Сравнение изображений (PSNR и SSIM)

[Smorodov 579]

На сайте с документацией, работу с GPU рассматривают на примере сравнения изображений (есть исходники для CPU и GPU):

http://docs.opencv.org/2.4.2/doc/tutorials/gpu/gpu-basics-similarity/gpu-basics-similarity.html

Код оттуда (на всякий случай):

#include <sstream> // String to number conversion

#include <opencv2/core/core.hpp> // Basic OpenCV structures
#include <opencv2/imgproc/imgproc.hpp>// Image processing methods for the CPU
#include <opencv2/highgui/highgui.hpp>// Read images
#include <opencv2/gpu/gpu.hpp> // GPU structures and methods

using namespace std;
using namespace cv;

double getPSNR(const Mat& I1, const Mat& I2); // CPU versions
Scalar getMSSIM( const Mat& I1, const Mat& I2);

double getPSNR_GPU(const Mat& I1, const Mat& I2); // Basic GPU versions
Scalar getMSSIM_GPU( const Mat& I1, const Mat& I2);

struct BufferPSNR // Optimized GPU versions
{ // Data allocations are very expensive on GPU. Use a buffer to solve: allocate once reuse later.
gpu::GpuMat gI1, gI2, gs, t1,t2;

gpu::GpuMat buf;
};
double getPSNR_GPU_optimized(const Mat& I1, const Mat& I2, BufferPSNR& ;

struct BufferMSSIM // Optimized GPU versions
{ // Data allocations are very expensive on GPU. Use a buffer to solve: allocate once reuse later.
gpu::GpuMat gI1, gI2, gs, t1,t2;

gpu::GpuMat I1_2, I2_2, I1_I2;
vector<gpu::GpuMat> vI1, vI2;

gpu::GpuMat mu1, mu2;
gpu::GpuMat mu1_2, mu2_2, mu1_mu2;

gpu::GpuMat sigma1_2, sigma2_2, sigma12;
gpu::GpuMat t3;

gpu::GpuMat ssim_map;

gpu::GpuMat buf;
};
Scalar getMSSIM_GPU_optimized( const Mat& i1, const Mat& i2, BufferMSSIM& ;

void help()
{
cout
<< "\n--------------------------------------------------------------------------" << endl
<< "This program shows how to port your CPU code to GPU or write that from scratch." << endl
<< "You can see the performance improvement for the similarity check methods (PSNR and SSIM)." << endl
<< "Usage:" << endl
<< "./gpu-basics-similarity referenceImage comparedImage numberOfTimesToRunTest(like 10)." << endl
<< "--------------------------------------------------------------------------" << endl
<< endl;
}

int main(int argc, char *argv[])
{
help();
Mat I1 = imread(argv[1]); // Read the two images
Mat I2 = imread(argv[2]);

if (!I1.data || !I2.data) // Check for success
{
cout << "Couldn't read the image";
return 0;
}

BufferPSNR bufferPSNR;
BufferMSSIM bufferMSSIM;

int TIMES;
stringstream sstr(argv[3]);
sstr >> TIMES;
double time, result;

//------------------------------- PSNR CPU ----------------------------------------------------
time = (double)getTickCount();

for (int i = 0; i < TIMES; ++i)
result = getPSNR(I1,I2);

time = 1000*((double)getTickCount() - time)/getTickFrequency();
time /= TIMES;

cout << "Time of PSNR CPU (averaged for " << TIMES << " runs): " << time << " milliseconds."
<< " With result of: " << result << endl;

//------------------------------- PSNR GPU ----------------------------------------------------
time = (double)getTickCount();

for (int i = 0; i < TIMES; ++i)
result = getPSNR_GPU(I1,I2);

time = 1000*((double)getTickCount() - time)/getTickFrequency();
time /= TIMES;

cout << "Time of PSNR GPU (averaged for " << TIMES << " runs): " << time << " milliseconds."
<< " With result of: " << result << endl;

//------------------------------- PSNR GPU Optimized--------------------------------------------
time = (double)getTickCount(); // Initial call
result = getPSNR_GPU_optimized(I1, I2, bufferPSNR);
time = 1000*((double)getTickCount() - time)/getTickFrequency();
cout << "Initial call GPU optimized: " << time <<" milliseconds."
<< " With result of: " << result << endl;

time = (double)getTickCount();
for (int i = 0; i < TIMES; ++i)
result = getPSNR_GPU_optimized(I1, I2, bufferPSNR);

time = 1000*((double)getTickCount() - time)/getTickFrequency();
time /= TIMES;

cout << "Time of PSNR GPU OPTIMIZED ( / " << TIMES << " runs): " << time
<< " milliseconds." << " With result of: " << result << endl << endl;


