format files to ANSI C style with coolformat

change the download channels to oclchannles()
fix bugs of arithm functions
perf fix of bilateral
bug fix of split test case
add build_warps functions

modules/ocl/perf/interpolation.hpp

@@ -42,7 +42,7 @@
 #ifndef __OPENCV_TEST_INTERPOLATION_HPP__
 #define __OPENCV_TEST_INTERPOLATION_HPP__
 
-template <typename T> T readVal(const cv::Mat& src, int y, int x, int c, int border_type, cv::Scalar borderVal = cv::Scalar())
+template <typename T> T readVal(const cv::Mat &src, int y, int x, int c, int border_type, cv::Scalar borderVal = cv::Scalar())
 {
     if (border_type == cv::BORDER_CONSTANT)
         return (y >= 0 && y < src.rows && x >= 0 && x < src.cols) ? src.at<T>(y, x * src.channels() + c) : cv::saturate_cast<T>(borderVal.val[c]);
@@ -52,7 +52,7 @@ template <typename T> T readVal(const cv::Mat& src, int y, int x, int c, int bor
 
 template <typename T> struct NearestInterpolator
 {
-    static T getValue(const cv::Mat& src, float y, float x, int c, int border_type, cv::Scalar borderVal = cv::Scalar())
+    static T getValue(const cv::Mat &src, float y, float x, int c, int border_type, cv::Scalar borderVal = cv::Scalar())
     {
         return readVal<T>(src, cvFloor(y), cvFloor(x), c, border_type, borderVal);
     }
@@ -60,7 +60,7 @@ template <typename T> struct NearestInterpolator
 
 template <typename T> struct LinearInterpolator
 {
-    static T getValue(const cv::Mat& src, float y, float x, int c, int border_type, cv::Scalar borderVal = cv::Scalar())
+    static T getValue(const cv::Mat &src, float y, float x, int c, int border_type, cv::Scalar borderVal = cv::Scalar())
     {
         x -= 0.5f;
         y -= 0.5f;
@@ -85,7 +85,7 @@ template <typename T> struct CubicInterpolator
 {
     static float getValue(float p[4], float x)
     {
-        return p[1] + 0.5 * x * (p[2] - p[0] + x*(2.0*p[0] - 5.0*p[1] + 4.0*p[2] - p[3] + x*(3.0*(p[1] - p[2]) + p[3] - p[0])));
+        return p[1] + 0.5 * x * (p[2] - p[0] + x * (2.0 * p[0] - 5.0 * p[1] + 4.0 * p[2] - p[3] + x * (3.0 * (p[1] - p[2]) + p[3] - p[0])));
     }
 
     static float getValue(float p[4][4], float x, float y)
@@ -100,7 +100,7 @@ template <typename T> struct CubicInterpolator
         return getValue(arr, y);
     }
 
-    static T getValue(const cv::Mat& src, float y, float x, int c, int border_type, cv::Scalar borderVal = cv::Scalar())
+    static T getValue(const cv::Mat &src, float y, float x, int c, int border_type, cv::Scalar borderVal = cv::Scalar())
     {
         int ix = cvRound(x);
         int iy = cvRound(y);

modules/ocl/perf/main.cpp

@@ -50,46 +50,46 @@ using namespace cvtest;
 using namespace testing;
 
 void print_info()
-{    
+{
     printf("\n");
 #if defined _WIN32
 #   if defined _WIN64
-        puts("OS: Windows 64");
+    puts("OS: Windows 64");
 #   else
-        puts("OS: Windows 32");
+    puts("OS: Windows 32");
 #   endif
 #elif defined linux
 #   if defined _LP64
-        puts("OS: Linux 64");
+    puts("OS: Linux 64");
 #   else
-        puts("OS: Linux 32");
+    puts("OS: Linux 32");
 #   endif
 #elif defined __APPLE__
 #   if defined _LP64
-        puts("OS: Apple 64");
+    puts("OS: Apple 64");
 #   else
-        puts("OS: Apple 32");
+    puts("OS: Apple 32");
 #   endif
 #endif
 
 }
 
-int main(int argc, char** argv)
+int main(int argc, char **argv)
 {
-	std::vector<cv::ocl::Info> oclinfo;
+    std::vector<cv::ocl::Info> oclinfo;
     TS::ptr()->init("ocl");
     InitGoogleTest(&argc, argv);
 
     print_info();
-	int devnums = getDevice(oclinfo);
-	if(devnums<1)
-	{
-		std::cout << "no device found\n";
-		return -1;
-	}
-	//if you want to use undefault device, set it here
-	//setDevice(oclinfo[0]);
-	setBinpath(CLBINPATH);
+    int devnums = getDevice(oclinfo);
+    if(devnums < 1)
+    {
+        std::cout << "no device found\n";
+        return -1;
+    }
+    //if you want to use undefault device, set it here
+    //setDevice(oclinfo[0]);
+    setBinpath(CLBINPATH);
     return RUN_ALL_TESTS();
 }
 

modules/ocl/perf/perf_blend.cpp

@@ -55,66 +55,66 @@ using namespace std;
 
