0 +1,753 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Copyright (C) 2015, Itseez Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CORE_CVDEF_H
|
||||
#define OPENCV_CORE_CVDEF_H
|
||||
|
||||
//! @addtogroup core_utils
|
||||
//! @{
|
||||
|
||||
#if !defined CV_DOXYGEN && !defined CV_IGNORE_DEBUG_BUILD_GUARD
|
||||
#if (defined(_MSC_VER) && (defined(DEBUG) || defined(_DEBUG))) || \
|
||||
(defined(_GLIBCXX_DEBUG) || defined(_GLIBCXX_DEBUG_PEDANTIC))
|
||||
// Guard to prevent using of binary incompatible binaries / runtimes
|
||||
// https://github.com/opencv/opencv/pull/9161
|
||||
#define CV__DEBUG_NS_BEGIN namespace debug_build_guard {
|
||||
#define CV__DEBUG_NS_END }
|
||||
namespace cv { namespace debug_build_guard { } using namespace debug_build_guard; }
|
||||
#endif
|
||||
#endif
|
||||
|
||||
#ifndef CV__DEBUG_NS_BEGIN
|
||||
#define CV__DEBUG_NS_BEGIN
|
||||
#define CV__DEBUG_NS_END
|
||||
#endif
|
||||
|
||||
|
||||
#ifdef __OPENCV_BUILD
|
||||
#include "cvconfig.h"
|
||||
#endif
|
||||
|
||||
#ifndef __CV_EXPAND
|
||||
#define __CV_EXPAND(x) x
|
||||
#endif
|
||||
|
||||
#ifndef __CV_CAT
|
||||
#define __CV_CAT__(x, y) x ## y
|
||||
#define __CV_CAT_(x, y) __CV_CAT__(x, y)
|
||||
#define __CV_CAT(x, y) __CV_CAT_(x, y)
|
||||
#endif
|
||||
|
||||
#define __CV_VA_NUM_ARGS_HELPER(_1, _2, _3, _4, _5, _6, _7, _8, _9, _10, N, ...) N
|
||||
#define __CV_VA_NUM_ARGS(...) __CV_VA_NUM_ARGS_HELPER(__VA_ARGS__, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1, 0)
|
||||
|
||||
#if defined __GNUC__
|
||||
#define CV_Func __func__
|
||||
#elif defined _MSC_VER
|
||||
#define CV_Func __FUNCTION__
|
||||
#else
|
||||
#define CV_Func ""
|
||||
#endif
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
//////////////// static assert /////////////////
|
||||
#define CVAUX_CONCAT_EXP(a, b) a##b
|
||||
#define CVAUX_CONCAT(a, b) CVAUX_CONCAT_EXP(a,b)
|
||||
|
||||
#if defined(__clang__)
|
||||
# ifndef __has_extension
|
||||
# define __has_extension __has_feature /* compatibility, for older versions of clang */
|
||||
# endif
|
||||
# if __has_extension(cxx_static_assert)
|
||||
# define CV_StaticAssert(condition, reason) static_assert((condition), reason " " #condition)
|
||||
# elif __has_extension(c_static_assert)
|
||||
# define CV_StaticAssert(condition, reason) _Static_assert((condition), reason " " #condition)
|
||||
# endif
|
||||
#elif defined(__GNUC__)
|
||||
# if (defined(__GXX_EXPERIMENTAL_CXX0X__) || __cplusplus >= 201103L)
|
||||
# define CV_StaticAssert(condition, reason) static_assert((condition), reason " " #condition)
|
||||
# endif
|
||||
#elif defined(_MSC_VER)
|
||||
# if _MSC_VER >= 1600 /* MSVC 10 */
|
||||
# define CV_StaticAssert(condition, reason) static_assert((condition), reason " " #condition)
|
||||
# endif
|
||||
#endif
|
||||
#ifndef CV_StaticAssert
|
||||
# if !defined(__clang__) && defined(__GNUC__) && (__GNUC__*100 + __GNUC_MINOR__ > 302)
|
||||
# define CV_StaticAssert(condition, reason) ({ extern int __attribute__((error("CV_StaticAssert: " reason " " #condition))) CV_StaticAssert(); ((condition) ? 0 : CV_StaticAssert()); })
|
||||
# else
|
||||
template <bool x> struct CV_StaticAssert_failed;
|
||||
template <> struct CV_StaticAssert_failed<true> { enum { val = 1 }; };
|
||||
template<int x> struct CV_StaticAssert_test {};
|
||||
# define CV_StaticAssert(condition, reason)\
|
||||
typedef cv::CV_StaticAssert_test< sizeof(cv::CV_StaticAssert_failed< static_cast<bool>(condition) >) > CVAUX_CONCAT(CV_StaticAssert_failed_at_, __LINE__)
|
||||
# endif
|
||||
#endif
|
||||
|
||||
// Suppress warning "-Wdeprecated-declarations" / C4996
|
||||
#if defined(_MSC_VER)
|
||||
#define CV_DO_PRAGMA(x) __pragma(x)
|
||||
#elif defined(__GNUC__)
|
||||
#define CV_DO_PRAGMA(x) _Pragma (#x)
|
||||
#else
|
||||
#define CV_DO_PRAGMA(x)
|
||||
#endif
|
||||
|
||||
#ifdef _MSC_VER
|
||||
#define CV_SUPPRESS_DEPRECATED_START \
|
||||
CV_DO_PRAGMA(warning(push)) \
|
||||
CV_DO_PRAGMA(warning(disable: 4996))
|
||||
#define CV_SUPPRESS_DEPRECATED_END CV_DO_PRAGMA(warning(pop))
|
||||
#elif defined (__clang__) || ((__GNUC__) && (__GNUC__*100 + __GNUC_MINOR__ > 405))
|
||||
#define CV_SUPPRESS_DEPRECATED_START \
|
||||
CV_DO_PRAGMA(GCC diagnostic push) \
|
||||
CV_DO_PRAGMA(GCC diagnostic ignored "-Wdeprecated-declarations")
|
||||
#define CV_SUPPRESS_DEPRECATED_END CV_DO_PRAGMA(GCC diagnostic pop)
|
||||
#else
|
||||
#define CV_SUPPRESS_DEPRECATED_START
|
||||
#define CV_SUPPRESS_DEPRECATED_END
|
||||
#endif
|
||||
|
||||
#define CV_UNUSED(name) (void)name
|
||||
|
||||
#if defined __GNUC__ && !defined __EXCEPTIONS
|
||||
#define CV_TRY
|
||||
#define CV_CATCH(A, B) for (A B; false; )
|
||||
#define CV_CATCH_ALL if (false)
|
||||
#define CV_THROW(A) abort()
|
||||
#define CV_RETHROW() abort()
|
||||
#else
|
||||
#define CV_TRY try
|
||||
#define CV_CATCH(A, B) catch(const A & B)
|
||||
#define CV_CATCH_ALL catch(...)
|
||||
#define CV_THROW(A) throw A
|
||||
#define CV_RETHROW() throw
|
||||
#endif
|
||||
|
||||
//! @endcond
|
||||
|
||||
// undef problematic defines sometimes defined by system headers (windows.h in particular)
|
||||
#undef small
|
||||
#undef min
|
||||
#undef max
|
||||
#undef abs
|
||||
#undef Complex
|
||||
|
||||
#include <limits.h>
|
||||
#include "opencv2/core/hal/interface.h"
|
||||
|
||||
#if defined __ICL
|
||||
# define CV_ICC __ICL
|
||||
#elif defined __ICC
|
||||
# define CV_ICC __ICC
|
||||
#elif defined __ECL
|
||||
# define CV_ICC __ECL
|
||||
#elif defined __ECC
|
||||
# define CV_ICC __ECC
|
||||
#elif defined __INTEL_COMPILER
|
||||
# define CV_ICC __INTEL_COMPILER
|
||||
#endif
|
||||
|
||||
#ifndef CV_INLINE
|
||||
# if defined __cplusplus
|
||||
# define CV_INLINE static inline
|
||||
# elif defined _MSC_VER
|
||||
# define CV_INLINE __inline
|
||||
# else
|
||||
# define CV_INLINE static
|
||||
# endif
|
||||
#endif
|
||||
|
||||
#if defined CV_DISABLE_OPTIMIZATION || (defined CV_ICC && !defined CV_ENABLE_UNROLLED)
|
||||
# define CV_ENABLE_UNROLLED 0
|
||||
#else
|
||||
# define CV_ENABLE_UNROLLED 1
|
||||
#endif
|
||||
|
||||
#ifdef __GNUC__
|
||||
# define CV_DECL_ALIGNED(x) __attribute__ ((aligned (x)))
|
||||
#elif defined _MSC_VER
|
||||
# define CV_DECL_ALIGNED(x) __declspec(align(x))
|
||||
#else
|
||||
# define CV_DECL_ALIGNED(x)
|
||||
#endif
|
||||
|
||||
/* CPU features and intrinsics support */
|
||||
#define CV_CPU_NONE 0
|
||||
#define CV_CPU_MMX 1
|
||||
#define CV_CPU_SSE 2
|
||||
#define CV_CPU_SSE2 3
|
||||
#define CV_CPU_SSE3 4
|
||||
#define CV_CPU_SSSE3 5
|
||||
#define CV_CPU_SSE4_1 6
|
||||
#define CV_CPU_SSE4_2 7
|
||||
#define CV_CPU_POPCNT 8
|
||||
#define CV_CPU_FP16 9
|
||||
#define CV_CPU_AVX 10
|
||||
#define CV_CPU_AVX2 11
|
||||
#define CV_CPU_FMA3 12
|
||||
|
||||
#define CV_CPU_AVX_512F 13
|
||||
#define CV_CPU_AVX_512BW 14
|
||||
#define CV_CPU_AVX_512CD 15
|
||||
#define CV_CPU_AVX_512DQ 16
|
||||
#define CV_CPU_AVX_512ER 17
|
||||
#define CV_CPU_AVX_512IFMA512 18 // deprecated
|
||||
#define CV_CPU_AVX_512IFMA 18
|
||||
#define CV_CPU_AVX_512PF 19
|
||||
#define CV_CPU_AVX_512VBMI 20
|
||||
#define CV_CPU_AVX_512VL 21
|
||||
|
||||
#define CV_CPU_NEON 100
|
||||
|
||||
#define CV_CPU_VSX 200
|
||||
#define CV_CPU_VSX3 201
|
||||
|
||||
// CPU features groups
|
||||
#define CV_CPU_AVX512_SKX 256
|
||||
|
||||
// when adding to this list remember to update the following enum
|
||||
#define CV_HARDWARE_MAX_FEATURE 512
|
||||
|
||||
/** @brief Available CPU features.
|
||||
*/
|
||||
enum CpuFeatures {
|
||||
CPU_MMX = 1,
|
||||
CPU_SSE = 2,
|
||||
CPU_SSE2 = 3,
|
||||
CPU_SSE3 = 4,
|
||||
CPU_SSSE3 = 5,
|
||||
CPU_SSE4_1 = 6,
|
||||
CPU_SSE4_2 = 7,
|
||||
CPU_POPCNT = 8,
|
||||
CPU_FP16 = 9,
|
||||
CPU_AVX = 10,
|
||||
CPU_AVX2 = 11,
|
||||
CPU_FMA3 = 12,
|
||||
|
||||
CPU_AVX_512F = 13,
|
||||
CPU_AVX_512BW = 14,
|
||||
CPU_AVX_512CD = 15,
|
||||
CPU_AVX_512DQ = 16,
|
||||
CPU_AVX_512ER = 17,
|
||||
CPU_AVX_512IFMA512 = 18, // deprecated
|
||||
CPU_AVX_512IFMA = 18,
|
||||
CPU_AVX_512PF = 19,
|
||||
CPU_AVX_512VBMI = 20,
|
||||
CPU_AVX_512VL = 21,
|
||||
|
||||
CPU_NEON = 100,
|
||||
|
||||
CPU_VSX = 200,
|
||||
CPU_VSX3 = 201,
|
||||
|
||||
CPU_AVX512_SKX = 256, //!< Skylake-X with AVX-512F/CD/BW/DQ/VL
|
||||
|
||||
CPU_MAX_FEATURE = 512 // see CV_HARDWARE_MAX_FEATURE
|
||||
};
|
||||
|
||||
|
||||
#include "cv_cpu_dispatch.h"
|
||||
|
||||
|
||||
/* fundamental constants */
|
||||
#define CV_PI 3.1415926535897932384626433832795
|
||||
#define CV_2PI 6.283185307179586476925286766559
|
||||
#define CV_LOG2 0.69314718055994530941723212145818
|
||||
|
||||
#if defined __ARM_FP16_FORMAT_IEEE \
|
||||
&& !defined __CUDACC__
|
||||
# define CV_FP16_TYPE 1
|
||||
#else
|
||||
# define CV_FP16_TYPE 0
|
||||
#endif
|
||||
|
||||
typedef union Cv16suf
|
||||
{
|
||||
short i;
|
||||
ushort u;
|
||||
#if CV_FP16_TYPE
|
||||
__fp16 h;
|
||||
#endif
|
||||
}
|
||||
Cv16suf;
|
||||
|
||||
typedef union Cv32suf
|
||||
{
|
||||
int i;
|
||||
unsigned u;
|
||||
float f;
|
||||
}
|
||||
Cv32suf;
|
||||
|
||||
typedef union Cv64suf
|
||||
{
|
||||
int64 i;
|
||||
uint64 u;
|
||||
double f;
|
||||
}
|
||||
Cv64suf;
|
||||
|
||||
#define OPENCV_ABI_COMPATIBILITY 300
|
||||
|
||||
#ifdef __OPENCV_BUILD
|
||||
# define DISABLE_OPENCV_24_COMPATIBILITY
|
||||
# define OPENCV_DISABLE_DEPRECATED_COMPATIBILITY
|
||||
#endif
|
||||
|
||||
#ifdef CVAPI_EXPORTS
|
||||
# if (defined _WIN32 || defined WINCE || defined __CYGWIN__)
|
||||
# define CV_EXPORTS __declspec(dllexport)
|
||||
# elif defined __GNUC__ && __GNUC__ >= 4
|
||||
# define CV_EXPORTS __attribute__ ((visibility ("default")))
|
||||
# endif
|
||||
#endif
|
||||
|
||||
#ifndef CV_EXPORTS
|
||||
# define CV_EXPORTS
|
||||
#endif
|
||||
|
||||
#ifdef _MSC_VER
|
||||
# define CV_EXPORTS_TEMPLATE
|
||||
#else
|
||||
# define CV_EXPORTS_TEMPLATE CV_EXPORTS
|
||||
#endif
|
||||
|
||||
#ifndef CV_DEPRECATED
|
||||
# if defined(__GNUC__)
|
||||
# define CV_DEPRECATED __attribute__ ((deprecated))
|
||||
# elif defined(_MSC_VER)
|
||||
# define CV_DEPRECATED __declspec(deprecated)
|
||||
# else
|
||||
# define CV_DEPRECATED
|
||||
# endif
|
||||
#endif
|
||||
|
||||
#ifndef CV_DEPRECATED_EXTERNAL
|
||||
# if defined(__OPENCV_BUILD)
|
||||
# define CV_DEPRECATED_EXTERNAL /* nothing */
|
||||
# else
|
||||
# define CV_DEPRECATED_EXTERNAL CV_DEPRECATED
|
||||
# endif
|
||||
#endif
|
||||
|
||||
|
||||
#ifndef CV_EXTERN_C
|
||||
# ifdef __cplusplus
|
||||
# define CV_EXTERN_C extern "C"
|
||||
# else
|
||||
# define CV_EXTERN_C
|
||||
# endif
|
||||
#endif
|
||||
|
||||
/* special informative macros for wrapper generators */
|
||||
#define CV_EXPORTS_W CV_EXPORTS
|
||||
#define CV_EXPORTS_W_SIMPLE CV_EXPORTS
|
||||
#define CV_EXPORTS_AS(synonym) CV_EXPORTS
|
||||
#define CV_EXPORTS_W_MAP CV_EXPORTS
|
||||
#define CV_IN_OUT
|
||||
#define CV_OUT
|
||||
#define CV_PROP
|
||||
#define CV_PROP_RW
|
||||
#define CV_WRAP
|
||||
#define CV_WRAP_AS(synonym)
|
||||
|
||||
/****************************************************************************************\
|
||||
* Matrix type (Mat) *
|
||||
\****************************************************************************************/
|
||||
|
||||
#define CV_MAT_CN_MASK ((CV_CN_MAX - 1) << CV_CN_SHIFT)
|
||||
#define CV_MAT_CN(flags) ((((flags) & CV_MAT_CN_MASK) >> CV_CN_SHIFT) + 1)
|
||||
#define CV_MAT_TYPE_MASK (CV_DEPTH_MAX*CV_CN_MAX - 1)
|
||||
#define CV_MAT_TYPE(flags) ((flags) & CV_MAT_TYPE_MASK)
|
||||
#define CV_MAT_CONT_FLAG_SHIFT 14
|
||||
#define CV_MAT_CONT_FLAG (1 << CV_MAT_CONT_FLAG_SHIFT)
|
||||
#define CV_IS_MAT_CONT(flags) ((flags) & CV_MAT_CONT_FLAG)
|
||||
#define CV_IS_CONT_MAT CV_IS_MAT_CONT
|
||||
#define CV_SUBMAT_FLAG_SHIFT 15
|
||||
#define CV_SUBMAT_FLAG (1 << CV_SUBMAT_FLAG_SHIFT)
|
||||
#define CV_IS_SUBMAT(flags) ((flags) & CV_MAT_SUBMAT_FLAG)
|
||||
|
||||
/** Size of each channel item,
|
||||
0x8442211 = 1000 0100 0100 0010 0010 0001 0001 ~ array of sizeof(arr_type_elem) */
|
||||
#define CV_ELEM_SIZE1(type) \
|
||||
((((sizeof(size_t)<<28)|0x8442211) >> CV_MAT_DEPTH(type)*4) & 15)
|
||||
|
||||
/** 0x3a50 = 11 10 10 01 01 00 00 ~ array of log2(sizeof(arr_type_elem)) */
|
||||
#define CV_ELEM_SIZE(type) \
|
||||
(CV_MAT_CN(type) << ((((sizeof(size_t)/4+1)*16384|0x3a50) >> CV_MAT_DEPTH(type)*2) & 3))
|
||||
|
||||
#ifndef MIN
|
||||
# define MIN(a,b) ((a) > (b) ? (b) : (a))
|
||||
#endif
|
||||
|
||||
#ifndef MAX
|
||||
# define MAX(a,b) ((a) < (b) ? (b) : (a))
|
||||
#endif
|
||||
|
||||
/****************************************************************************************\
|
||||
* static analysys *
|
||||
\****************************************************************************************/
|
||||
|
||||
// In practice, some macro are not processed correctly (noreturn is not detected).