//------------------------------- SSIM CPU -----------------------------------------------------
Scalar x;
time = (double)getTickCount();

for (int i = 0; i < TIMES; ++i)
x = getMSSIM(I1,I2);

time = 1000*((double)getTickCount() - time)/getTickFrequency();
time /= TIMES;

cout << "Time of MSSIM CPU (averaged for " << TIMES << " runs): " << time << " milliseconds."
<< " With result of B" << x.val[0] << " G" << x.val[1] << " R" << x.val[2] << endl;

//------------------------------- SSIM GPU -----------------------------------------------------
time = (double)getTickCount();

for (int i = 0; i < TIMES; ++i)
x = getMSSIM_GPU(I1,I2);

time = 1000*((double)getTickCount() - time)/getTickFrequency();
time /= TIMES;

cout << "Time of MSSIM GPU (averaged for " << TIMES << " runs): " << time << " milliseconds."
<< " With result of B" << x.val[0] << " G" << x.val[1] << " R" << x.val[2] << endl;

//------------------------------- SSIM GPU Optimized--------------------------------------------
time = (double)getTickCount();
x = getMSSIM_GPU_optimized(I1,I2, bufferMSSIM);
time = 1000*((double)getTickCount() - time)/getTickFrequency();
cout << "Time of MSSIM GPU Initial Call " << time << " milliseconds."
<< " With result of B" << x.val[0] << " G" << x.val[1] << " R" << x.val[2] << endl;

time = (double)getTickCount();

for (int i = 0; i < TIMES; ++i)
x = getMSSIM_GPU_optimized(I1,I2, bufferMSSIM);

time = 1000*((double)getTickCount() - time)/getTickFrequency();
time /= TIMES;

cout << "Time of MSSIM GPU OPTIMIZED ( / " << TIMES << " runs): " << time << " milliseconds."
<< " With result of B" << x.val[0] << " G" << x.val[1] << " R" << x.val[2] << endl << endl;
return 0;
}


double getPSNR(const Mat& I1, const Mat& I2)
{
Mat s1;
absdiff(I1, I2, s1); // |I1 - I2|
s1.convertTo(s1, CV_32F); // cannot make a square on 8 bits
s1 = s1.mul(s1); // |I1 - I2|^2

Scalar s = sum(s1); // sum elements per channel

double sse = s.val[0] + s.val[1] + s.val[2]; // sum channels

if( sse <= 1e-10) // for small values return zero
return 0;
else
{
double mse =sse /(double)(I1.channels() * I1.total());
double psnr = 10.0*log10((255*255)/mse);
return psnr;
}
}



double getPSNR_GPU_optimized(const Mat& I1, const Mat& I2, BufferPSNR&
{
b.gI1.upload(I1);
b.gI2.upload(I2);

b.gI1.convertTo(b.t1, CV_32F);
b.gI2.convertTo(b.t2, CV_32F);

gpu::absdiff(b.t1.reshape(1), b.t2.reshape(1), b.gs);
gpu::multiply(b.gs, b.gs, b.gs);

double sse = gpu::sum(b.gs, b.buf)[0];

if( sse <= 1e-10) // for small values return zero
return 0;
else
{
double mse = sse /(double)(I1.channels() * I1.total());
double psnr = 10.0*log10((255*255)/mse);
return psnr;
}
}

double getPSNR_GPU(const Mat& I1, const Mat& I2)
{
gpu::GpuMat gI1, gI2, gs, t1,t2;

gI1.upload(I1);
gI2.upload(I2);

gI1.convertTo(t1, CV_32F);
gI2.convertTo(t2, CV_32F);

gpu::absdiff(t1.reshape(1), t2.reshape(1), gs);
gpu::multiply(gs, gs, gs);

Scalar s = gpu::sum(gs);
double sse = s.val[0] + s.val[1] + s.val[2];

if( sse <= 1e-10) // for small values return zero
return 0;
else
{
double mse =sse /(double)(gI1.channels() * I1.total());
double psnr = 10.0*log10((255*255)/mse);
return psnr;
}
}

Scalar getMSSIM( const Mat& i1, const Mat& i2)
{
const double C1 = 6.5025, C2 = 58.5225;
/***************************** INITS **********************************/
int d = CV_32F;

Mat I1, I2;
i1.convertTo(I1, d); // cannot calculate on one byte large values
i2.convertTo(I2, d);