 PARAM_TEST_CASE(Blend, MatType, int)
 {
-	int type;
-	int channels;
-	std::vector<cv::ocl::Info> oclinfo;
-	
-	virtual void SetUp()
-	{
-	
-		type = GET_PARAM(0);
-		channels = GET_PARAM(1);
-		//int devnums = getDevice(oclinfo);
-		//CV_Assert(devnums > 0);
-		//cv::ocl::setBinpath(CLBINPATH);
-	}
+    int type;
+    int channels;
+    std::vector<cv::ocl::Info> oclinfo;
+
+    virtual void SetUp()
+    {
+
+        type = GET_PARAM(0);
+        channels = GET_PARAM(1);
+        //int devnums = getDevice(oclinfo);
+        //CV_Assert(devnums > 0);
+        //cv::ocl::setBinpath(CLBINPATH);
+    }
 };
 
 TEST_P(Blend, Performance)
 {
-	cv::Size size(MWIDTH, MHEIGHT);
-	cv::Mat img1_host = randomMat(size, CV_MAKETYPE(type, channels), 0, type == CV_8U ? 255.0 : 1.0);
-	cv::Mat img2_host = randomMat(size, CV_MAKETYPE(type, channels), 0, type == CV_8U ? 255.0 : 1.0);
-	cv::Mat weights1 = randomMat(size, CV_32F, 0, 1);
-	cv::Mat weights2 = randomMat(size, CV_32F, 0, 1);
-	cv::ocl::oclMat gimg1(size, CV_MAKETYPE(type, channels)), gimg2(size, CV_MAKETYPE(type, channels)), gweights1(size, CV_32F), gweights2(size, CV_32F);
-	cv::ocl::oclMat gdst(size, CV_MAKETYPE(type, channels));
-	
-	
-	double totalgputick_all = 0;
-	double totalgputick_kernel = 0;
-	double t1 = 0;
-	double t2 = 0;
-	
-	for (int j = 0; j < LOOP_TIMES + 1; j ++) //LOOP_TIMES=100
-	{
-		t1 = (double)cvGetTickCount();
-		cv::ocl::oclMat gimg1 = cv::ocl::oclMat(img1_host);
-		cv::ocl::oclMat gimg2 = cv::ocl::oclMat(img2_host);
-		cv::ocl::oclMat gweights1 = cv::ocl::oclMat(weights1);
-		cv::ocl::oclMat gweights2 = cv::ocl::oclMat(weights1);
-		
-		t2 = (double)cvGetTickCount();
-		cv::ocl::blendLinear(gimg1, gimg2, gweights1, gweights2, gdst);
-		t2 = (double)cvGetTickCount() - t2;
-		
-		cv::Mat m;
-		gdst.download(m);
-		t1 = (double)cvGetTickCount() - t1;
-		
-		if (j == 0)
-		{
-			continue;
-		}
-		
-		totalgputick_all = t1 + totalgputick_all;
-		totalgputick_kernel = t2 + totalgputick_kernel;
-	};
-	
-	cout << "average gpu total  runtime is  " << totalgputick_all / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
-	
-	cout << "average gpu runtime without data transfering  is  " << totalgputick_kernel / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
-	
+    cv::Size size(MWIDTH, MHEIGHT);
+    cv::Mat img1_host = randomMat(size, CV_MAKETYPE(type, channels), 0, type == CV_8U ? 255.0 : 1.0);
+    cv::Mat img2_host = randomMat(size, CV_MAKETYPE(type, channels), 0, type == CV_8U ? 255.0 : 1.0);
+    cv::Mat weights1 = randomMat(size, CV_32F, 0, 1);
+    cv::Mat weights2 = randomMat(size, CV_32F, 0, 1);
+    cv::ocl::oclMat gimg1(size, CV_MAKETYPE(type, channels)), gimg2(size, CV_MAKETYPE(type, channels)), gweights1(size, CV_32F), gweights2(size, CV_32F);
+    cv::ocl::oclMat gdst(size, CV_MAKETYPE(type, channels));
+
+
+    double totalgputick_all = 0;
+    double totalgputick_kernel = 0;
+    double t1 = 0;
+    double t2 = 0;
+
+    for (int j = 0; j < LOOP_TIMES + 1; j ++) //LOOP_TIMES=100
+    {
+        t1 = (double)cvGetTickCount();
+        cv::ocl::oclMat gimg1 = cv::ocl::oclMat(img1_host);
+        cv::ocl::oclMat gimg2 = cv::ocl::oclMat(img2_host);
+        cv::ocl::oclMat gweights1 = cv::ocl::oclMat(weights1);
+        cv::ocl::oclMat gweights2 = cv::ocl::oclMat(weights1);
+
+        t2 = (double)cvGetTickCount();
+        cv::ocl::blendLinear(gimg1, gimg2, gweights1, gweights2, gdst);
+        t2 = (double)cvGetTickCount() - t2;
+
+        cv::Mat m;
+        gdst.download(m);
+        t1 = (double)cvGetTickCount() - t1;
+
+        if (j == 0)
+        {
+            continue;
+        }
+
+        totalgputick_all = t1 + totalgputick_all;
+        totalgputick_kernel = t2 + totalgputick_kernel;
+    };
+
+    cout << "average gpu total  runtime is  " << totalgputick_all / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
+
+    cout << "average gpu runtime without data transfering  is  " << totalgputick_kernel / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
+
 }
 
 INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Blend, Combine(

modules/ocl/perf/perf_canny.cpp

@@ -85,70 +85,70 @@ IMPLEMENT_PARAM_CLASS(L2gradient, bool);
 
 PARAM_TEST_CASE(Canny1, AppertureSize, L2gradient)
 {
-	int apperture_size;
-	bool useL2gradient;
-	//std::vector<cv::ocl::Info> oclinfo;
-
-	virtual void SetUp()
-	{
-		apperture_size = GET_PARAM(0);
-		useL2gradient = GET_PARAM(1);
-		
-		//int devnums = getDevice(oclinfo);
-		//CV_Assert(devnums > 0);
-	}
+    int apperture_size;
+    bool useL2gradient;
+    //std::vector<cv::ocl::Info> oclinfo;
+
+    virtual void SetUp()
+    {
+        apperture_size = GET_PARAM(0);
+        useL2gradient = GET_PARAM(1);
+
+        //int devnums = getDevice(oclinfo);
+        //CV_Assert(devnums > 0);
+    }
 };
 
 TEST_P(Canny1, Performance)
 {
-	cv::Mat img = readImage(FILTER_IMAGE,cv::IMREAD_GRAYSCALE);
-	ASSERT_FALSE(img.empty());
-
-	double low_thresh = 100.0;
-	double high_thresh = 150.0;
-
-	cv::Mat edges_gold;
-	cv::ocl::oclMat edges;
-
-    double totalgputick=0;
-	double totalgputick_kernel=0;
-	
-	double t1=0;
-	double t2=0;
-	for(int j = 0; j < LOOP_TIMES+1; j ++)
-	{
-
-		t1 = (double)cvGetTickCount();//gpu start1		
-			
-		cv::ocl::oclMat ocl_img = cv::ocl::oclMat(img);//upload
-			
-		t2=(double)cvGetTickCount();//kernel
-		cv::ocl::Canny(ocl_img, edges, low_thresh, high_thresh, apperture_size, useL2gradient);
-		t2 = (double)cvGetTickCount() - t2;//kernel
-			
-		cv::Mat cpu_dst;
-		edges.download (cpu_dst);//download
-			
-		t1 = (double)cvGetTickCount() - t1;//gpu end1
-
-		if(j == 0)
-			continue;
-
-		totalgputick=t1+totalgputick;
-
-		totalgputick_kernel=t2+totalgputick_kernel;	
+    cv::Mat img = readImage(FILTER_IMAGE, cv::IMREAD_GRAYSCALE);
+    ASSERT_FALSE(img.empty());
 
-	}
+    double low_thresh = 100.0;
+    double high_thresh = 150.0;
+
+    cv::Mat edges_gold;
+    cv::ocl::oclMat edges;
+
+    double totalgputick = 0;
+    double totalgputick_kernel = 0;
+
+    double t1 = 0;
+    double t2 = 0;
+    for(int j = 0; j < LOOP_TIMES + 1; j ++)
+    {
+
+        t1 = (double)cvGetTickCount();//gpu start1
+
+        cv::ocl::oclMat ocl_img = cv::ocl::oclMat(img);//upload
+
+        t2 = (double)cvGetTickCount(); //kernel
+        cv::ocl::Canny(ocl_img, edges, low_thresh, high_thresh, apperture_size, useL2gradient);
+        t2 = (double)cvGetTickCount() - t2;//kernel
+
+        cv::Mat cpu_dst;
+        edges.download (cpu_dst);//download
+
+        t1 = (double)cvGetTickCount() - t1;//gpu end1
+
+        if(j == 0)
+            continue;
+
+        totalgputick = t1 + totalgputick;
+
+        totalgputick_kernel = t2 + totalgputick_kernel;
+
+    }
 
-	cout << "average gpu runtime is  " << totalgputick/((double)cvGetTickFrequency()* LOOP_TIMES *1000.) << "ms" << endl;
-	cout << "average gpu runtime without data transfer is  " << totalgputick_kernel/((double)cvGetTickFrequency()* LOOP_TIMES *1000.) << "ms" << endl;
+    cout << "average gpu runtime is  " << totalgputick / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
+    cout << "average gpu runtime without data transfer is  " << totalgputick_kernel / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
 
 
 }
 
 INSTANTIATE_TEST_CASE_P(GPU_ImgProc, Canny1, testing::Combine(
-						testing::Values(AppertureSize(3), AppertureSize(5)),
-						testing::Values(L2gradient(false), L2gradient(true))));
+                            testing::Values(AppertureSize(3), AppertureSize(5)),
+                            testing::Values(L2gradient(false), L2gradient(true))));
 