|
||||
// We need to use simplified definition for them.
|
||||
#ifndef CV_STATIC_ANALYSIS
|
||||
# if defined(__KLOCWORK__) || defined(__clang_analyzer__) || defined(__COVERITY__)
|
||||
# define CV_STATIC_ANALYSIS 1
|
||||
# endif
|
||||
#else
|
||||
# if defined(CV_STATIC_ANALYSIS) && !(__CV_CAT(1, CV_STATIC_ANALYSIS) == 1) // defined and not empty
|
||||
# if 0 == CV_STATIC_ANALYSIS
|
||||
# undef CV_STATIC_ANALYSIS
|
||||
# endif
|
||||
# endif
|
||||
#endif
|
||||
|
||||
/****************************************************************************************\
|
||||
* Thread sanitizer *
|
||||
\****************************************************************************************/
|
||||
#ifndef CV_THREAD_SANITIZER
|
||||
# if defined(__has_feature)
|
||||
# if __has_feature(thread_sanitizer)
|
||||
# define CV_THREAD_SANITIZER
|
||||
# endif
|
||||
# endif
|
||||
#endif
|
||||
|
||||
/****************************************************************************************\
|
||||
* exchange-add operation for atomic operations on reference counters *
|
||||
\****************************************************************************************/
|
||||
|
||||
#ifdef CV_XADD
|
||||
// allow to use user-defined macro
|
||||
#elif defined __GNUC__ || defined __clang__
|
||||
# if defined __clang__ && __clang_major__ >= 3 && !defined __ANDROID__ && !defined __EMSCRIPTEN__ && !defined(__CUDACC__)
|
||||
# ifdef __ATOMIC_ACQ_REL
|
||||
# define CV_XADD(addr, delta) __c11_atomic_fetch_add((_Atomic(int)*)(addr), delta, __ATOMIC_ACQ_REL)
|
||||
# else
|
||||
# define CV_XADD(addr, delta) __atomic_fetch_add((_Atomic(int)*)(addr), delta, 4)
|
||||
# endif
|
||||
# else
|
||||
# if defined __ATOMIC_ACQ_REL && !defined __clang__
|
||||
// version for gcc >= 4.7
|
||||
# define CV_XADD(addr, delta) (int)__atomic_fetch_add((unsigned*)(addr), (unsigned)(delta), __ATOMIC_ACQ_REL)
|
||||
# else
|
||||
# define CV_XADD(addr, delta) (int)__sync_fetch_and_add((unsigned*)(addr), (unsigned)(delta))
|
||||
# endif
|
||||
# endif
|
||||
#elif defined _MSC_VER && !defined RC_INVOKED
|
||||
# include <intrin.h>
|
||||
# define CV_XADD(addr, delta) (int)_InterlockedExchangeAdd((long volatile*)addr, delta)
|
||||
#else
|
||||
CV_INLINE CV_XADD(int* addr, int delta) { int tmp = *addr; *addr += delta; return tmp; }
|
||||
#endif
|
||||
|
||||
|
||||
/****************************************************************************************\
|
||||
* CV_NORETURN attribute *
|
||||
\****************************************************************************************/
|
||||
|
||||
#ifndef CV_NORETURN
|
||||
# if defined(__GNUC__)
|
||||
# define CV_NORETURN __attribute__((__noreturn__))
|
||||
# elif defined(_MSC_VER) && (_MSC_VER >= 1300)
|
||||
# define CV_NORETURN __declspec(noreturn)
|
||||
# else
|
||||
# define CV_NORETURN /* nothing by default */
|
||||
# endif
|
||||
#endif
|
||||
|
||||
|
||||
/****************************************************************************************\
|
||||
* CV_NODISCARD attribute *
|
||||
* encourages the compiler to issue a warning if the return value is discarded (C++17) *
|
||||
\****************************************************************************************/
|
||||
#ifndef CV_NODISCARD
|
||||
# if defined(__GNUC__)
|
||||
# define CV_NODISCARD __attribute__((__warn_unused_result__)) // at least available with GCC 3.4
|
||||
# elif defined(__clang__) && defined(__has_attribute)
|
||||
# if __has_attribute(__warn_unused_result__)
|
||||
# define CV_NODISCARD __attribute__((__warn_unused_result__))
|
||||
# endif
|
||||
# endif
|
||||
#endif
|
||||
#ifndef CV_NODISCARD
|
||||
# define CV_NODISCARD /* nothing by default */
|
||||
#endif
|
||||
|
||||
|
||||
/****************************************************************************************\
|
||||
* C++ 11 *
|
||||
\****************************************************************************************/
|
||||
#ifndef CV_CXX11
|
||||
# if __cplusplus >= 201103L || (defined(_MSC_VER) && _MSC_VER >= 1800)
|
||||
# define CV_CXX11 1
|
||||
# endif
|
||||
#else
|
||||
# if CV_CXX11 == 0
|
||||
# undef CV_CXX11
|
||||
# endif
|
||||
#endif
|
||||
|
||||
|
||||
/****************************************************************************************\
|
||||
* C++ Move semantics *
|
||||
\****************************************************************************************/
|
||||
|
||||
#ifndef CV_CXX_MOVE_SEMANTICS
|
||||
# if __cplusplus >= 201103L || defined(__GXX_EXPERIMENTAL_CXX0X__) || (defined(_MSC_VER) && _MSC_VER >= 1600)
|
||||
# define CV_CXX_MOVE_SEMANTICS 1
|
||||
# elif defined(__clang)
|
||||
# if __has_feature(cxx_rvalue_references)
|
||||
# define CV_CXX_MOVE_SEMANTICS 1
|
||||
# endif
|
||||
# endif
|
||||
#else
|
||||
# if CV_CXX_MOVE_SEMANTICS == 0
|
||||
# undef CV_CXX_MOVE_SEMANTICS
|
||||
# endif
|
||||
#endif
|
||||
|
||||
/****************************************************************************************\
|
||||
* C++11 std::array *
|
||||
\****************************************************************************************/
|
||||
|
||||
#ifndef CV_CXX_STD_ARRAY
|
||||
# if __cplusplus >= 201103L || (defined(_MSC_VER) && _MSC_VER >= 1900/*MSVS 2015*/)
|
||||
# define CV_CXX_STD_ARRAY 1
|
||||
# include <array>
|
||||
# endif
|
||||
#else
|
||||
# if CV_CXX_STD_ARRAY == 0
|
||||
# undef CV_CXX_STD_ARRAY
|
||||
# endif
|
||||
#endif
|
||||
|
||||
|
||||
/****************************************************************************************\
|
||||
* C++11 override / final *
|
||||
\****************************************************************************************/
|
||||
|
||||
#ifndef CV_OVERRIDE
|
||||
# ifdef CV_CXX11
|
||||
# define CV_OVERRIDE override
|
||||
# endif
|
||||
#endif
|
||||
#ifndef CV_OVERRIDE
|
||||
# define CV_OVERRIDE
|
||||
#endif
|
||||
|
||||
#ifndef CV_FINAL
|
||||
# ifdef CV_CXX11
|
||||
# define CV_FINAL final
|
||||
# endif
|
||||
#endif
|
||||
#ifndef CV_FINAL
|
||||
# define CV_FINAL
|
||||
#endif
|
||||
|
||||
|
||||
|
||||
// Integer types portatibility
|
||||
#ifdef OPENCV_STDINT_HEADER
|
||||
#include OPENCV_STDINT_HEADER
|
||||
#elif defined(__cplusplus)
|
||||
#if defined(_MSC_VER) && _MSC_VER < 1600 /* MSVS 2010 */
|
||||
namespace cv {
|
||||
typedef signed char int8_t;
|
||||
typedef unsigned char uint8_t;
|
||||
typedef signed short int16_t;
|
||||
typedef unsigned short uint16_t;
|
||||
typedef signed int int32_t;
|
||||
typedef unsigned int uint32_t;
|
||||
typedef signed __int64 int64_t;
|
||||
typedef unsigned __int64 uint64_t;
|
||||
}
|
||||
#elif defined(_MSC_VER) || __cplusplus >= 201103L
|
||||
#include <cstdint>
|
||||
namespace cv {
|
||||
using std::int8_t;
|
||||
using std::uint8_t;
|
||||
using std::int16_t;
|
||||
using std::uint16_t;
|
||||
using std::int32_t;
|
||||
using std::uint32_t;
|
||||
using std::int64_t;
|
||||
using std::uint64_t;
|
||||
}
|
||||
#else
|
||||
#include <stdint.h>
|
||||
namespace cv {
|
||||
typedef ::int8_t int8_t;
|
||||
typedef ::uint8_t uint8_t;
|
||||
typedef ::int16_t int16_t;
|
||||
typedef ::uint16_t uint16_t;
|
||||
typedef ::int32_t int32_t;
|
||||
typedef ::uint32_t uint32_t;
|
||||
typedef ::int64_t int64_t;
|
||||
typedef ::uint64_t uint64_t;
|
||||
}
|
||||
#endif
|
||||
#else // pure C
|
||||
#include <stdint.h>
|
||||
#endif
|
||||
|
||||
#ifdef __cplusplus
|
||||
namespace cv
|
||||
{
|
||||
|
||||
class float16_t
|
||||
{
|
||||
public:
|
||||
#if CV_FP16_TYPE
|
||||
|
||||
float16_t() {}
|
||||
explicit float16_t(float x) { h = (__fp16)x; }
|
||||
operator float() const { return (float)h; }
|
||||
static float16_t fromBits(ushort w)
|
||||
{
|
||||
Cv16suf u;
|
||||
u.u = w;
|
||||
float16_t result;
|
||||
result.h = u.h;
|
||||
return result;
|
||||
}
|
||||
static float16_t zero()
|
||||
{
|
||||
float16_t result;
|
||||
result.h = (__fp16)0;
|
||||
return result;
|
||||
}
|
||||
ushort bits() const
|
||||
{
|
||||
Cv16suf u;
|
||||
u.h = h;
|
||||
return u.u;
|
||||
}
|
||||
protected:
|
||||
__fp16 h;
|
||||
|
||||
#else
|
||||
float16_t() {}
|
||||
explicit float16_t(float x)
|
||||
{
|
||||
#if CV_AVX2
|
||||
__m128 v = _mm_load_ss(&x);
|
||||
w = (ushort)_mm_cvtsi128_si32(_mm_cvtps_ph(v, 0));
|
||||
#else
|
||||
Cv32suf in;
|
||||
in.f = x;
|
||||
unsigned sign = in.u & 0x80000000;
|
||||
in.u ^= sign;
|
||||
|
||||
if( in.u >= 0x47800000 )
|
||||
w = (ushort)(in.u > 0x7f800000 ? 0x7e00 : 0x7c00);
|
||||
else
|
||||
{
|
||||
if (in.u < 0x38800000)
|
||||
{
|
||||
in.f += 0.5f;
|
||||
w = (ushort)(in.u - 0x3f000000);
|
||||
}
|
||||
else
|
||||
{
|
||||
unsigned t = in.u + 0xc8000fff;
|
||||
w = (ushort)((t + ((in.u >> 13) & 1)) >> 13);
|
||||
}
|
||||
}
|
||||
|
||||
w = (ushort)(w | (sign >> 16));
|
||||
#endif
|
||||
}
|
||||
|
||||
operator float() const
|
||||
{
|
||||
#if CV_AVX2
|
||||
float f;
|
||||
_mm_store_ss(&f, _mm_cvtph_ps(_mm_cvtsi32_si128(w)));
|
||||
return f;
|
||||
#else
|
||||
Cv32suf out;
|
||||
|
||||
unsigned t = ((w & 0x7fff) << 13) + 0x38000000;
|
||||
unsigned sign = (w & 0x8000) << 16;
|
||||
unsigned e = w & 0x7c00;
|
||||
|
||||
out.u = t + (1 << 23);
|
||||
out.u = (e >= 0x7c00 ? t + 0x38000000 :
|
||||
e == 0 ? (out.f -= 6.103515625e-05f, out.u) : t) | sign;
|
||||
return out.f;
|
||||
#endif
|
||||
}
|
||||
|
||||
static float16_t fromBits(ushort b)
|
||||
{
|
||||
float16_t result;
|
||||
result.w = b;
|
||||
return result;
|
||||
}
|
||||
static float16_t zero()
|
||||
{
|
||||
float16_t result;
|
||||
result.w = (ushort)0;
|
||||
return result;
|
||||
}
|
||||
ushort bits() const { return w; }
|
||||
protected:
|
||||
ushort w;
|
||||
|
||||
#endif
|
||||
};
|
||||
|
||||
}
|
||||
#endif
|
||||
|
||||
//! @}
|
||||
|
||||
#ifndef __cplusplus
|
||||
#include "opencv2/core/fast_math.hpp" // define cvRound(double)
|
||||
#endif
|
||||
|
||||
#endif // OPENCV_CORE_CVDEF_H
|
||||
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,285 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CORE_CVSTDINL_HPP
|
||||
#define OPENCV_CORE_CVSTDINL_HPP
|
||||
|
||||
#include <complex>
|
||||