Mat I2_2 = I2.mul(I2); // I2^2
Mat I1_2 = I1.mul(I1); // I1^2
Mat I1_I2 = I1.mul(I2); // I1 * I2

/*************************** END INITS **********************************/

Mat mu1, mu2; // PRELIMINARY COMPUTING
GaussianBlur(I1, mu1, Size(11, 11), 1.5);
GaussianBlur(I2, mu2, Size(11, 11), 1.5);

Mat mu1_2 = mu1.mul(mu1);
Mat mu2_2 = mu2.mul(mu2);
Mat mu1_mu2 = mu1.mul(mu2);

Mat sigma1_2, sigma2_2, sigma12;

GaussianBlur(I1_2, sigma1_2, Size(11, 11), 1.5);
sigma1_2 -= mu1_2;

GaussianBlur(I2_2, sigma2_2, Size(11, 11), 1.5);
sigma2_2 -= mu2_2;

GaussianBlur(I1_I2, sigma12, Size(11, 11), 1.5);
sigma12 -= mu1_mu2;

///////////////////////////////// FORMULA ////////////////////////////////
Mat t1, t2, t3;

t1 = 2 * mu1_mu2 + C1;
t2 = 2 * sigma12 + C2;
t3 = t1.mul(t2); // t3 = ((2*mu1_mu2 + C1).*(2*sigma12 + C2))

t1 = mu1_2 + mu2_2 + C1;
t2 = sigma1_2 + sigma2_2 + C2;
t1 = t1.mul(t2); // t1 =((mu1_2 + mu2_2 + C1).*(sigma1_2 + sigma2_2 + C2))

Mat ssim_map;
divide(t3, t1, ssim_map); // ssim_map = t3./t1;

Scalar mssim = mean( ssim_map ); // mssim = average of ssim map
return mssim;
}

Scalar getMSSIM_GPU( const Mat& i1, const Mat& i2)
{
const float C1 = 6.5025f, C2 = 58.5225f;
/***************************** INITS **********************************/
gpu::GpuMat gI1, gI2, gs1, t1,t2;

gI1.upload(i1);
gI2.upload(i2);

gI1.convertTo(t1, CV_MAKE_TYPE(CV_32F, gI1.channels()));
gI2.convertTo(t2, CV_MAKE_TYPE(CV_32F, gI2.channels()));

vector<gpu::GpuMat> vI1, vI2;
gpu::split(t1, vI1);
gpu::split(t2, vI2);
Scalar mssim;

for( int i = 0; i < gI1.channels(); ++i )
{
gpu::GpuMat I2_2, I1_2, I1_I2;

gpu::multiply(vI2[i], vI2[i], I2_2); // I2^2
gpu::multiply(vI1[i], vI1[i], I1_2); // I1^2
gpu::multiply(vI1[i], vI2[i], I1_I2); // I1 * I2

/*************************** END INITS **********************************/
gpu::GpuMat mu1, mu2; // PRELIMINARY COMPUTING
gpu::GaussianBlur(vI1[i], mu1, Size(11, 11), 1.5);
gpu::GaussianBlur(vI2[i], mu2, Size(11, 11), 1.5);

gpu::GpuMat mu1_2, mu2_2, mu1_mu2;
gpu::multiply(mu1, mu1, mu1_2);
gpu::multiply(mu2, mu2, mu2_2);
gpu::multiply(mu1, mu2, mu1_mu2);

gpu::GpuMat sigma1_2, sigma2_2, sigma12;

gpu::GaussianBlur(I1_2, sigma1_2, Size(11, 11), 1.5);
sigma1_2 -= mu1_2;

gpu::GaussianBlur(I2_2, sigma2_2, Size(11, 11), 1.5);
sigma2_2 -= mu2_2;

gpu::GaussianBlur(I1_I2, sigma12, Size(11, 11), 1.5);
sigma12 -= mu1_mu2;

///////////////////////////////// FORMULA ////////////////////////////////
gpu::GpuMat t1, t2, t3;

t1 = 2 * mu1_mu2 + C1;
t2 = 2 * sigma12 + C2;
gpu::multiply(t1, t2, t3); // t3 = ((2*mu1_mu2 + C1).*(2*sigma12 + C2))

t1 = mu1_2 + mu2_2 + C1;
t2 = sigma1_2 + sigma2_2 + C2;
gpu::multiply(t1, t2, t1); // t1 =((mu1_2 + mu2_2 + C1).*(sigma1_2 + sigma2_2 + C2))

gpu::GpuMat ssim_map;
gpu::divide(t3, t1, ssim_map); // ssim_map = t3./t1;