 
 

modules/ocl/perf/perf_columnsum.cpp

@@ -16,7 +16,7 @@
 //
 // @Authors
 //	   Fangfang Bai fangfang@multicorewareinc.com
-//    
+//
 //
 // Redistribution and use in source and binary forms, with or without modification,
 // are permitted provided that the following conditions are met:
@@ -63,53 +63,53 @@ using namespace std;
 
 PARAM_TEST_CASE(ColumnSum)
 {
-	cv::Mat src;
-	//std::vector<cv::ocl::Info> oclinfo;
-
-	virtual void SetUp()
-	{
-		//int devnums = getDevice(oclinfo);
-		//CV_Assert(devnums > 0);
-	}
+    cv::Mat src;
+    //std::vector<cv::ocl::Info> oclinfo;
+
+    virtual void SetUp()
+    {
+        //int devnums = getDevice(oclinfo);
+        //CV_Assert(devnums > 0);
+    }
 };
 
 TEST_F(ColumnSum, Performance)
 {
-	cv::Size size(MWIDTH,MHEIGHT);
+    cv::Size size(MWIDTH, MHEIGHT);
     cv::Mat src = randomMat(size, CV_32FC1);
     cv::ocl::oclMat d_dst;
 
-	double totalgputick=0;
-	double totalgputick_kernel=0;
-	double t1=0;
-	double t2=0;
+    double totalgputick = 0;
+    double totalgputick_kernel = 0;
+    double t1 = 0;
+    double t2 = 0;
 
-	for(int j = 0; j < LOOP_TIMES+1; j ++)
-	{
+    for(int j = 0; j < LOOP_TIMES + 1; j ++)
+    {
 
-		t1 = (double)cvGetTickCount();//gpu start1
+        t1 = (double)cvGetTickCount();//gpu start1
 
-        cv::ocl::oclMat d_src(src);		
+        cv::ocl::oclMat d_src(src);
 
-		t2=(double)cvGetTickCount();//kernel
-		cv::ocl::columnSum(d_src,d_dst);
-		t2 = (double)cvGetTickCount() - t2;//kernel
+        t2 = (double)cvGetTickCount(); //kernel
+        cv::ocl::columnSum(d_src, d_dst);
+        t2 = (double)cvGetTickCount() - t2;//kernel
 
-		cv::Mat cpu_dst;
-		d_dst.download (cpu_dst);//download
+        cv::Mat cpu_dst;
+        d_dst.download (cpu_dst);//download
 
-		t1 = (double)cvGetTickCount() - t1;//gpu end1
+        t1 = (double)cvGetTickCount() - t1;//gpu end1
 
-		if(j == 0)
-			continue;
+        if(j == 0)
+            continue;
 
-		totalgputick=t1+totalgputick;
-		totalgputick_kernel=t2+totalgputick_kernel;	
+        totalgputick = t1 + totalgputick;
+        totalgputick_kernel = t2 + totalgputick_kernel;
 
-	}
+    }
 
-	cout << "average gpu runtime is  " << totalgputick/((double)cvGetTickFrequency()* LOOP_TIMES *1000.) << "ms" << endl;
-	cout << "average gpu runtime without data transfer is  " << totalgputick_kernel/((double)cvGetTickFrequency()* LOOP_TIMES *1000.) << "ms" << endl;
+    cout << "average gpu runtime is  " << totalgputick / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
+    cout << "average gpu runtime without data transfer is  " << totalgputick_kernel / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
 
 
 
@@ -117,4 +117,4 @@ TEST_F(ColumnSum, Performance)
 
 
 
-#endif 
\ No newline at end of file
+#endif
\ No newline at end of file

modules/ocl/perf/perf_fft.cpp

@@ -48,75 +48,75 @@ using namespace std;
 #ifdef HAVE_CLAMDFFT
 ////////////////////////////////////////////////////////////////////////////
 // Dft
-PARAM_TEST_CASE(Dft, cv::Size, bool) 
+PARAM_TEST_CASE(Dft, cv::Size, bool)
 {
-	cv::Size dft_size;
-	bool	 dft_rows;
-	vector<cv::ocl::Info> info;
-	virtual void SetUp()
-	{
-		dft_size = GET_PARAM(0);
-		dft_rows = GET_PARAM(1);
-		cv::ocl::getDevice(info);
-	}
+    cv::Size dft_size;
+    bool	 dft_rows;
+    vector<cv::ocl::Info> info;
+    virtual void SetUp()
+    {
+        dft_size = GET_PARAM(0);
+        dft_rows = GET_PARAM(1);
+        cv::ocl::getDevice(info);
+    }
 };
 
 TEST_P(Dft, C2C)
 {
-	cv::Mat a = randomMat(dft_size, CV_32FC2, 0.0, 10.0);
-	int flags = 0;
-	flags |= dft_rows ? cv::DFT_ROWS : 0;
+    cv::Mat a = randomMat(dft_size, CV_32FC2, 0.0, 10.0);
+    int flags = 0;
+    flags |= dft_rows ? cv::DFT_ROWS : 0;
 