#include <ostream>
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
#ifdef _MSC_VER
|
||||
#pragma warning( push )
|
||||
#pragma warning( disable: 4127 )
|
||||
#endif
|
||||
|
||||
namespace cv
|
||||
{
|
||||
|
||||
template<typename _Tp> class DataType< std::complex<_Tp> >
|
||||
{
|
||||
public:
|
||||
typedef std::complex<_Tp> value_type;
|
||||
typedef value_type work_type;
|
||||
typedef _Tp channel_type;
|
||||
|
||||
enum { generic_type = 0,
|
||||
depth = DataType<channel_type>::depth,
|
||||
channels = 2,
|
||||
fmt = DataType<channel_type>::fmt + ((channels - 1) << 8),
|
||||
type = CV_MAKETYPE(depth, channels) };
|
||||
|
||||
typedef Vec<channel_type, channels> vec_type;
|
||||
};
|
||||
|
||||
inline
|
||||
String::String(const std::string& str)
|
||||
: cstr_(0), len_(0)
|
||||
{
|
||||
size_t len = str.size();
|
||||
if (len) memcpy(allocate(len), str.c_str(), len);
|
||||
}
|
||||
|
||||
inline
|
||||
String::String(const std::string& str, size_t pos, size_t len)
|
||||
: cstr_(0), len_(0)
|
||||
{
|
||||
size_t strlen = str.size();
|
||||
pos = min(pos, strlen);
|
||||
len = min(strlen - pos, len);
|
||||
if (!len) return;
|
||||
memcpy(allocate(len), str.c_str() + pos, len);
|
||||
}
|
||||
|
||||
inline
|
||||
String& String::operator = (const std::string& str)
|
||||
{
|
||||
deallocate();
|
||||
size_t len = str.size();
|
||||
if (len) memcpy(allocate(len), str.c_str(), len);
|
||||
return *this;
|
||||
}
|
||||
|
||||
inline
|
||||
String& String::operator += (const std::string& str)
|
||||
{
|
||||
*this = *this + str;
|
||||
return *this;
|
||||
}
|
||||
|
||||
inline
|
||||
String::operator std::string() const
|
||||
{
|
||||
return std::string(cstr_, len_);
|
||||
}
|
||||
|
||||
inline
|
||||
String operator + (const String& lhs, const std::string& rhs)
|
||||
{
|
||||
String s;
|
||||
size_t rhslen = rhs.size();
|
||||
s.allocate(lhs.len_ + rhslen);
|
||||
if (lhs.len_) memcpy(s.cstr_, lhs.cstr_, lhs.len_);
|
||||
if (rhslen) memcpy(s.cstr_ + lhs.len_, rhs.c_str(), rhslen);
|
||||
return s;
|
||||
}
|
||||
|
||||
inline
|
||||
String operator + (const std::string& lhs, const String& rhs)
|
||||
{
|
||||
String s;
|
||||
size_t lhslen = lhs.size();
|
||||
s.allocate(lhslen + rhs.len_);
|
||||
if (lhslen) memcpy(s.cstr_, lhs.c_str(), lhslen);
|
||||
if (rhs.len_) memcpy(s.cstr_ + lhslen, rhs.cstr_, rhs.len_);
|
||||
return s;
|
||||
}
|
||||
|
||||
inline
|
||||
FileNode::operator std::string() const
|
||||
{
|
||||
String value;
|
||||
read(*this, value, value);
|
||||
return value;
|
||||
}
|
||||
|
||||
template<> inline
|
||||
void operator >> (const FileNode& n, std::string& value)
|
||||
{
|
||||
read(n, value, std::string());
|
||||
}
|
||||
|
||||
template<> inline
|
||||
FileStorage& operator << (FileStorage& fs, const std::string& value)
|
||||
{
|
||||
return fs << cv::String(value);
|
||||
}
|
||||
|
||||
static inline
|
||||
std::ostream& operator << (std::ostream& os, const String& str)
|
||||
{
|
||||
return os << str.c_str();
|
||||
}
|
||||
|
||||
static inline
|
||||
std::ostream& operator << (std::ostream& out, Ptr<Formatted> fmtd)
|
||||
{
|
||||
fmtd->reset();
|
||||
for(const char* str = fmtd->next(); str; str = fmtd->next())
|
||||
out << str;
|
||||
return out;
|
||||
}
|
||||
|
||||
static inline
|
||||
std::ostream& operator << (std::ostream& out, const Mat& mtx)
|
||||
{
|
||||
return out << Formatter::get()->format(mtx);
|
||||
}
|
||||
|
||||
static inline
|
||||
std::ostream& operator << (std::ostream& out, const UMat& m)
|
||||
{
|
||||
return out << m.getMat(ACCESS_READ);
|
||||
}
|
||||
|
||||
template<typename _Tp> static inline
|
||||
std::ostream& operator << (std::ostream& out, const Complex<_Tp>& c)
|
||||
{
|
||||
return out << "(" << c.re << "," << c.im << ")";
|
||||
}
|
||||
|
||||
template<typename _Tp> static inline
|
||||
std::ostream& operator << (std::ostream& out, const std::vector<Point_<_Tp> >& vec)
|
||||
{
|
||||
return out << Formatter::get()->format(Mat(vec));
|
||||
}
|
||||
|
||||
|
||||
template<typename _Tp> static inline
|
||||
std::ostream& operator << (std::ostream& out, const std::vector<Point3_<_Tp> >& vec)
|
||||
{
|
||||
return out << Formatter::get()->format(Mat(vec));
|
||||
}
|
||||
|
||||
|
||||
template<typename _Tp, int m, int n> static inline
|
||||
std::ostream& operator << (std::ostream& out, const Matx<_Tp, m, n>& matx)
|
||||
{
|
||||
return out << Formatter::get()->format(Mat(matx));
|
||||
}
|
||||
|
||||
template<typename _Tp> static inline
|
||||
std::ostream& operator << (std::ostream& out, const Point_<_Tp>& p)
|
||||
{
|
||||
out << "[" << p.x << ", " << p.y << "]";
|
||||
return out;
|
||||
}
|
||||
|
||||
template<typename _Tp> static inline
|
||||
std::ostream& operator << (std::ostream& out, const Point3_<_Tp>& p)
|
||||
{
|
||||
out << "[" << p.x << ", " << p.y << ", " << p.z << "]";
|
||||
return out;
|
||||
}
|
||||
|
||||
template<typename _Tp, int n> static inline
|
||||
std::ostream& operator << (std::ostream& out, const Vec<_Tp, n>& vec)
|
||||
{
|
||||
out << "[";
|
||||
if (cv::traits::Depth<_Tp>::value <= CV_32S)
|
||||
{
|
||||
for (int i = 0; i < n - 1; ++i) {
|
||||
out << (int)vec[i] << ", ";
|
||||
}
|
||||
out << (int)vec[n-1] << "]";
|
||||
}
|
||||
else
|
||||
{
|
||||
for (int i = 0; i < n - 1; ++i) {
|
||||
out << vec[i] << ", ";
|
||||
}
|
||||
out << vec[n-1] << "]";
|
||||
}
|
||||
|
||||
return out;
|
||||
}
|
||||
|
||||
template<typename _Tp> static inline
|
||||
std::ostream& operator << (std::ostream& out, const Size_<_Tp>& size)
|
||||
{
|
||||
return out << "[" << size.width << " x " << size.height << "]";
|
||||
}
|
||||
|
||||
template<typename _Tp> static inline
|
||||
std::ostream& operator << (std::ostream& out, const Rect_<_Tp>& rect)
|
||||
{
|
||||
return out << "[" << rect.width << " x " << rect.height << " from (" << rect.x << ", " << rect.y << ")]";
|
||||
}
|
||||
|
||||
static inline std::ostream& operator << (std::ostream& out, const MatSize& msize)
|
||||
{
|
||||
int i, dims = msize.dims();
|
||||
for( i = 0; i < dims; i++ )
|
||||
{
|
||||
out << msize[i];
|
||||
if( i < dims-1 )
|
||||
out << " x ";
|
||||
}
|
||||
return out;
|
||||
}
|
||||
|
||||
static inline std::ostream &operator<< (std::ostream &s, cv::Range &r)
|
||||
{
|
||||
return s << "[" << r.start << " : " << r.end << ")";
|
||||
}
|
||||
|
||||
} // cv
|
||||
|
||||
#ifdef _MSC_VER
|
||||
#pragma warning( pop )
|
||||
#endif
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif // OPENCV_CORE_CVSTDINL_HPP
|
||||
|
|
@ -0,0 +1,184 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2010-2013, Advanced Micro Devices, Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors as is and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the copyright holders or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CORE_DIRECTX_HPP
|
||||
#define OPENCV_CORE_DIRECTX_HPP
|
||||
|
||||
#include "mat.hpp"
|
||||
#include "ocl.hpp"
|
||||
|
||||
#if !defined(__d3d11_h__)
|
||||
struct ID3D11Device;
|
||||
struct ID3D11Texture2D;
|
||||
#endif
|
||||
|
||||
#if !defined(__d3d10_h__)
|
||||
struct ID3D10Device;
|
||||
struct ID3D10Texture2D;
|
||||
#endif
|
||||
|
||||
#if !defined(_D3D9_H_)
|
||||
struct IDirect3DDevice9;
|
||||
struct IDirect3DDevice9Ex;
|
||||
struct IDirect3DSurface9;
|
||||
#endif
|
||||
|
||||
|
||||
namespace cv { namespace directx {
|
||||
|
||||
namespace ocl {
|
||||
using namespace cv::ocl;
|
||||
|
||||
//! @addtogroup core_directx
|
||||
// This section describes OpenCL and DirectX interoperability.
|
||||
//
|
||||
// To enable DirectX support, configure OpenCV using CMake with WITH_DIRECTX=ON . Note, DirectX is
|
||||
// supported only on Windows.
|
||||
//
|
||||
// To use OpenCL functionality you should first initialize OpenCL context from DirectX resource.
|
||||
//
|
||||
//! @{
|
||||
|
||||
// TODO static functions in the Context class
|
||||
//! @brief Creates OpenCL context from D3D11 device
|
||||
//
|
||||
//! @param pD3D11Device - pointer to D3D11 device
|
||||
//! @return Returns reference to OpenCL Context
|
||||
CV_EXPORTS Context& initializeContextFromD3D11Device(ID3D11Device* pD3D11Device);
|
||||
|
||||
//! @brief Creates OpenCL context from D3D10 device
|
||||
//
|
||||
//! @param pD3D10Device - pointer to D3D10 device
|
||||
//! @return Returns reference to OpenCL Context
|
||||
CV_EXPORTS Context& initializeContextFromD3D10Device(ID3D10Device* pD3D10Device);
|
||||
|
||||
//! @brief Creates OpenCL context from Direct3DDevice9Ex device
|
||||
//
|
||||
//! @param pDirect3DDevice9Ex - pointer to Direct3DDevice9Ex device
|
||||
//! @return Returns reference to OpenCL Context
|
||||
CV_EXPORTS Context& initializeContextFromDirect3DDevice9Ex(IDirect3DDevice9Ex* pDirect3DDevice9Ex);
|
||||
|
||||
//! @brief Creates OpenCL context from Direct3DDevice9 device
|
||||
//
|
||||
//! @param pDirect3DDevice9 - pointer to Direct3Device9 device
|
||||
//! @return Returns reference to OpenCL Context
|
||||
CV_EXPORTS Context& initializeContextFromDirect3DDevice9(IDirect3DDevice9* pDirect3DDevice9);
|
||||
|
||||
//! @}
|
||||
|
||||
} // namespace cv::directx::ocl
|
||||
|
||||
//! @addtogroup core_directx
|
||||
//! @{
|
||||
|
||||
//! @brief Converts InputArray to ID3D11Texture2D. If destination texture format is DXGI_FORMAT_NV12 then
|
||||
//! input UMat expected to be in BGR format and data will be downsampled and color-converted to NV12.
|
||||
//
|
||||
//! @note Note: Destination texture must be allocated by application. Function does memory copy from src to
|
||||
//! pD3D11Texture2D
|
||||
//
|
||||
//! @param src - source InputArray
|
||||
//! @param pD3D11Texture2D - destination D3D11 texture
|
||||
CV_EXPORTS void convertToD3D11Texture2D(InputArray src, ID3D11Texture2D* pD3D11Texture2D);
|
||||
|
||||
//! @brief Converts ID3D11Texture2D to OutputArray. If input texture format is DXGI_FORMAT_NV12 then
|
||||
//! data will be upsampled and color-converted to BGR format.
|
||||
//
|
||||
//! @note Note: Destination matrix will be re-allocated if it has not enough memory to match texture size.
