void run(int)
{
RNG& rng = theRNG();
for( int i = 0; i < 10; i++ )
{
int m = rng.uniform(2, 11);
int n = rng.uniform(2, 11);
int depth = rng.uniform(0, 2) + CV_32F;
Mat src8u(m, n, depth), src(m, n, depth), dst(m, n, CV_MAKETYPE(depth, 2));
Mat z = Mat::zeros(m, n, depth), dstz;
randu(src8u, Scalar::all(0), Scalar::all(10));
src8u.convertTo(src, src.type());
dst = Scalar::all(123);
Mat mv[] = {src, z}, srcz;
merge(mv, 2, srcz);
dft(srcz, dstz);
dft(src, dst, DFT_COMPLEX_OUTPUT);
if (cvtest::norm(dst, dstz, NORM_INF) > 1e-3)
{
cout << "actual:\n" << dst << endl << endl;
cout << "reference:\n" << dstz << endl << endl;
CV_Error(CV_StsError, "");
}
}
}
Moments ocl_moments(oclMat& src, bool binary) //for image
{
CV_Assert(src.oclchannels() == 1);
if(src.type() == CV_64FC1 && !Context::getContext()->supportsFeature(FEATURE_CL_DOUBLE))
{
CV_Error(CV_StsUnsupportedFormat, "Moments - double is not supported by your GPU!");
}
if(binary)
{
oclMat mask;
if(src.type() != CV_8UC1)
{
src.convertTo(mask, CV_8UC1);
}
oclMat src8u(src.size(), CV_8UC1);
src8u.setTo(Scalar(255), mask);
src = src8u;
}
const int TILE_SIZE = 256;
CvMoments mom;
memset(&mom, 0, sizeof(mom));
cv::Size size = src.size();
int blockx, blocky;
blockx = (size.width + TILE_SIZE - 1)/TILE_SIZE;
blocky = (size.height + TILE_SIZE - 1)/TILE_SIZE;
oclMat dst_m;
int tile_height = TILE_SIZE;
size_t localThreads[3] = {1, tile_height, 1};
size_t globalThreads[3] = {blockx, size.height, 1};
if(Context::getContext()->supportsFeature(FEATURE_CL_DOUBLE))
{
dst_m.create(blocky * 10, blockx, CV_64FC1);
}else
{
dst_m.create(blocky * 10, blockx, CV_32FC1);
}
int src_step = (int)(src.step/src.elemSize());
int dstm_step = (int)(dst_m.step/dst_m.elemSize());
std::vector<std::pair<size_t , const void *> > args,args_sum;
args.push_back( std::make_pair( sizeof(cl_mem) , (void *)&src.data ));
args.push_back( std::make_pair( sizeof(cl_int) , (void *)&src.rows ));
args.push_back( std::make_pair( sizeof(cl_int) , (void *)&src.cols ));
args.push_back( std::make_pair( sizeof(cl_int) , (void *)&src_step ));
args.push_back( std::make_pair( sizeof(cl_mem) , (void *)&dst_m.data ));
args.push_back( std::make_pair( sizeof(cl_int) , (void *)&dst_m.cols ));
args.push_back( std::make_pair( sizeof(cl_int) , (void *)&dstm_step ));
int binary_;
if(binary)
binary_ = 1;
else
binary_ = 0;
args.push_back( std::make_pair( sizeof(cl_int) , (void *)&binary_));
char builOption[128];
if(binary || src.type() == CV_8UC1)
{
snprintf(builOption, 128, "-D CV_8UC1");
}else if(src.type() == CV_16UC1)
{
snprintf(builOption, 128, "-D CV_16UC1");
}else if(src.type() == CV_16SC1)
{
snprintf(builOption, 128, "-D CV_16SC1");
}else if(src.type() == CV_32FC1)
{
snprintf(builOption, 128, "-D CV_32FC1");
}else if(src.type() == CV_64FC1)
{
snprintf(builOption, 128, "-D CV_64FC1");
}else
{
CV_Error( CV_StsUnsupportedFormat, "" );
}
openCLExecuteKernel(Context::getContext(), &moments, "CvMoments", globalThreads, localThreads, args, -1, -1, builOption);
Mat tmp(dst_m);
tmp.convertTo(tmp, CV_64FC1);
double tmp_m[10] = {0};
for(int j = 0; j < tmp.rows; j += 10)
{
for(int i = 0; i < tmp.cols; i++)
{
tmp_m[0] += tmp.at<double>(j, i);
tmp_m[1] += tmp.at<double>(j + 1, i);
tmp_m[2] += tmp.at<double>(j + 2, i);
tmp_m[3] += tmp.at<double>(j + 3, i);
tmp_m[4] += tmp.at<double>(j + 4, i);
tmp_m[5] += tmp.at<double>(j + 5, i);
tmp_m[6] += tmp.at<double>(j + 6, i);
tmp_m[7] += tmp.at<double>(j + 7, i);
tmp_m[8] += tmp.at<double>(j + 8, i);
tmp_m[9] += tmp.at<double>(j + 9, i);
}
}
mom.m00 = tmp_m[0];
mom.m10 = tmp_m[1];
mom.m01 = tmp_m[2];
mom.m20 = tmp_m[3];
mom.m11 = tmp_m[4];
mom.m02 = tmp_m[5];
mom.m30 = tmp_m[6];
mom.m21 = tmp_m[7];
mom.m12 = tmp_m[8];
mom.m03 = tmp_m[9];
icvCompleteMomentState( &mom );
return mom;
}