-	cv::ocl::oclMat d_b;
+    cv::ocl::oclMat d_b;
 
-	double totalgputick=0;
-	double totalgputick_kernel=0;
-	double t1=0;
-	double t2=0;
+    double totalgputick = 0;
+    double totalgputick_kernel = 0;
+    double t1 = 0;
+    double t2 = 0;
 
-	for(int j = 0; j < LOOP_TIMES+1; j ++)
-	{
+    for(int j = 0; j < LOOP_TIMES + 1; j ++)
+    {
 
-		t1 = (double)cvGetTickCount();//gpu start1
+        t1 = (double)cvGetTickCount();//gpu start1
 
-		cv::ocl::oclMat ga=cv::ocl::oclMat(a);//upload
+        cv::ocl::oclMat ga = cv::ocl::oclMat(a); //upload
 
-		t2=(double)cvGetTickCount();//kernel
-		cv::ocl::dft(ga, d_b, a.size(), flags);
-		t2 = (double)cvGetTickCount() - t2;//kernel
+        t2 = (double)cvGetTickCount(); //kernel
+        cv::ocl::dft(ga, d_b, a.size(), flags);
+        t2 = (double)cvGetTickCount() - t2;//kernel
 
-		cv::Mat cpu_dst;
-		d_b.download (cpu_dst);//download
+        cv::Mat cpu_dst;
+        d_b.download (cpu_dst);//download
 
-		t1 = (double)cvGetTickCount() - t1;//gpu end1
+        t1 = (double)cvGetTickCount() - t1;//gpu end1
 
-		if(j == 0)
-			continue;
+        if(j == 0)
+            continue;
 
-		totalgputick=t1+totalgputick;	
-		totalgputick_kernel=t2+totalgputick_kernel;	
+        totalgputick = t1 + totalgputick;
+        totalgputick_kernel = t2 + totalgputick_kernel;
 
-	}
+    }
 
-	cout << "average gpu runtime is  " << totalgputick/((double)cvGetTickFrequency()* LOOP_TIMES *1000.) << "ms" << endl;
-	cout << "average gpu runtime without data transfer is  " << totalgputick_kernel/((double)cvGetTickFrequency()* LOOP_TIMES *1000.) << "ms" << endl;
+    cout << "average gpu runtime is  " << totalgputick / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
+    cout << "average gpu runtime without data transfer is  " << totalgputick_kernel / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
 }
 
 
 
 TEST_P(Dft, R2CthenC2R)
 {
-	cv::Mat a = randomMat(dft_size, CV_32FC1, 0.0, 10.0);
+    cv::Mat a = randomMat(dft_size, CV_32FC1, 0.0, 10.0);
 
-	int flags = 0;
-	//flags |= dft_rows ? cv::DFT_ROWS : 0; // not supported yet
+    int flags = 0;
+    //flags |= dft_rows ? cv::DFT_ROWS : 0; // not supported yet
 
-	cv::ocl::oclMat d_b, d_c;
+    cv::ocl::oclMat d_b, d_c;
 
-	cv::ocl::dft(cv::ocl::oclMat(a), d_b, a.size(), flags);
-	cv::ocl::dft(d_b, d_c, a.size(), flags + cv::DFT_INVERSE + cv::DFT_REAL_OUTPUT);
+    cv::ocl::dft(cv::ocl::oclMat(a), d_b, a.size(), flags);
+    cv::ocl::dft(d_b, d_c, a.size(), flags + cv::DFT_INVERSE + cv::DFT_REAL_OUTPUT);
 
-	EXPECT_MAT_NEAR(a, d_c, a.size().area() * 1e-4, "");
+    EXPECT_MAT_NEAR(a, d_c, a.size().area() * 1e-4, "");
 }
 
 //INSTANTIATE_TEST_CASE_P(ocl_DFT, Dft, testing::Combine(

modules/ocl/perf/perf_gemm.cpp

@@ -48,66 +48,66 @@ using namespace std;
 #ifdef HAVE_CLAMDBLAS
 ////////////////////////////////////////////////////////////////////////////
 // GEMM
-PARAM_TEST_CASE(Gemm, int, cv::Size, int) 
+PARAM_TEST_CASE(Gemm, int, cv::Size, int)
 {
-	int      type;
-	cv::Size mat_size;
-	int		 flags;
-	vector<cv::ocl::Info> info;
-	virtual void SetUp()
-	{
-		type     = GET_PARAM(0);
-		mat_size = GET_PARAM(1);
-		flags    = GET_PARAM(2);
-
-		cv::ocl::getDevice(info);
-	}
+    int      type;
+    cv::Size mat_size;
+    int		 flags;
+    vector<cv::ocl::Info> info;
+    virtual void SetUp()
+    {
+        type     = GET_PARAM(0);
+        mat_size = GET_PARAM(1);
+        flags    = GET_PARAM(2);
+
+        cv::ocl::getDevice(info);
+    }
 };
 