|
||||
//! function does memory copy from pD3D11Texture2D to dst
|
||||
//
|
||||
//! @param pD3D11Texture2D - source D3D11 texture
|
||||
//! @param dst - destination OutputArray
|
||||
CV_EXPORTS void convertFromD3D11Texture2D(ID3D11Texture2D* pD3D11Texture2D, OutputArray dst);
|
||||
|
||||
//! @brief Converts InputArray to ID3D10Texture2D
|
||||
//
|
||||
//! @note Note: function does memory copy from src to
|
||||
//! pD3D10Texture2D
|
||||
//
|
||||
//! @param src - source InputArray
|
||||
//! @param pD3D10Texture2D - destination D3D10 texture
|
||||
CV_EXPORTS void convertToD3D10Texture2D(InputArray src, ID3D10Texture2D* pD3D10Texture2D);
|
||||
|
||||
//! @brief Converts ID3D10Texture2D to OutputArray
|
||||
//
|
||||
//! @note Note: function does memory copy from pD3D10Texture2D
|
||||
//! to dst
|
||||
//
|
||||
//! @param pD3D10Texture2D - source D3D10 texture
|
||||
//! @param dst - destination OutputArray
|
||||
CV_EXPORTS void convertFromD3D10Texture2D(ID3D10Texture2D* pD3D10Texture2D, OutputArray dst);
|
||||
|
||||
//! @brief Converts InputArray to IDirect3DSurface9
|
||||
//
|
||||
//! @note Note: function does memory copy from src to
|
||||
//! pDirect3DSurface9
|
||||
//
|
||||
//! @param src - source InputArray
|
||||
//! @param pDirect3DSurface9 - destination D3D10 texture
|
||||
//! @param surfaceSharedHandle - shared handle
|
||||
CV_EXPORTS void convertToDirect3DSurface9(InputArray src, IDirect3DSurface9* pDirect3DSurface9, void* surfaceSharedHandle = NULL);
|
||||
|
||||
//! @brief Converts IDirect3DSurface9 to OutputArray
|
||||
//
|
||||
//! @note Note: function does memory copy from pDirect3DSurface9
|
||||
//! to dst
|
||||
//
|
||||
//! @param pDirect3DSurface9 - source D3D10 texture
|
||||
//! @param dst - destination OutputArray
|
||||
//! @param surfaceSharedHandle - shared handle
|
||||
CV_EXPORTS void convertFromDirect3DSurface9(IDirect3DSurface9* pDirect3DSurface9, OutputArray dst, void* surfaceSharedHandle = NULL);
|
||||
|
||||
//! @brief Get OpenCV type from DirectX type
|
||||
//! @param iDXGI_FORMAT - enum DXGI_FORMAT for D3D10/D3D11
|
||||
//! @return OpenCV type or -1 if there is no equivalent
|
||||
CV_EXPORTS int getTypeFromDXGI_FORMAT(const int iDXGI_FORMAT); // enum DXGI_FORMAT for D3D10/D3D11
|
||||
|
||||
//! @brief Get OpenCV type from DirectX type
|
||||
//! @param iD3DFORMAT - enum D3DTYPE for D3D9
|
||||
//! @return OpenCV type or -1 if there is no equivalent
|
||||
CV_EXPORTS int getTypeFromD3DFORMAT(const int iD3DFORMAT); // enum D3DTYPE for D3D9
|
||||
|
||||
//! @}
|
||||
|
||||
} } // namespace cv::directx
|
||||
|
||||
#endif // OPENCV_CORE_DIRECTX_HPP
|
||||
|
|
@ -0,0 +1,280 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
|
||||
#ifndef OPENCV_CORE_EIGEN_HPP
|
||||
#define OPENCV_CORE_EIGEN_HPP
|
||||
|
||||
#include "opencv2/core.hpp"
|
||||
|
||||
#if defined _MSC_VER && _MSC_VER >= 1200
|
||||
#pragma warning( disable: 4714 ) //__forceinline is not inlined
|
||||
#pragma warning( disable: 4127 ) //conditional expression is constant
|
||||
#pragma warning( disable: 4244 ) //conversion from '__int64' to 'int', possible loss of data
|
||||
#endif
|
||||
|
||||
namespace cv
|
||||
{
|
||||
|
||||
//! @addtogroup core_eigen
|
||||
//! @{
|
||||
|
||||
template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols> static inline
|
||||
void eigen2cv( const Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCols>& src, OutputArray dst )
|
||||
{
|
||||
if( !(src.Flags & Eigen::RowMajorBit) )
|
||||
{
|
||||
Mat _src(src.cols(), src.rows(), traits::Type<_Tp>::value,
|
||||
(void*)src.data(), src.outerStride()*sizeof(_Tp));
|
||||
transpose(_src, dst);
|
||||
}
|
||||
else
|
||||
{
|
||||
Mat _src(src.rows(), src.cols(), traits::Type<_Tp>::value,
|
||||
(void*)src.data(), src.outerStride()*sizeof(_Tp));
|
||||
_src.copyTo(dst);
|
||||
}
|
||||
}
|
||||
|
||||
// Matx case
|
||||
template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols> static inline
|
||||
void eigen2cv( const Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCols>& src,
|
||||
Matx<_Tp, _rows, _cols>& dst )
|
||||
{
|
||||
if( !(src.Flags & Eigen::RowMajorBit) )
|
||||
{
|
||||
dst = Matx<_Tp, _cols, _rows>(static_cast<const _Tp*>(src.data())).t();
|
||||
}
|
||||
else
|
||||
{
|
||||
dst = Matx<_Tp, _rows, _cols>(static_cast<const _Tp*>(src.data()));
|
||||
}
|
||||
}
|
||||
|
||||
template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols> static inline
|
||||
void cv2eigen( const Mat& src,
|
||||
Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCols>& dst )
|
||||
{
|
||||
CV_DbgAssert(src.rows == _rows && src.cols == _cols);
|
||||
if( !(dst.Flags & Eigen::RowMajorBit) )
|
||||
{
|
||||
const Mat _dst(src.cols, src.rows, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
if( src.type() == _dst.type() )
|
||||
transpose(src, _dst);
|
||||
else if( src.cols == src.rows )
|
||||
{
|
||||
src.convertTo(_dst, _dst.type());
|
||||
transpose(_dst, _dst);
|
||||
}
|
||||
else
|
||||
Mat(src.t()).convertTo(_dst, _dst.type());
|
||||
}
|
||||
else
|
||||
{
|
||||
const Mat _dst(src.rows, src.cols, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
src.convertTo(_dst, _dst.type());
|
||||
}
|
||||
}
|
||||
|
||||
// Matx case
|
||||
template<typename _Tp, int _rows, int _cols, int _options, int _maxRows, int _maxCols> static inline
|
||||
void cv2eigen( const Matx<_Tp, _rows, _cols>& src,
|
||||
Eigen::Matrix<_Tp, _rows, _cols, _options, _maxRows, _maxCols>& dst )
|
||||
{
|
||||
if( !(dst.Flags & Eigen::RowMajorBit) )
|
||||
{
|
||||
const Mat _dst(_cols, _rows, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
transpose(src, _dst);
|
||||
}
|
||||
else
|
||||
{
|
||||
const Mat _dst(_rows, _cols, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
Mat(src).copyTo(_dst);
|
||||
}
|
||||
}
|
||||
|
||||
template<typename _Tp> static inline
|
||||
void cv2eigen( const Mat& src,
|
||||
Eigen::Matrix<_Tp, Eigen::Dynamic, Eigen::Dynamic>& dst )
|
||||
{
|
||||
dst.resize(src.rows, src.cols);
|
||||
if( !(dst.Flags & Eigen::RowMajorBit) )
|
||||
{
|
||||
const Mat _dst(src.cols, src.rows, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
if( src.type() == _dst.type() )
|
||||
transpose(src, _dst);
|
||||
else if( src.cols == src.rows )
|
||||
{
|
||||
src.convertTo(_dst, _dst.type());
|
||||
transpose(_dst, _dst);
|
||||
}
|
||||
else
|
||||
Mat(src.t()).convertTo(_dst, _dst.type());
|
||||
}
|
||||
else
|
||||
{
|
||||
const Mat _dst(src.rows, src.cols, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
src.convertTo(_dst, _dst.type());
|
||||
}
|
||||
}
|
||||
|
||||
// Matx case
|
||||
template<typename _Tp, int _rows, int _cols> static inline
|
||||
void cv2eigen( const Matx<_Tp, _rows, _cols>& src,
|
||||
Eigen::Matrix<_Tp, Eigen::Dynamic, Eigen::Dynamic>& dst )
|
||||
{
|
||||
dst.resize(_rows, _cols);
|
||||
if( !(dst.Flags & Eigen::RowMajorBit) )
|
||||
{
|
||||
const Mat _dst(_cols, _rows, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
transpose(src, _dst);
|
||||
}
|
||||
else
|
||||
{
|
||||
const Mat _dst(_rows, _cols, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
Mat(src).copyTo(_dst);
|
||||
}
|
||||
}
|
||||
|
||||
template<typename _Tp> static inline
|
||||
void cv2eigen( const Mat& src,
|
||||
Eigen::Matrix<_Tp, Eigen::Dynamic, 1>& dst )
|
||||
{
|
||||
CV_Assert(src.cols == 1);
|
||||
dst.resize(src.rows);
|
||||
|
||||
if( !(dst.Flags & Eigen::RowMajorBit) )
|
||||
{
|
||||
const Mat _dst(src.cols, src.rows, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
if( src.type() == _dst.type() )
|
||||
transpose(src, _dst);
|
||||
else
|
||||
Mat(src.t()).convertTo(_dst, _dst.type());
|
||||
}
|
||||
else
|
||||
{
|
||||
const Mat _dst(src.rows, src.cols, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
src.convertTo(_dst, _dst.type());
|
||||
}
|
||||
}
|
||||
|
||||
// Matx case
|
||||
template<typename _Tp, int _rows> static inline
|
||||
void cv2eigen( const Matx<_Tp, _rows, 1>& src,
|
||||
Eigen::Matrix<_Tp, Eigen::Dynamic, 1>& dst )
|
||||
{
|
||||
dst.resize(_rows);
|
||||
|
||||
if( !(dst.Flags & Eigen::RowMajorBit) )
|
||||
{
|
||||
const Mat _dst(1, _rows, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
transpose(src, _dst);
|
||||
}
|
||||
else
|
||||
{
|
||||
const Mat _dst(_rows, 1, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
src.copyTo(_dst);
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
template<typename _Tp> static inline
|
||||
void cv2eigen( const Mat& src,
|
||||
Eigen::Matrix<_Tp, 1, Eigen::Dynamic>& dst )
|
||||
{
|
||||
CV_Assert(src.rows == 1);
|
||||
dst.resize(src.cols);
|
||||
if( !(dst.Flags & Eigen::RowMajorBit) )
|
||||
{
|
||||
const Mat _dst(src.cols, src.rows, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
if( src.type() == _dst.type() )
|
||||
transpose(src, _dst);
|
||||
else
|
||||
Mat(src.t()).convertTo(_dst, _dst.type());
|
||||
}
|
||||
else
|
||||
{
|
||||
const Mat _dst(src.rows, src.cols, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
src.convertTo(_dst, _dst.type());
|
||||
}
|
||||
}
|
||||
|
||||
//Matx
|
||||
template<typename _Tp, int _cols> static inline
|
||||
void cv2eigen( const Matx<_Tp, 1, _cols>& src,
|
||||
Eigen::Matrix<_Tp, 1, Eigen::Dynamic>& dst )
|
||||
{
|
||||
dst.resize(_cols);
|
||||
if( !(dst.Flags & Eigen::RowMajorBit) )
|
||||
{
|
||||
const Mat _dst(_cols, 1, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
transpose(src, _dst);
|
||||
}
|
||||
else
|
||||
{
|
||||
const Mat _dst(1, _cols, traits::Type<_Tp>::value,
|
||||
dst.data(), (size_t)(dst.outerStride()*sizeof(_Tp)));
|
||||
Mat(src).copyTo(_dst);
|
||||
}
|
||||
}
|
||||
|
||||
//! @}
|
||||
|
||||
} // cv
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,271 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Copyright (C) 2015, Itseez Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_CORE_FAST_MATH_HPP
|
||||
#define OPENCV_CORE_FAST_MATH_HPP
|
||||
|
||||
#include "opencv2/core/cvdef.h"
|
||||
|
||||
#if ((defined _MSC_VER && defined _M_X64) || (defined __GNUC__ && defined __x86_64__ \
|
||||
&& defined __SSE2__ && !defined __APPLE__)) && !defined(__CUDACC__)
|
||||
#include <emmintrin.h>
|
||||
#endif
|
||||
|
||||
|
||||
//! @addtogroup core_utils
|
||||
//! @{
|
||||
|
||||
/****************************************************************************************\
|
||||
* fast math *
|
||||
\****************************************************************************************/
|
||||
|
||||
#ifdef __cplusplus
|
||||
# include <cmath>
|
||||
#else
|
||||
# ifdef __BORLANDC__
|
||||
# include <fastmath.h>
|
||||
# else
|
||||
# include <math.h>
|
||||
# endif
|
||||
#endif
|
||||
|
||||
#ifdef HAVE_TEGRA_OPTIMIZATION
|
||||
# include "tegra_round.hpp"
|
||||
#endif
|
||||
|
||||
#if defined __GNUC__ && defined __arm__ && (defined __ARM_PCS_VFP || defined __ARM_VFPV3__ || defined __ARM_NEON__) && !defined __SOFTFP__ && !defined(__CUDACC__)
|
||||
// 1. general scheme
|
||||
#define ARM_ROUND(_value, _asm_string) \
|
||||
int res; \
|
||||
float temp; \
|
||||
CV_UNUSED(temp); \
|
||||
__asm__(_asm_string : [res] "=r" (res), [temp] "=w" (temp) : [value] "w" (_value)); \
|
||||
return res
|
||||
// 2. version for double
|
||||
#ifdef __clang__
|
||||
#define ARM_ROUND_DBL(value) ARM_ROUND(value, "vcvtr.s32.f64 %[temp], %[value] \n vmov %[res], %[temp]")
|
||||
#else
|
||||
#define ARM_ROUND_DBL(value) ARM_ROUND(value, "vcvtr.s32.f64 %[temp], %P[value] \n vmov %[res], %[temp]")
|
||||
#endif
|
||||
// 3. version for float
|
||||
#define ARM_ROUND_FLT(value) ARM_ROUND(value, "vcvtr.s32.f32 %[temp], %[value]\n vmov %[res], %[temp]")
|
||||
#endif
|
||||
|
||||
/** @brief Rounds floating-point number to the nearest integer
|
||||
|
||||
@param value floating-point number. If the value is outside of INT_MIN ... INT_MAX range, the
|
||||
result is not defined.
|
||||
*/
|
||||
CV_INLINE int
|
||||
cvRound( double value )
|
||||
{
|
||||
#if ((defined _MSC_VER && defined _M_X64) || (defined __GNUC__ && defined __x86_64__ \
|
||||
&& defined __SSE2__ && !defined __APPLE__) || CV_SSE2) && !defined(__CUDACC__)
|
||||
__m128d t = _mm_set_sd( value );
|
||||
return _mm_cvtsd_si32(t);
|
||||
#elif defined _MSC_VER && defined _M_IX86
|
||||
int t;
|
||||
__asm
|
||||
{
|
||||
fld value;
|
||||
fistp t;
|
||||
}
|
||||
return t;
|
||||
#elif ((defined _MSC_VER && defined _M_ARM) || defined CV_ICC || \
|
||||
defined __GNUC__) && defined HAVE_TEGRA_OPTIMIZATION
|
||||
TEGRA_ROUND_DBL(value);
|
||||
#elif defined CV_ICC || defined __GNUC__
|
||||
# if defined ARM_ROUND_DBL
|
||||
ARM_ROUND_DBL(value);
|
||||
# else
|
||||
return (int)lrint(value);
|
||||
# endif
|
||||
#else
|
||||
/* it's ok if round does not comply with IEEE754 standard;
|
||||
the tests should allow +/-1 difference when the tested functions use round */
|
||||
return (int)(value + (value >= 0 ? 0.5 : -0.5));
|
||||
#endif
|
||||
}
|
||||
|
||||
|
||||
/** @brief Rounds floating-point number to the nearest integer not larger than the original.
