opencv/modules/gpu/perf_cpu/perf_matop.cpp
2012-05-22 10:29:58 +00:00

163 lines
4.3 KiB
C++

#include "perf_cpu_precomp.hpp"
#ifdef HAVE_CUDA
//////////////////////////////////////////////////////////////////////
// Merge
GPU_PERF_TEST(Merge, cv::gpu::DeviceInfo, cv::Size, perf::MatType)
{
cv::Size size = GET_PARAM(1);
int type = GET_PARAM(2);
const int num_channels = 4;
std::vector<cv::Mat> src(num_channels);
for (int i = 0; i < num_channels; ++i)
src[i] = cv::Mat(size, type, cv::Scalar::all(i));
cv::Mat dst;
TEST_CYCLE()
{
cv::merge(src, dst);
}
}
INSTANTIATE_TEST_CASE_P(MatOp, Merge, testing::Combine(
ALL_DEVICES,
GPU_TYPICAL_MAT_SIZES,
testing::Values(CV_8UC1, CV_16UC1, CV_32FC1)));
//////////////////////////////////////////////////////////////////////
// Split
GPU_PERF_TEST(Split, cv::gpu::DeviceInfo, cv::Size, perf::MatType)
{
cv::Size size = GET_PARAM(1);
int type = GET_PARAM(2);
const int num_channels = 4;
cv::Mat src(size, CV_MAKETYPE(type, num_channels), cv::Scalar(1, 2, 3, 4));
std::vector<cv::Mat> dst(num_channels);
for (int i = 0; i < num_channels; ++i)
dst[i] = cv::Mat(size, type);
TEST_CYCLE()
{
cv::split(src, dst);
}
}
INSTANTIATE_TEST_CASE_P(MatOp, Split, testing::Combine(
ALL_DEVICES,
GPU_TYPICAL_MAT_SIZES,
testing::Values(CV_8UC1, CV_16UC1, CV_32FC1)));
//////////////////////////////////////////////////////////////////////
// SetTo
GPU_PERF_TEST(SetTo, cv::gpu::DeviceInfo, cv::Size, perf::MatType)
{
cv::Size size = GET_PARAM(1);
int type = GET_PARAM(2);
cv::Mat src(size, type);
cv::Scalar val(1, 2, 3, 4);
TEST_CYCLE()
{
src.setTo(val);
}
}
INSTANTIATE_TEST_CASE_P(MatOp, SetTo, testing::Combine(
ALL_DEVICES,
GPU_TYPICAL_MAT_SIZES,
testing::Values(CV_8UC1, CV_8UC3, CV_8UC4, CV_16UC1, CV_16UC3, CV_16UC4, CV_32FC1, CV_32FC3, CV_32FC4)));
//////////////////////////////////////////////////////////////////////
// SetToMasked
GPU_PERF_TEST(SetToMasked, cv::gpu::DeviceInfo, cv::Size, perf::MatType)
{
cv::Size size = GET_PARAM(1);
int type = GET_PARAM(2);
cv::Mat src(size, type);
cv::Mat mask(size, CV_8UC1);
declare.in(src, WARMUP_RNG);
fill(mask, 0, 2);
cv::Scalar val(1, 2, 3, 4);
TEST_CYCLE()
{
src.setTo(val, mask);
}
}
INSTANTIATE_TEST_CASE_P(MatOp, SetToMasked, testing::Combine(
ALL_DEVICES,
GPU_TYPICAL_MAT_SIZES,
testing::Values(CV_8UC1, CV_8UC3, CV_8UC4, CV_16UC1, CV_16UC3, CV_16UC4, CV_32FC1, CV_32FC3, CV_32FC4)));
//////////////////////////////////////////////////////////////////////
// CopyToMasked
GPU_PERF_TEST(CopyToMasked, cv::gpu::DeviceInfo, cv::Size, perf::MatType)
{
cv::Size size = GET_PARAM(1);
int type = GET_PARAM(2);
cv::Mat src(size, type);
cv::Mat mask(size, CV_8UC1);
declare.in(src, WARMUP_RNG);
fill(mask, 0, 2);
cv::Mat dst;
TEST_CYCLE()
{
src.copyTo(dst, mask);
}
}
INSTANTIATE_TEST_CASE_P(MatOp, CopyToMasked, testing::Combine(
ALL_DEVICES,
GPU_TYPICAL_MAT_SIZES,
testing::Values(CV_8UC1, CV_8UC3, CV_8UC4, CV_16UC1, CV_16UC3, CV_16UC4, CV_32FC1, CV_32FC3, CV_32FC4)));
//////////////////////////////////////////////////////////////////////
// ConvertTo
GPU_PERF_TEST(ConvertTo, cv::gpu::DeviceInfo, cv::Size, perf::MatType, perf::MatType)
{
cv::Size size = GET_PARAM(1);
int type1 = GET_PARAM(2);
int type2 = GET_PARAM(3);
cv::Mat src(size, type1);
declare.in(src, WARMUP_RNG);
cv::Mat dst;
TEST_CYCLE()
{
src.convertTo(dst, type2, 0.5, 1.0);
}
}
INSTANTIATE_TEST_CASE_P(MatOp, ConvertTo, testing::Combine(
ALL_DEVICES,
GPU_TYPICAL_MAT_SIZES,
testing::Values(CV_8UC1, CV_16UC1, CV_32FC1),
testing::Values(CV_8UC1, CV_16UC1, CV_32FC1)));
#endif