 TEST_P(Gemm, Performance)
 {
-	cv::Mat a = randomMat(mat_size, type, 0.0, 10.0);
-	cv::Mat b = randomMat(mat_size, type, 0.0, 10.0);
-	cv::Mat c = randomMat(mat_size, type, 0.0, 10.0);
-	cv::ocl::oclMat ocl_dst;	
+    cv::Mat a = randomMat(mat_size, type, 0.0, 10.0);
+    cv::Mat b = randomMat(mat_size, type, 0.0, 10.0);
+    cv::Mat c = randomMat(mat_size, type, 0.0, 10.0);
+    cv::ocl::oclMat ocl_dst;
 
-	double totalgputick=0;
-	double totalgputick_kernel=0;
-	double t1=0;
-	double t2=0;
+    double totalgputick = 0;
+    double totalgputick_kernel = 0;
+    double t1 = 0;
+    double t2 = 0;
 
-	for(int j = 0; j < LOOP_TIMES+1; j ++)
-	{
+    for(int j = 0; j < LOOP_TIMES + 1; j ++)
+    {
 
-		t1 = (double)cvGetTickCount();//gpu start1
+        t1 = (double)cvGetTickCount();//gpu start1
 
-		cv::ocl::oclMat ga = cv::ocl::oclMat(a);//upload
-		cv::ocl::oclMat gb = cv::ocl::oclMat(b);//upload
-		cv::ocl::oclMat gc = cv::ocl::oclMat(c);//upload
+        cv::ocl::oclMat ga = cv::ocl::oclMat(a);//upload
+        cv::ocl::oclMat gb = cv::ocl::oclMat(b);//upload
+        cv::ocl::oclMat gc = cv::ocl::oclMat(c);//upload
 
-		t2=(double)cvGetTickCount();//kernel
-		cv::ocl::gemm(ga, gb, 1.0,gc, 1.0, ocl_dst, flags);
-		t2 = (double)cvGetTickCount() - t2;//kernel
+        t2 = (double)cvGetTickCount(); //kernel
+        cv::ocl::gemm(ga, gb, 1.0, gc, 1.0, ocl_dst, flags);
+        t2 = (double)cvGetTickCount() - t2;//kernel
 
-		cv::Mat cpu_dst;
-		ocl_dst.download (cpu_dst);//download
+        cv::Mat cpu_dst;
+        ocl_dst.download (cpu_dst);//download
 
-		t1 = (double)cvGetTickCount() - t1;//gpu end
+        t1 = (double)cvGetTickCount() - t1;//gpu end
 
-		if(j == 0)
-			continue;
+        if(j == 0)
+            continue;
 
-		totalgputick=t1+totalgputick;	
-		totalgputick_kernel=t2+totalgputick_kernel;	
+        totalgputick = t1 + totalgputick;
+        totalgputick_kernel = t2 + totalgputick_kernel;
 
-	}
-	cout << "average gpu runtime is  " << totalgputick/((double)cvGetTickFrequency()* LOOP_TIMES *1000.) << "ms" << endl;
-    cout << "average gpu runtime without data transfer is  " << totalgputick_kernel/((double)cvGetTickFrequency()* LOOP_TIMES *1000.) << "ms" << endl;
+    }
+    cout << "average gpu runtime is  " << totalgputick / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
+    cout << "average gpu runtime without data transfer is  " << totalgputick_kernel / ((double)cvGetTickFrequency()* LOOP_TIMES * 1000.) << "ms" << endl;
 }
 
 
 INSTANTIATE_TEST_CASE_P(ocl_gemm, Gemm, testing::Combine(
-						testing::Values(CV_32FC1, CV_32FC2/* , CV_64FC1, CV_64FC2*/),
-						testing::Values(cv::Size(512, 512), cv::Size(1024, 1024)),
-						testing::Values(0, cv::GEMM_1_T, cv::GEMM_2_T, cv::GEMM_1_T + cv::GEMM_2_T)));
+                            testing::Values(CV_32FC1, CV_32FC2/* , CV_64FC1, CV_64FC2*/),
+                            testing::Values(cv::Size(512, 512), cv::Size(1024, 1024)),
+                            testing::Values(0, cv::GEMM_1_T, cv::GEMM_2_T, cv::GEMM_1_T + cv::GEMM_2_T)));
 #endif
\ No newline at end of file

modules/ocl/perf/perf_haar.cpp

@@ -53,118 +53,125 @@ using namespace testing;
 using namespace std;
 using namespace cv;
 
-struct getRect { Rect operator ()(const CvAvgComp& e) const { return e.rect; } };
+struct getRect
+{
+    Rect operator ()(const CvAvgComp &e) const
+    {
+        return e.rect;
+    }
+};
 