|
||||
|
||||
The function computes an integer i such that:
|
||||
\f[i \le \texttt{value} < i+1\f]
|
||||
@param value floating-point number. If the value is outside of INT_MIN ... INT_MAX range, the
|
||||
result is not defined.
|
||||
*/
|
||||
CV_INLINE int cvFloor( double value )
|
||||
{
|
||||
int i = (int)value;
|
||||
return i - (i > value);
|
||||
}
|
||||
|
||||
/** @brief Rounds floating-point number to the nearest integer not smaller than the original.
|
||||
|
||||
The function computes an integer i such that:
|
||||
\f[i \le \texttt{value} < i+1\f]
|
||||
@param value floating-point number. If the value is outside of INT_MIN ... INT_MAX range, the
|
||||
result is not defined.
|
||||
*/
|
||||
CV_INLINE int cvCeil( double value )
|
||||
{
|
||||
int i = (int)value;
|
||||
return i + (i < value);
|
||||
}
|
||||
|
||||
/** @brief Determines if the argument is Not A Number.
|
||||
|
||||
@param value The input floating-point value
|
||||
|
||||
The function returns 1 if the argument is Not A Number (as defined by IEEE754 standard), 0
|
||||
otherwise. */
|
||||
CV_INLINE int cvIsNaN( double value )
|
||||
{
|
||||
Cv64suf ieee754;
|
||||
ieee754.f = value;
|
||||
return ((unsigned)(ieee754.u >> 32) & 0x7fffffff) +
|
||||
((unsigned)ieee754.u != 0) > 0x7ff00000;
|
||||
}
|
||||
|
||||
/** @brief Determines if the argument is Infinity.
|
||||
|
||||
@param value The input floating-point value
|
||||
|
||||
The function returns 1 if the argument is a plus or minus infinity (as defined by IEEE754 standard)
|
||||
and 0 otherwise. */
|
||||
CV_INLINE int cvIsInf( double value )
|
||||
{
|
||||
Cv64suf ieee754;
|
||||
ieee754.f = value;
|
||||
return ((unsigned)(ieee754.u >> 32) & 0x7fffffff) == 0x7ff00000 &&
|
||||
(unsigned)ieee754.u == 0;
|
||||
}
|
||||
|
||||
#ifdef __cplusplus
|
||||
|
||||
/** @overload */
|
||||
CV_INLINE int cvRound(float value)
|
||||
{
|
||||
#if ((defined _MSC_VER && defined _M_X64) || (defined __GNUC__ && defined __x86_64__ \
|
||||
&& defined __SSE2__ && !defined __APPLE__) || CV_SSE2) && !defined(__CUDACC__)
|
||||
__m128 t = _mm_set_ss( value );
|
||||
return _mm_cvtss_si32(t);
|
||||
#elif defined _MSC_VER && defined _M_IX86
|
||||
int t;
|
||||
__asm
|
||||
{
|
||||
fld value;
|
||||
fistp t;
|
||||
}
|
||||
return t;
|
||||
#elif ((defined _MSC_VER && defined _M_ARM) || defined CV_ICC || \
|
||||
defined __GNUC__) && defined HAVE_TEGRA_OPTIMIZATION
|
||||
TEGRA_ROUND_FLT(value);
|
||||
#elif defined CV_ICC || defined __GNUC__
|
||||
# if defined ARM_ROUND_FLT
|
||||
ARM_ROUND_FLT(value);
|
||||
# else
|
||||
return (int)lrintf(value);
|
||||
# endif
|
||||
#else
|
||||
/* it's ok if round does not comply with IEEE754 standard;
|
||||
the tests should allow +/-1 difference when the tested functions use round */
|
||||
return (int)(value + (value >= 0 ? 0.5f : -0.5f));
|
||||
#endif
|
||||
}
|
||||
|
||||
/** @overload */
|
||||
CV_INLINE int cvRound( int value )
|
||||
{
|
||||
return value;
|
||||
}
|
||||
|
||||
/** @overload */
|
||||
CV_INLINE int cvFloor( float value )
|
||||
{
|
||||
int i = (int)value;
|
||||
return i - (i > value);
|
||||
}
|
||||
|
||||
/** @overload */
|
||||
CV_INLINE int cvFloor( int value )
|
||||
{
|
||||
return value;
|
||||
}
|
||||
|
||||
/** @overload */
|
||||
CV_INLINE int cvCeil( float value )
|
||||
{
|
||||
int i = (int)value;
|
||||
return i + (i < value);
|
||||
}
|
||||
|
||||
/** @overload */
|
||||
CV_INLINE int cvCeil( int value )
|
||||
{
|
||||
return value;
|
||||
}
|
||||
|
||||
/** @overload */
|
||||
CV_INLINE int cvIsNaN( float value )
|
||||
{
|
||||
Cv32suf ieee754;
|
||||
ieee754.f = value;
|
||||
return (ieee754.u & 0x7fffffff) > 0x7f800000;
|
||||
}
|
||||
|
||||
/** @overload */
|
||||
CV_INLINE int cvIsInf( float value )
|
||||
{
|
||||
Cv32suf ieee754;
|
||||
ieee754.f = value;
|
||||
return (ieee754.u & 0x7fffffff) == 0x7f800000;
|
||||
}
|
||||
|
||||
#endif // __cplusplus
|
||||
|
||||
//! @} core_utils
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,250 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Copyright (C) 2015, Itseez Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_HAL_HPP
|
||||
#define OPENCV_HAL_HPP
|
||||
|
||||
#include "opencv2/core/cvdef.h"
|
||||
#include "opencv2/core/cvstd.hpp"
|
||||
#include "opencv2/core/hal/interface.h"
|
||||
|
||||
namespace cv { namespace hal {
|
||||
|
||||
//! @addtogroup core_hal_functions
|
||||
//! @{
|
||||
|
||||
CV_EXPORTS int normHamming(const uchar* a, int n);
|
||||
CV_EXPORTS int normHamming(const uchar* a, const uchar* b, int n);
|
||||
|
||||
CV_EXPORTS int normHamming(const uchar* a, int n, int cellSize);
|
||||
CV_EXPORTS int normHamming(const uchar* a, const uchar* b, int n, int cellSize);
|
||||
|
||||
CV_EXPORTS int LU32f(float* A, size_t astep, int m, float* b, size_t bstep, int n);
|
||||
CV_EXPORTS int LU64f(double* A, size_t astep, int m, double* b, size_t bstep, int n);
|
||||
CV_EXPORTS bool Cholesky32f(float* A, size_t astep, int m, float* b, size_t bstep, int n);
|
||||
CV_EXPORTS bool Cholesky64f(double* A, size_t astep, int m, double* b, size_t bstep, int n);
|
||||
CV_EXPORTS void SVD32f(float* At, size_t astep, float* W, float* U, size_t ustep, float* Vt, size_t vstep, int m, int n, int flags);
|
||||
CV_EXPORTS void SVD64f(double* At, size_t astep, double* W, double* U, size_t ustep, double* Vt, size_t vstep, int m, int n, int flags);
|
||||
CV_EXPORTS int QR32f(float* A, size_t astep, int m, int n, int k, float* b, size_t bstep, float* hFactors);
|
||||
CV_EXPORTS int QR64f(double* A, size_t astep, int m, int n, int k, double* b, size_t bstep, double* hFactors);
|
||||
|
||||
CV_EXPORTS void gemm32f(const float* src1, size_t src1_step, const float* src2, size_t src2_step,
|
||||
float alpha, const float* src3, size_t src3_step, float beta, float* dst, size_t dst_step,
|
||||
int m_a, int n_a, int n_d, int flags);
|
||||
CV_EXPORTS void gemm64f(const double* src1, size_t src1_step, const double* src2, size_t src2_step,
|
||||
double alpha, const double* src3, size_t src3_step, double beta, double* dst, size_t dst_step,
|
||||
int m_a, int n_a, int n_d, int flags);
|
||||
CV_EXPORTS void gemm32fc(const float* src1, size_t src1_step, const float* src2, size_t src2_step,
|
||||
float alpha, const float* src3, size_t src3_step, float beta, float* dst, size_t dst_step,
|
||||
int m_a, int n_a, int n_d, int flags);
|
||||
CV_EXPORTS void gemm64fc(const double* src1, size_t src1_step, const double* src2, size_t src2_step,
|
||||
double alpha, const double* src3, size_t src3_step, double beta, double* dst, size_t dst_step,
|
||||
int m_a, int n_a, int n_d, int flags);
|
||||
|
||||
CV_EXPORTS int normL1_(const uchar* a, const uchar* b, int n);
|
||||
CV_EXPORTS float normL1_(const float* a, const float* b, int n);
|
||||
CV_EXPORTS float normL2Sqr_(const float* a, const float* b, int n);
|
||||
|
||||
CV_EXPORTS void exp32f(const float* src, float* dst, int n);
|
||||
CV_EXPORTS void exp64f(const double* src, double* dst, int n);
|
||||
CV_EXPORTS void log32f(const float* src, float* dst, int n);
|
||||
CV_EXPORTS void log64f(const double* src, double* dst, int n);
|
||||
|
||||
CV_EXPORTS void fastAtan32f(const float* y, const float* x, float* dst, int n, bool angleInDegrees);
|
||||
CV_EXPORTS void fastAtan64f(const double* y, const double* x, double* dst, int n, bool angleInDegrees);
|
||||
CV_EXPORTS void magnitude32f(const float* x, const float* y, float* dst, int n);
|
||||
CV_EXPORTS void magnitude64f(const double* x, const double* y, double* dst, int n);
|
||||
CV_EXPORTS void sqrt32f(const float* src, float* dst, int len);
|
||||
CV_EXPORTS void sqrt64f(const double* src, double* dst, int len);
|
||||
CV_EXPORTS void invSqrt32f(const float* src, float* dst, int len);
|
||||
CV_EXPORTS void invSqrt64f(const double* src, double* dst, int len);
|
||||
|
||||
CV_EXPORTS void split8u(const uchar* src, uchar** dst, int len, int cn );
|
||||
CV_EXPORTS void split16u(const ushort* src, ushort** dst, int len, int cn );
|
||||
CV_EXPORTS void split32s(const int* src, int** dst, int len, int cn );
|
||||
CV_EXPORTS void split64s(const int64* src, int64** dst, int len, int cn );
|
||||
|
||||
CV_EXPORTS void merge8u(const uchar** src, uchar* dst, int len, int cn );
|
||||
CV_EXPORTS void merge16u(const ushort** src, ushort* dst, int len, int cn );
|
||||
CV_EXPORTS void merge32s(const int** src, int* dst, int len, int cn );
|
||||
CV_EXPORTS void merge64s(const int64** src, int64* dst, int len, int cn );
|
||||
|
||||
CV_EXPORTS void add8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void add8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void add16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void add16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void add32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void add32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void add64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* );
|
||||
|
||||
CV_EXPORTS void sub8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void sub8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void sub16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void sub16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void sub32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void sub32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void sub64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* );
|
||||
|
||||
CV_EXPORTS void max8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void max8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void max16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void max16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void max32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void max32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void max64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* );
|
||||
|
||||
CV_EXPORTS void min8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void min8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void min16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void min16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void min32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void min32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void min64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* );
|
||||
|
||||
CV_EXPORTS void absdiff8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void absdiff8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void absdiff16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void absdiff16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void absdiff32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void absdiff32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void absdiff64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* );
|
||||
|
||||
CV_EXPORTS void and8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void or8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void xor8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
|
||||
CV_EXPORTS void not8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* );
|
||||
|
||||
CV_EXPORTS void cmp8u(const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
|
||||
CV_EXPORTS void cmp8s(const schar* src1, size_t step1, const schar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
|
||||
CV_EXPORTS void cmp16u(const ushort* src1, size_t step1, const ushort* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
|
||||
CV_EXPORTS void cmp16s(const short* src1, size_t step1, const short* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
|
||||
CV_EXPORTS void cmp32s(const int* src1, size_t step1, const int* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
|
||||
CV_EXPORTS void cmp32f(const float* src1, size_t step1, const float* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
|
||||
CV_EXPORTS void cmp64f(const double* src1, size_t step1, const double* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _cmpop);
|
||||
|
||||
CV_EXPORTS void mul8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void mul8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void mul16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void mul16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void mul32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void mul32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void mul64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* scale);
|
||||
|
||||
CV_EXPORTS void div8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void div8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void div16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void div16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void div32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void div32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void div64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* scale);