Scalar s = gpu::sum(ssim_map);
mssim.val[i] = s.val[0] / (ssim_map.rows * ssim_map.cols);

}
return mssim;
}

Scalar getMSSIM_GPU_optimized( const Mat& i1, const Mat& i2, BufferMSSIM&
{
int cn = i1.channels();

const float C1 = 6.5025f, C2 = 58.5225f;
/***************************** INITS **********************************/

b.gI1.upload(i1);
b.gI2.upload(i2);

gpu::Stream stream;

stream.enqueueConvert(b.gI1, b.t1, CV_32F);
stream.enqueueConvert(b.gI2, b.t2, CV_32F);

gpu::split(b.t1, b.vI1, stream);
gpu::split(b.t2, b.vI2, stream);
Scalar mssim;

for( int i = 0; i < b.gI1.channels(); ++i )
{
gpu::multiply(b.vI2[i], b.vI2[i], b.I2_2, stream); // I2^2
gpu::multiply(b.vI1[i], b.vI1[i], b.I1_2, stream); // I1^2
gpu::multiply(b.vI1[i], b.vI2[i], b.I1_I2, stream); // I1 * I2

gpu::GaussianBlur(b.vI1[i], b.mu1, Size(11, 11), 1.5, 0, BORDER_DEFAULT, -1, stream);
gpu::GaussianBlur(b.vI2[i], b.mu2, Size(11, 11), 1.5, 0, BORDER_DEFAULT, -1, stream);

gpu::multiply(b.mu1, b.mu1, b.mu1_2, stream);
gpu::multiply(b.mu2, b.mu2, b.mu2_2, stream);
gpu::multiply(b.mu1, b.mu2, b.mu1_mu2, stream);

gpu::GaussianBlur(b.I1_2, b.sigma1_2, Size(11, 11), 1.5, 0, BORDER_DEFAULT, -1, stream);
gpu::subtract(b.sigma1_2, b.mu1_2, b.sigma1_2, stream);
//b.sigma1_2 -= b.mu1_2; - This would result in an extra data transfer operation

gpu::GaussianBlur(b.I2_2, b.sigma2_2, Size(11, 11), 1.5, 0, BORDER_DEFAULT, -1, stream);
gpu::subtract(b.sigma2_2, b.mu2_2, b.sigma2_2, stream);
//b.sigma2_2 -= b.mu2_2;

gpu::GaussianBlur(b.I1_I2, b.sigma12, Size(11, 11), 1.5, 0, BORDER_DEFAULT, -1, stream);
gpu::subtract(b.sigma12, b.mu1_mu2, b.sigma12, stream);
//b.sigma12 -= b.mu1_mu2;

//here too it would be an extra data transfer due to call of operator*(Scalar, Mat)
gpu::multiply(b.mu1_mu2, 2, b.t1, stream); //b.t1 = 2 * b.mu1_mu2 + C1;
gpu::add(b.t1, C1, b.t1, stream);
gpu::multiply(b.sigma12, 2, b.t2, stream); //b.t2 = 2 * b.sigma12 + C2;
gpu::add(b.t2, C2, b.t2, stream);

gpu::multiply(b.t1, b.t2, b.t3, stream); // t3 = ((2*mu1_mu2 + C1).*(2*sigma12 + C2))

gpu::add(b.mu1_2, b.mu2_2, b.t1, stream);
gpu::add(b.t1, C1, b.t1, stream);

gpu::add(b.sigma1_2, b.sigma2_2, b.t2, stream);
gpu::add(b.t2, C2, b.t2, stream);


gpu::multiply(b.t1, b.t2, b.t1, stream); // t1 =((mu1_2 + mu2_2 + C1).*(sigma1_2 + sigma2_2 + C2))
gpu::divide(b.t3, b.t1, b.ssim_map, stream); // ssim_map = t3./t1;

stream.waitForCompletion();

Scalar s = gpu::sum(b.ssim_map, b.buf);
mssim.val[i] = s.val[0] / (b.ssim_map.rows * b.ssim_map.cols);

}
return mssim;
}

#include <iostream>                   // Console I/O

[Pavia00 32]

А почему так медленно?

[Smorodov 579]

В статье,MSSIM в GPU-шном варианте 20мс (слабый GPU) или 50мс. в CPU-шном. на кадр (3 канала).

PSNR - 4 мс. в CPU-ном.

В MSSIM операций много, одних блюров 5 шт.