 PARAM_TEST_CASE(HaarTestBase, int, int)
 {
-	//std::vector<cv::ocl::Info> oclinfo;
-	cv::ocl::OclCascadeClassifier cascade, nestedCascade;
-	cv::CascadeClassifier cpucascade, cpunestedCascade;
-	//    Mat img;
-
-	double scale;
-	int index;
-
-	virtual void SetUp()
-	{
-		scale = 1.0;
-		index=0;
-		string cascadeName="../../../data/haarcascades/haarcascade_frontalface_alt.xml";
-
-		if( (!cascade.load( cascadeName )) || (!cpucascade.load(cascadeName)))
-		{
-			cout << "ERROR: Could not load classifier cascade" << endl;
-			cout << "Usage: facedetect [--cascade=<cascade_path>]\n"
-				"   [--scale[=<image scale>\n"
-				"   [filename|camera_index]\n" << endl ;
-			return;
-		}
-		//int devnums = getDevice(oclinfo);
-		//CV_Assert(devnums>0);
-		////if you want to use undefault device, set it here
-		////setDevice(oclinfo[0]);
-		//cv::ocl::setBinpath("E:\\");
-	}
+    //std::vector<cv::ocl::Info> oclinfo;
+    cv::ocl::OclCascadeClassifier cascade, nestedCascade;
+    cv::CascadeClassifier cpucascade, cpunestedCascade;
+    //    Mat img;
+
+    double scale;
+    int index;
+
+    virtual void SetUp()
+    {
+        scale = 1.0;
+        index = 0;
+        string cascadeName = "../../../data/haarcascades/haarcascade_frontalface_alt.xml";
+
+        if( (!cascade.load( cascadeName )) || (!cpucascade.load(cascadeName)))
+        {
+            cout << "ERROR: Could not load classifier cascade" << endl;
+            cout << "Usage: facedetect [--cascade=<cascade_path>]\n"
+                 "   [--scale[=<image scale>\n"
+                 "   [filename|camera_index]\n" << endl ;
+            return;
+        }
+        //int devnums = getDevice(oclinfo);
+        //CV_Assert(devnums>0);
+        ////if you want to use undefault device, set it here
+        ////setDevice(oclinfo[0]);
+        //cv::ocl::setBinpath("E:\\");
+    }
 };
 
 ////////////////////////////////faceDetect/////////////////////////////////////////////////
 
 struct Haar : HaarTestBase {};
 