|
||||
|
||||
CV_EXPORTS void recip8u( const uchar *, size_t, const uchar * src2, size_t step2, uchar* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void recip8s( const schar *, size_t, const schar * src2, size_t step2, schar* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void recip16u( const ushort *, size_t, const ushort * src2, size_t step2, ushort* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void recip16s( const short *, size_t, const short * src2, size_t step2, short* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void recip32s( const int *, size_t, const int * src2, size_t step2, int* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void recip32f( const float *, size_t, const float * src2, size_t step2, float* dst, size_t step, int width, int height, void* scale);
|
||||
CV_EXPORTS void recip64f( const double *, size_t, const double * src2, size_t step2, double* dst, size_t step, int width, int height, void* scale);
|
||||
|
||||
CV_EXPORTS void addWeighted8u( const uchar* src1, size_t step1, const uchar* src2, size_t step2, uchar* dst, size_t step, int width, int height, void* _scalars );
|
||||
CV_EXPORTS void addWeighted8s( const schar* src1, size_t step1, const schar* src2, size_t step2, schar* dst, size_t step, int width, int height, void* scalars );
|
||||
CV_EXPORTS void addWeighted16u( const ushort* src1, size_t step1, const ushort* src2, size_t step2, ushort* dst, size_t step, int width, int height, void* scalars );
|
||||
CV_EXPORTS void addWeighted16s( const short* src1, size_t step1, const short* src2, size_t step2, short* dst, size_t step, int width, int height, void* scalars );
|
||||
CV_EXPORTS void addWeighted32s( const int* src1, size_t step1, const int* src2, size_t step2, int* dst, size_t step, int width, int height, void* scalars );
|
||||
CV_EXPORTS void addWeighted32f( const float* src1, size_t step1, const float* src2, size_t step2, float* dst, size_t step, int width, int height, void* scalars );
|
||||
CV_EXPORTS void addWeighted64f( const double* src1, size_t step1, const double* src2, size_t step2, double* dst, size_t step, int width, int height, void* scalars );
|
||||
|
||||
struct CV_EXPORTS DFT1D
|
||||
{
|
||||
static Ptr<DFT1D> create(int len, int count, int depth, int flags, bool * useBuffer = 0);
|
||||
virtual void apply(const uchar *src, uchar *dst) = 0;
|
||||
virtual ~DFT1D() {}
|
||||
};
|
||||
|
||||
struct CV_EXPORTS DFT2D
|
||||
{
|
||||
static Ptr<DFT2D> create(int width, int height, int depth,
|
||||
int src_channels, int dst_channels,
|
||||
int flags, int nonzero_rows = 0);
|
||||
virtual void apply(const uchar *src_data, size_t src_step, uchar *dst_data, size_t dst_step) = 0;
|
||||
virtual ~DFT2D() {}
|
||||
};
|
||||
|
||||
struct CV_EXPORTS DCT2D
|
||||
{
|
||||
static Ptr<DCT2D> create(int width, int height, int depth, int flags);
|
||||
virtual void apply(const uchar *src_data, size_t src_step, uchar *dst_data, size_t dst_step) = 0;
|
||||
virtual ~DCT2D() {}
|
||||
};
|
||||
|
||||
//! @} core_hal
|
||||
|
||||
//=============================================================================
|
||||
// for binary compatibility with 3.0
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
CV_EXPORTS int LU(float* A, size_t astep, int m, float* b, size_t bstep, int n);
|
||||
CV_EXPORTS int LU(double* A, size_t astep, int m, double* b, size_t bstep, int n);
|
||||
CV_EXPORTS bool Cholesky(float* A, size_t astep, int m, float* b, size_t bstep, int n);
|
||||
CV_EXPORTS bool Cholesky(double* A, size_t astep, int m, double* b, size_t bstep, int n);
|
||||
|
||||
CV_EXPORTS void exp(const float* src, float* dst, int n);
|
||||
CV_EXPORTS void exp(const double* src, double* dst, int n);
|
||||
CV_EXPORTS void log(const float* src, float* dst, int n);
|
||||
CV_EXPORTS void log(const double* src, double* dst, int n);
|
||||
|
||||
CV_EXPORTS void fastAtan2(const float* y, const float* x, float* dst, int n, bool angleInDegrees);
|
||||
CV_EXPORTS void magnitude(const float* x, const float* y, float* dst, int n);
|
||||
CV_EXPORTS void magnitude(const double* x, const double* y, double* dst, int n);
|
||||
CV_EXPORTS void sqrt(const float* src, float* dst, int len);
|
||||
CV_EXPORTS void sqrt(const double* src, double* dst, int len);
|
||||
CV_EXPORTS void invSqrt(const float* src, float* dst, int len);
|
||||
CV_EXPORTS void invSqrt(const double* src, double* dst, int len);
|
||||
|
||||
//! @endcond
|
||||
|
||||
}} //cv::hal
|
||||
|
||||
#endif //OPENCV_HAL_HPP
|
||||
|
|
@ -0,0 +1,182 @@
|
|||
#ifndef OPENCV_CORE_HAL_INTERFACE_H
|
||||
#define OPENCV_CORE_HAL_INTERFACE_H
|
||||
|
||||
//! @addtogroup core_hal_interface
|
||||
//! @{
|
||||
|
||||
//! @name Return codes
|
||||
//! @{
|
||||
#define CV_HAL_ERROR_OK 0
|
||||
#define CV_HAL_ERROR_NOT_IMPLEMENTED 1
|
||||
#define CV_HAL_ERROR_UNKNOWN -1
|
||||
//! @}
|
||||
|
||||
#ifdef __cplusplus
|
||||
#include <cstddef>
|
||||
#else
|
||||
#include <stddef.h>
|
||||
#include <stdbool.h>
|
||||
#endif
|
||||
|
||||
//! @name Data types
|
||||
//! primitive types
|
||||
//! - schar - signed 1 byte integer
|
||||
//! - uchar - unsigned 1 byte integer
|
||||
//! - short - signed 2 byte integer
|
||||
//! - ushort - unsigned 2 byte integer
|
||||
//! - int - signed 4 byte integer
|
||||
//! - uint - unsigned 4 byte integer
|
||||
//! - int64 - signed 8 byte integer
|
||||
//! - uint64 - unsigned 8 byte integer
|
||||
//! @{
|
||||
#if !defined _MSC_VER && !defined __BORLANDC__
|
||||
# if defined __cplusplus && __cplusplus >= 201103L && !defined __APPLE__
|
||||
# include <cstdint>
|
||||
# ifdef __NEWLIB__
|
||||
typedef unsigned int uint;
|
||||
# else
|
||||
typedef std::uint32_t uint;
|
||||
# endif
|
||||
# else
|
||||
# include <stdint.h>
|
||||
typedef uint32_t uint;
|
||||
# endif
|
||||
#else
|
||||
typedef unsigned uint;
|
||||
#endif
|
||||
|
||||
typedef signed char schar;
|
||||
|
||||
#ifndef __IPL_H__
|
||||
typedef unsigned char uchar;
|
||||
typedef unsigned short ushort;
|
||||
#endif
|
||||
|
||||
#if defined _MSC_VER || defined __BORLANDC__
|
||||
typedef __int64 int64;
|
||||
typedef unsigned __int64 uint64;
|
||||
# define CV_BIG_INT(n) n##I64
|
||||
# define CV_BIG_UINT(n) n##UI64
|
||||
#else
|
||||
typedef int64_t int64;
|
||||
typedef uint64_t uint64;
|
||||
# define CV_BIG_INT(n) n##LL
|
||||
# define CV_BIG_UINT(n) n##ULL
|
||||
#endif
|
||||
|
||||
#define CV_CN_MAX 512
|
||||
#define CV_CN_SHIFT 3
|
||||
#define CV_DEPTH_MAX (1 << CV_CN_SHIFT)
|
||||
|
||||
#define CV_8U 0
|
||||
#define CV_8S 1
|
||||
#define CV_16U 2
|
||||
#define CV_16S 3
|
||||
#define CV_32S 4
|
||||
#define CV_32F 5
|
||||
#define CV_64F 6
|
||||
#define CV_USRTYPE1 7
|
||||
|
||||
#define CV_MAT_DEPTH_MASK (CV_DEPTH_MAX - 1)
|
||||
#define CV_MAT_DEPTH(flags) ((flags) & CV_MAT_DEPTH_MASK)
|
||||
|
||||
#define CV_MAKETYPE(depth,cn) (CV_MAT_DEPTH(depth) + (((cn)-1) << CV_CN_SHIFT))
|
||||
#define CV_MAKE_TYPE CV_MAKETYPE
|
||||
|
||||
#define CV_8UC1 CV_MAKETYPE(CV_8U,1)
|
||||
#define CV_8UC2 CV_MAKETYPE(CV_8U,2)
|
||||
#define CV_8UC3 CV_MAKETYPE(CV_8U,3)
|
||||
#define CV_8UC4 CV_MAKETYPE(CV_8U,4)
|
||||
#define CV_8UC(n) CV_MAKETYPE(CV_8U,(n))
|
||||
|
||||
#define CV_8SC1 CV_MAKETYPE(CV_8S,1)
|
||||
#define CV_8SC2 CV_MAKETYPE(CV_8S,2)
|
||||
#define CV_8SC3 CV_MAKETYPE(CV_8S,3)
|
||||
#define CV_8SC4 CV_MAKETYPE(CV_8S,4)
|
||||
#define CV_8SC(n) CV_MAKETYPE(CV_8S,(n))
|
||||
|
||||
#define CV_16UC1 CV_MAKETYPE(CV_16U,1)
|
||||
#define CV_16UC2 CV_MAKETYPE(CV_16U,2)
|
||||
#define CV_16UC3 CV_MAKETYPE(CV_16U,3)
|
||||
#define CV_16UC4 CV_MAKETYPE(CV_16U,4)
|
||||
#define CV_16UC(n) CV_MAKETYPE(CV_16U,(n))
|
||||
|
||||
#define CV_16SC1 CV_MAKETYPE(CV_16S,1)
|
||||
#define CV_16SC2 CV_MAKETYPE(CV_16S,2)
|
||||
#define CV_16SC3 CV_MAKETYPE(CV_16S,3)
|
||||
#define CV_16SC4 CV_MAKETYPE(CV_16S,4)
|
||||
#define CV_16SC(n) CV_MAKETYPE(CV_16S,(n))
|
||||
|
||||
#define CV_32SC1 CV_MAKETYPE(CV_32S,1)
|
||||
#define CV_32SC2 CV_MAKETYPE(CV_32S,2)
|
||||
#define CV_32SC3 CV_MAKETYPE(CV_32S,3)
|
||||
#define CV_32SC4 CV_MAKETYPE(CV_32S,4)
|
||||
#define CV_32SC(n) CV_MAKETYPE(CV_32S,(n))
|
||||
|
||||
#define CV_32FC1 CV_MAKETYPE(CV_32F,1)
|
||||
#define CV_32FC2 CV_MAKETYPE(CV_32F,2)
|
||||
#define CV_32FC3 CV_MAKETYPE(CV_32F,3)
|
||||
#define CV_32FC4 CV_MAKETYPE(CV_32F,4)
|
||||
#define CV_32FC(n) CV_MAKETYPE(CV_32F,(n))
|
||||
|
||||
#define CV_64FC1 CV_MAKETYPE(CV_64F,1)
|
||||
#define CV_64FC2 CV_MAKETYPE(CV_64F,2)
|
||||
#define CV_64FC3 CV_MAKETYPE(CV_64F,3)
|
||||
#define CV_64FC4 CV_MAKETYPE(CV_64F,4)
|
||||
#define CV_64FC(n) CV_MAKETYPE(CV_64F,(n))
|
||||
//! @}
|
||||
|
||||
//! @name Comparison operation
|
||||
//! @sa cv::CmpTypes
|
||||
//! @{
|
||||
#define CV_HAL_CMP_EQ 0
|
||||
#define CV_HAL_CMP_GT 1
|
||||
#define CV_HAL_CMP_GE 2
|
||||
#define CV_HAL_CMP_LT 3
|
||||
#define CV_HAL_CMP_LE 4
|
||||
#define CV_HAL_CMP_NE 5
|
||||
//! @}
|
||||
|
||||
//! @name Border processing modes
|
||||
//! @sa cv::BorderTypes
|
||||
//! @{
|
||||
#define CV_HAL_BORDER_CONSTANT 0
|
||||
#define CV_HAL_BORDER_REPLICATE 1
|
||||
#define CV_HAL_BORDER_REFLECT 2
|
||||
#define CV_HAL_BORDER_WRAP 3
|
||||
#define CV_HAL_BORDER_REFLECT_101 4
|
||||
#define CV_HAL_BORDER_TRANSPARENT 5
|
||||
#define CV_HAL_BORDER_ISOLATED 16
|
||||
//! @}
|
||||
|
||||
//! @name DFT flags
|
||||
//! @{
|
||||
#define CV_HAL_DFT_INVERSE 1
|
||||
#define CV_HAL_DFT_SCALE 2
|
||||
#define CV_HAL_DFT_ROWS 4
|
||||
#define CV_HAL_DFT_COMPLEX_OUTPUT 16
|
||||
#define CV_HAL_DFT_REAL_OUTPUT 32
|
||||
#define CV_HAL_DFT_TWO_STAGE 64
|
||||
#define CV_HAL_DFT_STAGE_COLS 128
|
||||
#define CV_HAL_DFT_IS_CONTINUOUS 512
|
||||
#define CV_HAL_DFT_IS_INPLACE 1024
|
||||
//! @}
|
||||
|
||||
//! @name SVD flags
|
||||
//! @{
|
||||
#define CV_HAL_SVD_NO_UV 1
|
||||
#define CV_HAL_SVD_SHORT_UV 2
|
||||
#define CV_HAL_SVD_MODIFY_A 4
|
||||
#define CV_HAL_SVD_FULL_UV 8
|
||||
//! @}
|
||||
|
||||
//! @name Gemm flags
|
||||
//! @{
|
||||
#define CV_HAL_GEMM_1_T 1
|
||||
#define CV_HAL_GEMM_2_T 2
|
||||
#define CV_HAL_GEMM_3_T 4
|
||||
//! @}
|
||||
|
||||
//! @}
|
||||
|
||||
#endif
|
||||
|
|
@ -0,0 +1,420 @@
|
|||
/*M///////////////////////////////////////////////////////////////////////////////////////
|
||||
//
|
||||
// IMPORTANT: READ BEFORE DOWNLOADING, COPYING, INSTALLING OR USING.
|
||||
//
|
||||
// By downloading, copying, installing or using the software you agree to this license.
|
||||
// If you do not agree to this license, do not download, install,
|
||||
// copy or use the software.
|
||||
//
|
||||
//
|
||||
// License Agreement
|
||||
// For Open Source Computer Vision Library
|
||||
//
|
||||
// Copyright (C) 2000-2008, Intel Corporation, all rights reserved.
|
||||
// Copyright (C) 2009, Willow Garage Inc., all rights reserved.
|
||||
// Copyright (C) 2013, OpenCV Foundation, all rights reserved.
|
||||
// Copyright (C) 2015, Itseez Inc., all rights reserved.
|
||||
// Third party copyrights are property of their respective owners.
|
||||
//
|
||||
// Redistribution and use in source and binary forms, with or without modification,
|
||||
// are permitted provided that the following conditions are met:
|
||||
//
|
||||
// * Redistribution's of source code must retain the above copyright notice,
|
||||
// this list of conditions and the following disclaimer.
|
||||
//
|
||||
// * Redistribution's in binary form must reproduce the above copyright notice,
|
||||
// this list of conditions and the following disclaimer in the documentation
|
||||
// and/or other materials provided with the distribution.
|
||||
//
|
||||
// * The name of the copyright holders may not be used to endorse or promote products
|
||||
// derived from this software without specific prior written permission.
|
||||
//
|
||||
// This software is provided by the copyright holders and contributors "as is" and
|
||||
// any express or implied warranties, including, but not limited to, the implied
|
||||
// warranties of merchantability and fitness for a particular purpose are disclaimed.
|
||||
// In no event shall the Intel Corporation or contributors be liable for any direct,
|
||||
// indirect, incidental, special, exemplary, or consequential damages
|
||||
// (including, but not limited to, procurement of substitute goods or services;
|
||||
// loss of use, data, or profits; or business interruption) however caused
|
||||
// and on any theory of liability, whether in contract, strict liability,
|
||||
// or tort (including negligence or otherwise) arising in any way out of
|
||||
// the use of this software, even if advised of the possibility of such damage.
|
||||
//
|
||||
//M*/
|
||||
|
||||
#ifndef OPENCV_HAL_INTRIN_HPP
|
||||
#define OPENCV_HAL_INTRIN_HPP
|
||||
|
||||
#include <cmath>
|
||||
#include <float.h>
|
||||
#include <stdlib.h>
|
||||
#include "opencv2/core/cvdef.h"
|
||||
|
||||
#define OPENCV_HAL_ADD(a, b) ((a) + (b))
|
||||
#define OPENCV_HAL_AND(a, b) ((a) & (b))
|
||||
#define OPENCV_HAL_NOP(a) (a)
|
||||
#define OPENCV_HAL_1ST(a, b) (a)
|
||||
|
||||
// unlike HAL API, which is in cv::hal,
|
||||
// we put intrinsics into cv namespace to make its
|
||||
// access from within opencv code more accessible
|
||||
namespace cv {
|
||||
|
||||
namespace hal {
|
||||
|
||||
enum StoreMode
|
||||
{
|
||||
STORE_UNALIGNED = 0,
|
||||
STORE_ALIGNED = 1,
|
||||
STORE_ALIGNED_NOCACHE = 2
|
||||
};
|
||||
|
||||
}
|
||||
|
||||
template<typename _Tp> struct V_TypeTraits
|
||||
{
|
||||
};
|
||||
|
||||
#define CV_INTRIN_DEF_TYPE_TRAITS(type, int_type_, uint_type_, abs_type_, w_type_, q_type_, sum_type_, nlanes128_) \
|
||||
template<> struct V_TypeTraits<type> \
|
||||
{ \
|
||||
typedef type value_type; \
|
||||
typedef int_type_ int_type; \
|
||||
typedef abs_type_ abs_type; \
|
||||
typedef uint_type_ uint_type; \
|
||||
typedef w_type_ w_type; \
|
||||
typedef q_type_ q_type; \
|
||||
typedef sum_type_ sum_type; \
|
||||
enum { nlanes128 = nlanes128_ }; \
|
||||
\
|
||||
static inline int_type reinterpret_int(type x) \
|
||||
{ \
|
||||
union { type l; int_type i; } v; \
|
||||
v.l = x; \
|
||||
return v.i; \
|
||||
} \
|
||||
\
|
||||
static inline type reinterpret_from_int(int_type x) \
|
||||
{ \
|
||||
union { type l; int_type i; } v; \
|
||||
v.i = x; \
|
||||
return v.l; \
|
||||
} \
|
||||
}
|
||||
|
||||
CV_INTRIN_DEF_TYPE_TRAITS(uchar, schar, uchar, uchar, ushort, unsigned, unsigned, 16);
|
||||
CV_INTRIN_DEF_TYPE_TRAITS(schar, schar, uchar, uchar, short, int, int, 16);
|
||||
CV_INTRIN_DEF_TYPE_TRAITS(ushort, short, ushort, ushort, unsigned, uint64, unsigned, 8);
|
||||
CV_INTRIN_DEF_TYPE_TRAITS(short, short, ushort, ushort, int, int64, int, 8);
|
||||
CV_INTRIN_DEF_TYPE_TRAITS(unsigned, int, unsigned, unsigned, uint64, void, unsigned, 4);
|
||||
CV_INTRIN_DEF_TYPE_TRAITS(int, int, unsigned, unsigned, int64, void, int, 4);
|
||||
CV_INTRIN_DEF_TYPE_TRAITS(float, int, unsigned, float, double, void, float, 4);
|
||||
CV_INTRIN_DEF_TYPE_TRAITS(uint64, int64, uint64, uint64, void, void, uint64, 2);
|
||||
CV_INTRIN_DEF_TYPE_TRAITS(int64, int64, uint64, uint64, void, void, int64, 2);
|
||||
CV_INTRIN_DEF_TYPE_TRAITS(double, int64, uint64, double, void, void, double, 2);
|
||||
|
||||
#ifndef CV_DOXYGEN
|
||||
|
||||
#ifdef CV_CPU_DISPATCH_MODE
|
||||
#define CV_CPU_OPTIMIZATION_HAL_NAMESPACE __CV_CAT(hal_, CV_CPU_DISPATCH_MODE)
|
||||
#define CV_CPU_OPTIMIZATION_HAL_NAMESPACE_BEGIN namespace __CV_CAT(hal_, CV_CPU_DISPATCH_MODE) {
|
||||
#define CV_CPU_OPTIMIZATION_HAL_NAMESPACE_END }
|
||||
#else
|
||||
#define CV_CPU_OPTIMIZATION_HAL_NAMESPACE hal_baseline
|
||||
#define CV_CPU_OPTIMIZATION_HAL_NAMESPACE_BEGIN namespace hal_baseline {
|
||||
#define CV_CPU_OPTIMIZATION_HAL_NAMESPACE_END }
|
||||
#endif
|
||||
|
||||
CV_CPU_OPTIMIZATION_HAL_NAMESPACE_BEGIN
|
||||
CV_CPU_OPTIMIZATION_HAL_NAMESPACE_END
|
||||
using namespace CV_CPU_OPTIMIZATION_HAL_NAMESPACE;
|
||||
#endif
|
||||
}
|
||||
|
||||
#ifdef CV_DOXYGEN
|
||||
# undef CV_AVX2
|
||||
# undef CV_SSE2
|
||||
# undef CV_NEON
|
||||
# undef CV_VSX
|
||||
# undef CV_FP16
|
||||
#endif
|
||||
|
||||
#if CV_SSE2 || CV_NEON || CV_VSX
|
||||
#define CV__SIMD_FORWARD 128
|
||||
#include "opencv2/core/hal/intrin_forward.hpp"
|
||||
#endif
|
||||
|
||||
#if CV_SSE2
|
||||
|
||||
#include "opencv2/core/hal/intrin_sse_em.hpp"
|
||||
#include "opencv2/core/hal/intrin_sse.hpp"
|
||||
|
||||
#elif CV_NEON
|
||||
|
||||
#include "opencv2/core/hal/intrin_neon.hpp"
|
||||
|
||||
#elif CV_VSX
|
||||
|
||||
#include "opencv2/core/hal/intrin_vsx.hpp"
|
||||
|
||||
#else
|
||||
|
||||
#define CV_SIMD128_CPP 1
|
||||
#include "opencv2/core/hal/intrin_cpp.hpp"
|
||||
|
||||
#endif
|
||||
|
||||
// AVX2 can be used together with SSE2, so
|
||||
// we define those two sets of intrinsics at once.
|
||||
// Most of the intrinsics do not conflict (the proper overloaded variant is
|
||||
// resolved by the argument types, e.g. v_float32x4 ~ SSE2, v_float32x8 ~ AVX2),
|
||||
// but some of AVX2 intrinsics get v256_ prefix instead of v_, e.g. v256_load() vs v_load().
|
||||
// Correspondingly, the wide intrinsics (which are mapped to the "widest"
|
||||
// available instruction set) will get vx_ prefix
|
||||
// (and will be mapped to v256_ counterparts) (e.g. vx_load() => v256_load())
|
||||
#if CV_AVX2
|
||||
|
||||
#define CV__SIMD_FORWARD 256
|
||||
#include "opencv2/core/hal/intrin_forward.hpp"
|
||||
#include "opencv2/core/hal/intrin_avx.hpp"
|
||||
|
||||
#endif
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
namespace cv {
|
||||
|
||||
#ifndef CV_DOXYGEN
|
||||
CV_CPU_OPTIMIZATION_HAL_NAMESPACE_BEGIN
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD128
|
||||
#define CV_SIMD128 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD128_64F
|
||||
#define CV_SIMD128_64F 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD256
|
||||
#define CV_SIMD256 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD256_64F
|
||||
#define CV_SIMD256_64F 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD512
|
||||
#define CV_SIMD512 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD512_64F
|
||||
#define CV_SIMD512_64F 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD128_FP16
|
||||
#define CV_SIMD128_FP16 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD256_FP16
|
||||
#define CV_SIMD256_FP16 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD512_FP16
|
||||
#define CV_SIMD512_FP16 0
|
||||
#endif
|
||||
|
||||
//==================================================================================================
|
||||
|
||||
#define CV_INTRIN_DEFINE_WIDE_INTRIN(typ, vtyp, short_typ, prefix, loadsfx) \
|
||||
inline vtyp vx_setall_##short_typ(typ v) { return prefix##_setall_##short_typ(v); } \
|
||||
inline vtyp vx_setzero_##short_typ() { return prefix##_setzero_##short_typ(); } \
|
||||
inline vtyp vx_##loadsfx(const typ* ptr) { return prefix##_##loadsfx(ptr); } \
|
||||
inline vtyp vx_##loadsfx##_aligned(const typ* ptr) { return prefix##_##loadsfx##_aligned(ptr); } \
|
||||
inline vtyp vx_##loadsfx##_low(const typ* ptr) { return prefix##_##loadsfx##_low(ptr); } \
|
||||
inline vtyp vx_##loadsfx##_halves(const typ* ptr0, const typ* ptr1) { return prefix##_##loadsfx##_halves(ptr0, ptr1); } \
|
||||
inline void vx_store(typ* ptr, const vtyp& v) { return v_store(ptr, v); } \
|
||||
inline void vx_store_aligned(typ* ptr, const vtyp& v) { return v_store_aligned(ptr, v); }
|
||||
|
||||
#define CV_INTRIN_DEFINE_WIDE_LOAD_EXPAND(typ, wtyp, prefix) \
|
||||
inline wtyp vx_load_expand(const typ* ptr) { return prefix##_load_expand(ptr); }
|
||||
|
||||
#define CV_INTRIN_DEFINE_WIDE_LOAD_EXPAND_Q(typ, qtyp, prefix) \
|
||||
inline qtyp vx_load_expand_q(const typ* ptr) { return prefix##_load_expand_q(ptr); }
|
||||
|
||||
#define CV_INTRIN_DEFINE_WIDE_INTRIN_WITH_EXPAND(typ, vtyp, short_typ, wtyp, qtyp, prefix, loadsfx) \
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN(typ, vtyp, short_typ, prefix, loadsfx) \
|
||||
CV_INTRIN_DEFINE_WIDE_LOAD_EXPAND(typ, wtyp, prefix) \
|
||||
CV_INTRIN_DEFINE_WIDE_LOAD_EXPAND_Q(typ, qtyp, prefix)
|
||||
|
||||
#define CV_INTRIN_DEFINE_WIDE_INTRIN_ALL_TYPES(prefix) \
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN_WITH_EXPAND(uchar, v_uint8, u8, v_uint16, v_uint32, prefix, load) \
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN_WITH_EXPAND(schar, v_int8, s8, v_int16, v_int32, prefix, load) \
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN(ushort, v_uint16, u16, prefix, load) \
|
||||
CV_INTRIN_DEFINE_WIDE_LOAD_EXPAND(ushort, v_uint32, prefix) \
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN(short, v_int16, s16, prefix, load) \
|
||||
CV_INTRIN_DEFINE_WIDE_LOAD_EXPAND(short, v_int32, prefix) \
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN(int, v_int32, s32, prefix, load) \
|
||||
CV_INTRIN_DEFINE_WIDE_LOAD_EXPAND(int, v_int64, prefix) \
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN(unsigned, v_uint32, u32, prefix, load) \
|
||||
CV_INTRIN_DEFINE_WIDE_LOAD_EXPAND(unsigned, v_uint64, prefix) \
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN(float, v_float32, f32, prefix, load) \
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN(int64, v_int64, s64, prefix, load) \
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN(uint64, v_uint64, u64, prefix, load) \
|
||||
CV_INTRIN_DEFINE_WIDE_LOAD_EXPAND(float16_t, v_float32, prefix)
|
||||
|
||||
template<typename _Tp> struct V_RegTraits
|
||||
{
|
||||
};
|
||||
|
||||
#define CV_DEF_REG_TRAITS(prefix, _reg, lane_type, suffix, _u_reg, _w_reg, _q_reg, _int_reg, _round_reg) \
|
||||
template<> struct V_RegTraits<_reg> \
|
||||
{ \
|
||||
typedef _reg reg; \
|
||||
typedef _u_reg u_reg; \
|
||||
typedef _w_reg w_reg; \
|
||||
typedef _q_reg q_reg; \
|
||||
typedef _int_reg int_reg; \
|
||||
typedef _round_reg round_reg; \
|
||||
}
|
||||
|
||||
#if CV_SIMD128 || CV_SIMD128_CPP
|
||||
CV_DEF_REG_TRAITS(v, v_uint8x16, uchar, u8, v_uint8x16, v_uint16x8, v_uint32x4, v_int8x16, void);
|
||||
CV_DEF_REG_TRAITS(v, v_int8x16, schar, s8, v_uint8x16, v_int16x8, v_int32x4, v_int8x16, void);
|
||||
CV_DEF_REG_TRAITS(v, v_uint16x8, ushort, u16, v_uint16x8, v_uint32x4, v_uint64x2, v_int16x8, void);
|
||||
CV_DEF_REG_TRAITS(v, v_int16x8, short, s16, v_uint16x8, v_int32x4, v_int64x2, v_int16x8, void);
|
||||
CV_DEF_REG_TRAITS(v, v_uint32x4, unsigned, u32, v_uint32x4, v_uint64x2, void, v_int32x4, void);
|
||||
CV_DEF_REG_TRAITS(v, v_int32x4, int, s32, v_uint32x4, v_int64x2, void, v_int32x4, void);
|
||||
#if CV_SIMD128_64F
|
||||
CV_DEF_REG_TRAITS(v, v_float32x4, float, f32, v_float32x4, v_float64x2, void, v_int32x4, v_int32x4);
|
||||
#else
|
||||
CV_DEF_REG_TRAITS(v, v_float32x4, float, f32, v_float32x4, void, void, v_int32x4, v_int32x4);
|
||||
#endif
|
||||
CV_DEF_REG_TRAITS(v, v_uint64x2, uint64, u64, v_uint64x2, void, void, v_int64x2, void);
|
||||
CV_DEF_REG_TRAITS(v, v_int64x2, int64, s64, v_uint64x2, void, void, v_int64x2, void);
|
||||
#if CV_SIMD128_64F
|
||||
CV_DEF_REG_TRAITS(v, v_float64x2, double, f64, v_float64x2, void, void, v_int64x2, v_int32x4);
|
||||
#endif
|
||||
#endif
|
||||
|
||||
#if CV_SIMD256
|
||||
CV_DEF_REG_TRAITS(v256, v_uint8x32, uchar, u8, v_uint8x32, v_uint16x16, v_uint32x8, v_int8x32, void);
|
||||
CV_DEF_REG_TRAITS(v256, v_int8x32, schar, s8, v_uint8x32, v_int16x16, v_int32x8, v_int8x32, void);
|
||||
CV_DEF_REG_TRAITS(v256, v_uint16x16, ushort, u16, v_uint16x16, v_uint32x8, v_uint64x4, v_int16x16, void);
|
||||
CV_DEF_REG_TRAITS(v256, v_int16x16, short, s16, v_uint16x16, v_int32x8, v_int64x4, v_int16x16, void);
|
||||
CV_DEF_REG_TRAITS(v256, v_uint32x8, unsigned, u32, v_uint32x8, v_uint64x4, void, v_int32x8, void);