-TEST_F(Haar, FaceDetect) 
-{    
-	string imgName = "../../../samples/c/lena.jpg";
-	Mat img = imread( imgName, 1 );
-
-	if(img.empty())
-	{ 
-		std::cout << "Couldn't read test" << index <<".jpg" << std::endl;
-		return ;
-	}
-
-	int i = 0;
-	double t = 0;
-	vector<Rect> faces, oclfaces;
-
-	const static Scalar colors[] =  { CV_RGB(0,0,255),
-		CV_RGB(0,128,255),
-		CV_RGB(0,255,255),
-		CV_RGB(0,255,0),
-		CV_RGB(255,128,0),
-		CV_RGB(255,255,0),
-		CV_RGB(255,0,0),
-		CV_RGB(255,0,255)} ;
-
-	Mat gray, smallImg(cvRound (img.rows/scale), cvRound(img.cols/scale), CV_8UC1 );
-	MemStorage storage(cvCreateMemStorage(0));
-	cvtColor( img, gray, CV_BGR2GRAY );
-	resize( gray, smallImg, smallImg.size(), 0, 0, INTER_LINEAR );
-	equalizeHist( smallImg, smallImg );
-
-	t = (double)cvGetTickCount();
-	for(int k= 0; k<LOOP_TIMES; k++)
-	{
-		cpucascade.detectMultiScale( smallImg, faces,  1.1,
-			3, 0
-			|CV_HAAR_SCALE_IMAGE
-			, Size(30,30), Size(0, 0) );
-	}
-	t = (double)cvGetTickCount() - t ;
-	printf( "cpudetection time = %g ms\n", t/(LOOP_TIMES*(double)cvGetTickFrequency()*1000.) );
-
-	cv::ocl::oclMat image;
-	CvSeq* _objects;
-	t = (double)cvGetTickCount();
-	for(int k= 0; k<LOOP_TIMES; k++)
-	{
-		image.upload(smallImg);
-		_objects = cascade.oclHaarDetectObjects( image, storage, 1.1,
-			3, 0
-			|CV_HAAR_SCALE_IMAGE
-			, Size(30,30), Size(0, 0) );
-	}
-	t = (double)cvGetTickCount() - t ;
-	printf( "ocldetection time = %g ms\n", t/(LOOP_TIMES*(double)cvGetTickFrequency()*1000.) );
-	vector<CvAvgComp> vecAvgComp;
-	Seq<CvAvgComp>(_objects).copyTo(vecAvgComp);
-	oclfaces.resize(vecAvgComp.size());
-	std::transform(vecAvgComp.begin(), vecAvgComp.end(), oclfaces.begin(), getRect());
-
-	//for( vector<Rect>::const_iterator r = faces.begin(); r != faces.end(); r++, i++ )
-	//{ 
-	//	Mat smallImgROI;
-	//	Point center;
-	//	Scalar color = colors[i%8];
-	//	int radius;
-	//	center.x = cvRound((r->x + r->width*0.5)*scale);
-	//	center.y = cvRound((r->y + r->height*0.5)*scale);
-	//	radius = cvRound((r->width + r->height)*0.25*scale);
-	//	circle( img, center, radius, color, 3, 8, 0 );
-	//}  
-	//namedWindow("result");
-	//imshow("result",img);
-	//waitKey(0);
-	//destroyAllWindows();
+TEST_F(Haar, FaceDetect)
+{
+    string imgName = "../../../samples/c/lena.jpg";
+    Mat img = imread( imgName, 1 );
+
+    if(img.empty())
+    {
+        std::cout << "Couldn't read test" << index << ".jpg" << std::endl;
+        return ;
+    }
+
+    int i = 0;
+    double t = 0;
+    vector<Rect> faces, oclfaces;
+
+    const static Scalar colors[] =  { CV_RGB(0, 0, 255),
+                                      CV_RGB(0, 128, 255),
+                                      CV_RGB(0, 255, 255),
+                                      CV_RGB(0, 255, 0),
+                                      CV_RGB(255, 128, 0),
+                                      CV_RGB(255, 255, 0),
+                                      CV_RGB(255, 0, 0),
+                                      CV_RGB(255, 0, 255)
+                                    } ;
+
+    Mat gray, smallImg(cvRound (img.rows / scale), cvRound(img.cols / scale), CV_8UC1 );
+    MemStorage storage(cvCreateMemStorage(0));
+    cvtColor( img, gray, CV_BGR2GRAY );
+    resize( gray, smallImg, smallImg.size(), 0, 0, INTER_LINEAR );
+    equalizeHist( smallImg, smallImg );
+
+    t = (double)cvGetTickCount();
+    for(int k = 0; k < LOOP_TIMES; k++)
+    {
+        cpucascade.detectMultiScale( smallImg, faces,  1.1,
+                                     3, 0
+                                     | CV_HAAR_SCALE_IMAGE
+                                     , Size(30, 30), Size(0, 0) );
+    }
+    t = (double)cvGetTickCount() - t ;
+    printf( "cpudetection time = %g ms\n", t / (LOOP_TIMES * (double)cvGetTickFrequency() * 1000.) );
+
+    cv::ocl::oclMat image;
+    CvSeq *_objects;
+    t = (double)cvGetTickCount();
+    for(int k = 0; k < LOOP_TIMES; k++)
+    {
+        image.upload(smallImg);
+        _objects = cascade.oclHaarDetectObjects( image, storage, 1.1,
+                   3, 0
+                   | CV_HAAR_SCALE_IMAGE
+                   , Size(30, 30), Size(0, 0) );
+    }
+    t = (double)cvGetTickCount() - t ;
+    printf( "ocldetection time = %g ms\n", t / (LOOP_TIMES * (double)cvGetTickFrequency() * 1000.) );
+    vector<CvAvgComp> vecAvgComp;
+    Seq<CvAvgComp>(_objects).copyTo(vecAvgComp);
+    oclfaces.resize(vecAvgComp.size());
+    std::transform(vecAvgComp.begin(), vecAvgComp.end(), oclfaces.begin(), getRect());
+
+    //for( vector<Rect>::const_iterator r = faces.begin(); r != faces.end(); r++, i++ )
+    //{
+    //	Mat smallImgROI;
+    //	Point center;
+    //	Scalar color = colors[i%8];
+    //	int radius;
+    //	center.x = cvRound((r->x + r->width*0.5)*scale);
+    //	center.y = cvRound((r->y + r->height*0.5)*scale);
+    //	radius = cvRound((r->width + r->height)*0.25*scale);
+    //	circle( img, center, radius, color, 3, 8, 0 );
+    //}
+    //namedWindow("result");
+    //imshow("result",img);
+    //waitKey(0);
+    //destroyAllWindows();
 
 }
 #endif // HAVE_OPENCL

modules/ocl/perf/precomp.cpp

@@ -42,4 +42,3 @@
 #include "precomp.hpp"
 
 
-	
\ No newline at end of file

modules/ocl/src/mcwutil.hpp

@@ -63,10 +63,10 @@ namespace cv
             DISABLE
         };
         void openCLExecuteKernel2(Context *clCxt , const char **source, string kernelName, size_t globalThreads[3],
-            size_t localThreads[3],  vector< pair<size_t, const void *> > &args, int channels, int depth, FLUSH_MODE finish_mode = DISABLE);
+                                  size_t localThreads[3],  vector< pair<size_t, const void *> > &args, int channels, int depth, FLUSH_MODE finish_mode = DISABLE);
         void openCLExecuteKernel2(Context *clCxt , const char **source, string kernelName, size_t globalThreads[3],
-            size_t localThreads[3],  vector< pair<size_t, const void *> > &args, int channels,
-            int depth, char *build_options, FLUSH_MODE finish_mode = DISABLE);
+                                  size_t localThreads[3],  vector< pair<size_t, const void *> > &args, int channels,
+                                  int depth, char *build_options, FLUSH_MODE finish_mode = DISABLE);
     }//namespace ocl
 
 }//namespace cv