|
||||
CV_DEF_REG_TRAITS(v256, v_int32x8, int, s32, v_uint32x8, v_int64x4, void, v_int32x8, void);
|
||||
CV_DEF_REG_TRAITS(v256, v_float32x8, float, f32, v_float32x8, v_float64x4, void, v_int32x8, v_int32x8);
|
||||
CV_DEF_REG_TRAITS(v256, v_uint64x4, uint64, u64, v_uint64x4, void, void, v_int64x4, void);
|
||||
CV_DEF_REG_TRAITS(v256, v_int64x4, int64, s64, v_uint64x4, void, void, v_int64x4, void);
|
||||
CV_DEF_REG_TRAITS(v256, v_float64x4, double, f64, v_float64x4, void, void, v_int64x4, v_int32x8);
|
||||
#endif
|
||||
|
||||
#if CV_SIMD512 && (!defined(CV__SIMD_FORCE_WIDTH) || CV__SIMD_FORCE_WIDTH == 512)
|
||||
#define CV__SIMD_NAMESPACE simd512
|
||||
namespace CV__SIMD_NAMESPACE {
|
||||
#define CV_SIMD 1
|
||||
#define CV_SIMD_64F CV_SIMD512_64F
|
||||
#define CV_SIMD_WIDTH 64
|
||||
// TODO typedef v_uint8 / v_int32 / etc types here
|
||||
} // namespace
|
||||
using namespace CV__SIMD_NAMESPACE;
|
||||
#elif CV_SIMD256 && (!defined(CV__SIMD_FORCE_WIDTH) || CV__SIMD_FORCE_WIDTH == 256)
|
||||
#define CV__SIMD_NAMESPACE simd256
|
||||
namespace CV__SIMD_NAMESPACE {
|
||||
#define CV_SIMD 1
|
||||
#define CV_SIMD_64F CV_SIMD256_64F
|
||||
#define CV_SIMD_FP16 CV_SIMD256_FP16
|
||||
#define CV_SIMD_WIDTH 32
|
||||
typedef v_uint8x32 v_uint8;
|
||||
typedef v_int8x32 v_int8;
|
||||
typedef v_uint16x16 v_uint16;
|
||||
typedef v_int16x16 v_int16;
|
||||
typedef v_uint32x8 v_uint32;
|
||||
typedef v_int32x8 v_int32;
|
||||
typedef v_uint64x4 v_uint64;
|
||||
typedef v_int64x4 v_int64;
|
||||
typedef v_float32x8 v_float32;
|
||||
#if CV_SIMD256_64F
|
||||
typedef v_float64x4 v_float64;
|
||||
#endif
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN_ALL_TYPES(v256)
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN(double, v_float64, f64, v256, load)
|
||||
inline void vx_cleanup() { v256_cleanup(); }
|
||||
} // namespace
|
||||
using namespace CV__SIMD_NAMESPACE;
|
||||
#elif (CV_SIMD128 || CV_SIMD128_CPP) && (!defined(CV__SIMD_FORCE_WIDTH) || CV__SIMD_FORCE_WIDTH == 128)
|
||||
#define CV__SIMD_NAMESPACE simd128
|
||||
namespace CV__SIMD_NAMESPACE {
|
||||
#define CV_SIMD CV_SIMD128
|
||||
#define CV_SIMD_64F CV_SIMD128_64F
|
||||
#define CV_SIMD_WIDTH 16
|
||||
typedef v_uint8x16 v_uint8;
|
||||
typedef v_int8x16 v_int8;
|
||||
typedef v_uint16x8 v_uint16;
|
||||
typedef v_int16x8 v_int16;
|
||||
typedef v_uint32x4 v_uint32;
|
||||
typedef v_int32x4 v_int32;
|
||||
typedef v_uint64x2 v_uint64;
|
||||
typedef v_int64x2 v_int64;
|
||||
typedef v_float32x4 v_float32;
|
||||
#if CV_SIMD128_64F
|
||||
typedef v_float64x2 v_float64;
|
||||
#endif
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN_ALL_TYPES(v)
|
||||
#if CV_SIMD128_64F
|
||||
CV_INTRIN_DEFINE_WIDE_INTRIN(double, v_float64, f64, v, load)
|
||||
#endif
|
||||
inline void vx_cleanup() { v_cleanup(); }
|
||||
} // namespace
|
||||
using namespace CV__SIMD_NAMESPACE;
|
||||
#endif
|
||||
|
||||
inline unsigned int trailingZeros32(unsigned int value) {
|
||||
#if defined(_MSC_VER)
|
||||
#if (_MSC_VER < 1700) || defined(_M_ARM)
|
||||
unsigned long index = 0;
|
||||
_BitScanForward(&index, value);
|
||||
return (unsigned int)index;
|
||||
#elif defined(__clang__)
|
||||
// clang-cl doesn't export _tzcnt_u32 for non BMI systems
|
||||
return value ? __builtin_ctz(value) : 32;
|
||||
#else
|
||||
return _tzcnt_u32(value);
|
||||
#endif
|
||||
#elif defined(__GNUC__) || defined(__GNUG__)
|
||||
return __builtin_ctz(value);
|
||||
#elif defined(__ICC) || defined(__INTEL_COMPILER)
|
||||
return _bit_scan_forward(value);
|
||||
#elif defined(__clang__)
|
||||
return llvm.cttz.i32(value, true);
|
||||
#else
|
||||
static const int MultiplyDeBruijnBitPosition[32] = {
|
||||
0, 1, 28, 2, 29, 14, 24, 3, 30, 22, 20, 15, 25, 17, 4, 8,
|
||||
31, 27, 13, 23, 21, 19, 16, 7, 26, 12, 18, 6, 11, 5, 10, 9 };
|
||||
return MultiplyDeBruijnBitPosition[((uint32_t)((value & -value) * 0x077CB531U)) >> 27];
|
||||
#endif
|
||||
}
|
||||
|
||||
#ifndef CV_DOXYGEN
|
||||
CV_CPU_OPTIMIZATION_HAL_NAMESPACE_END
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD_64F
|
||||
#define CV_SIMD_64F 0
|
||||
#endif
|
||||
|
||||
#ifndef CV_SIMD_FP16
|
||||
#define CV_SIMD_FP16 0 //!< Defined to 1 on native support of operations with float16x8_t / float16x16_t (SIMD256) types
|
||||
#endif
|
||||
|
||||
|
||||
#ifndef CV_SIMD
|
||||
#define CV_SIMD 0
|
||||
#endif
|
||||
|
||||
} // cv::
|
||||
|
||||
//! @endcond
|
||||
|
||||
#endif
|
||||
File diff suppressed because it is too large
Load Diff
File diff suppressed because it is too large
Load Diff
|
|
@ -0,0 +1,158 @@
|
|||
// This file is part of OpenCV project.
|
||||
// It is subject to the license terms in the LICENSE file found in the top-level directory
|
||||
// of this distribution and at http://opencv.org/license.html
|
||||
|
||||
#ifndef CV__SIMD_FORWARD
|
||||
#error "Need to pre-define forward width"
|
||||
#endif
|
||||
|
||||
namespace cv
|
||||
{
|
||||
|
||||
//! @cond IGNORED
|
||||
|
||||
CV_CPU_OPTIMIZATION_HAL_NAMESPACE_BEGIN
|
||||
|
||||
/** Types **/
|
||||
#if CV__SIMD_FORWARD == 512
|
||||
// [todo] 512
|
||||
#error "AVX512 Not implemented yet"
|
||||
#elif CV__SIMD_FORWARD == 256
|
||||
// 256
|
||||
#define __CV_VX(fun) v256_##fun
|
||||
#define __CV_V_UINT8 v_uint8x32
|
||||
#define __CV_V_INT8 v_int8x32
|
||||
#define __CV_V_UINT16 v_uint16x16
|
||||
#define __CV_V_INT16 v_int16x16
|
||||
#define __CV_V_UINT32 v_uint32x8
|
||||
#define __CV_V_INT32 v_int32x8
|
||||
#define __CV_V_UINT64 v_uint64x4
|
||||
#define __CV_V_INT64 v_int64x4
|
||||
#define __CV_V_FLOAT32 v_float32x8
|
||||
#define __CV_V_FLOAT64 v_float64x4
|
||||
struct v_uint8x32;
|
||||
struct v_int8x32;
|
||||
struct v_uint16x16;
|
||||
struct v_int16x16;
|
||||
struct v_uint32x8;
|
||||
struct v_int32x8;
|
||||
struct v_uint64x4;
|
||||
struct v_int64x4;
|
||||
struct v_float32x8;
|
||||
struct v_float64x4;
|
||||
#else
|
||||
// 128
|
||||
#define __CV_VX(fun) v_##fun
|
||||
#define __CV_V_UINT8 v_uint8x16
|
||||
#define __CV_V_INT8 v_int8x16
|
||||
#define __CV_V_UINT16 v_uint16x8
|
||||
#define __CV_V_INT16 v_int16x8
|
||||
#define __CV_V_UINT32 v_uint32x4
|
||||
#define __CV_V_INT32 v_int32x4
|
||||
#define __CV_V_UINT64 v_uint64x2
|
||||
#define __CV_V_INT64 v_int64x2
|
||||
#define __CV_V_FLOAT32 v_float32x4
|
||||
#define __CV_V_FLOAT64 v_float64x2
|
||||
struct v_uint8x16;
|
||||
struct v_int8x16;
|
||||
struct v_uint16x8;
|
||||
struct v_int16x8;
|
||||
struct v_uint32x4;
|
||||
struct v_int32x4;
|
||||
struct v_uint64x2;
|
||||
struct v_int64x2;
|
||||
struct v_float32x4;
|
||||
struct v_float64x2;
|
||||
#endif
|
||||
|
||||
/** Value reordering **/
|
||||
|
||||
// Expansion
|
||||
void v_expand(const __CV_V_UINT8&, __CV_V_UINT16&, __CV_V_UINT16&);
|
||||
void v_expand(const __CV_V_INT8&, __CV_V_INT16&, __CV_V_INT16&);
|
||||
void v_expand(const __CV_V_UINT16&, __CV_V_UINT32&, __CV_V_UINT32&);
|
||||
void v_expand(const __CV_V_INT16&, __CV_V_INT32&, __CV_V_INT32&);
|
||||
void v_expand(const __CV_V_UINT32&, __CV_V_UINT64&, __CV_V_UINT64&);
|
||||
void v_expand(const __CV_V_INT32&, __CV_V_INT64&, __CV_V_INT64&);
|
||||
// Low Expansion
|
||||
__CV_V_UINT16 v_expand_low(const __CV_V_UINT8&);
|
||||
__CV_V_INT16 v_expand_low(const __CV_V_INT8&);
|
||||
__CV_V_UINT32 v_expand_low(const __CV_V_UINT16&);
|
||||
__CV_V_INT32 v_expand_low(const __CV_V_INT16&);
|
||||
__CV_V_UINT64 v_expand_low(const __CV_V_UINT32&);
|
||||
__CV_V_INT64 v_expand_low(const __CV_V_INT32&);
|
||||
// High Expansion
|
||||
__CV_V_UINT16 v_expand_high(const __CV_V_UINT8&);
|
||||
__CV_V_INT16 v_expand_high(const __CV_V_INT8&);
|
||||
__CV_V_UINT32 v_expand_high(const __CV_V_UINT16&);
|
||||
__CV_V_INT32 v_expand_high(const __CV_V_INT16&);
|
||||
__CV_V_UINT64 v_expand_high(const __CV_V_UINT32&);
|
||||
__CV_V_INT64 v_expand_high(const __CV_V_INT32&);
|
||||
// Load & Low Expansion
|
||||
__CV_V_UINT16 __CV_VX(load_expand)(const uchar*);
|
||||
__CV_V_INT16 __CV_VX(load_expand)(const schar*);
|
||||
__CV_V_UINT32 __CV_VX(load_expand)(const ushort*);
|
||||
__CV_V_INT32 __CV_VX(load_expand)(const short*);
|
||||
__CV_V_UINT64 __CV_VX(load_expand)(const uint*);
|
||||
__CV_V_INT64 __CV_VX(load_expand)(const int*);
|
||||
// Load lower 8-bit and expand into 32-bit
|
||||
__CV_V_UINT32 __CV_VX(load_expand_q)(const uchar*);
|
||||
__CV_V_INT32 __CV_VX(load_expand_q)(const schar*);
|
||||
|
||||
// Saturating Pack
|
||||
__CV_V_UINT8 v_pack(const __CV_V_UINT16&, const __CV_V_UINT16&);
|
||||
__CV_V_INT8 v_pack(const __CV_V_INT16&, const __CV_V_INT16&);
|
||||
__CV_V_UINT16 v_pack(const __CV_V_UINT32&, const __CV_V_UINT32&);
|
||||
__CV_V_INT16 v_pack(const __CV_V_INT32&, const __CV_V_INT32&);
|
||||
// Non-saturating Pack
|
||||
__CV_V_UINT32 v_pack(const __CV_V_UINT64&, const __CV_V_UINT64&);
|
||||
__CV_V_INT32 v_pack(const __CV_V_INT64&, const __CV_V_INT64&);
|
||||
// Pack signed integers with unsigned saturation
|
||||
__CV_V_UINT8 v_pack_u(const __CV_V_INT16&, const __CV_V_INT16&);
|
||||
__CV_V_UINT16 v_pack_u(const __CV_V_INT32&, const __CV_V_INT32&);
|
||||
|
||||
/** Arithmetic, bitwise and comparison operations **/
|
||||
|
||||
// Non-saturating multiply
|
||||
#if CV_VSX
|
||||
template<typename Tvec>
|
||||
Tvec v_mul_wrap(const Tvec& a, const Tvec& b);
|
||||
#else
|
||||
__CV_V_UINT8 v_mul_wrap(const __CV_V_UINT8&, const __CV_V_UINT8&);
|
||||
__CV_V_INT8 v_mul_wrap(const __CV_V_INT8&, const __CV_V_INT8&);
|
||||
__CV_V_UINT16 v_mul_wrap(const __CV_V_UINT16&, const __CV_V_UINT16&);
|
||||
__CV_V_INT16 v_mul_wrap(const __CV_V_INT16&, const __CV_V_INT16&);
|
||||
#endif
|
||||
|
||||
// Multiply and expand
|
||||
#if CV_VSX
|
||||
template<typename Tvec, typename Twvec>
|
||||
void v_mul_expand(const Tvec& a, const Tvec& b, Twvec& c, Twvec& d);
|
||||
#else
|
||||
void v_mul_expand(const __CV_V_UINT8&, const __CV_V_UINT8&, __CV_V_UINT16&, __CV_V_UINT16&);
|
||||
void v_mul_expand(const __CV_V_INT8&, const __CV_V_INT8&, __CV_V_INT16&, __CV_V_INT16&);
|
||||
void v_mul_expand(const __CV_V_UINT16&, const __CV_V_UINT16&, __CV_V_UINT32&, __CV_V_UINT32&);
|
||||
void v_mul_expand(const __CV_V_INT16&, const __CV_V_INT16&, __CV_V_INT32&, __CV_V_INT32&);
|
||||
void v_mul_expand(const __CV_V_UINT32&, const __CV_V_UINT32&, __CV_V_UINT64&, __CV_V_UINT64&);
|
||||
void v_mul_expand(const __CV_V_INT32&, const __CV_V_INT32&, __CV_V_INT64&, __CV_V_INT64&);
|
||||
#endif
|
||||
|
||||
/** Cleanup **/
|
||||
#undef CV__SIMD_FORWARD
|
||||
#undef __CV_VX
|
||||
#undef __CV_V_UINT8
|
||||
#undef __CV_V_INT8
|
||||
#undef __CV_V_UINT16
|
||||
#undef __CV_V_INT16
|
||||
#undef __CV_V_UINT32
|
||||
#undef __CV_V_INT32
|
||||
#undef __CV_V_UINT64
|
||||
#undef __CV_V_INT64
|
||||
#undef __CV_V_FLOAT32
|
||||
#undef __CV_V_FLOAT64
|
||||
|
||||
CV_CPU_OPTIMIZATION_HAL_NAMESPACE_END
|
||||
|
||||
//! @endcond
|
||||
|
||||
} // cv::
|
||||
Some files were not shown because too many files have changed in this diff Show More
Loading…
Reference in New Issue