/*
Learn GMMs parameters.
*/
static void learnGMMs( const Mat& img, const Mat& mask, const Mat& pixelModel, GMM& bgGMM, GMM& fgGMM )
{
bgGMM.initParameter();
fgGMM.initParameter();
//Point p;
for( int ci = 0; ci < GMM::modelNumInGMM; ci++ )
{
for( int x = 0; x < img.rows; x++ )
{
for( int y = 0; y < img.cols; y++ )
{
if( pixelModel.at<int>(x,y) == ci )
{
if( mask.at<uchar>(x,y) == BG || mask.at<uchar>(x,y) == PBG )
bgGMM.calcParameter( ci, img.at<Vec3b>(x,y) );
else
fgGMM.calcParameter( ci, img.at<Vec3b>(x,y) );
}
}
}
}
bgGMM.endLearning();
fgGMM.endLearning();
}
/*
Learn GMMs parameters.
*/
void learnGMMs( const Mat& img, const Mat& mask, const Mat& compIdxs, GMM& bgdGMM, GMM& fgdGMM )
{
bgdGMM.initLearning();
fgdGMM.initLearning();
Point p;
for( int ci = 0; ci < GMM::componentsCount; ci++ )
{
for( p.y = 0; p.y < img.rows; p.y++ )
{
for( p.x = 0; p.x < img.cols; p.x++ )
{
if( compIdxs.at<int>(p) == ci )
{
if( mask.at<uchar>(p) == GC_BGD || mask.at<uchar>(p) == GC_PR_BGD )
bgdGMM.addSample( ci, img.at<Vec3b>(p) );
else
fgdGMM.addSample( ci, img.at<Vec3b>(p) );
}
}
}
}
bgdGMM.endLearning();
fgdGMM.endLearning();
}
static void initGMMs( const Mat& img, const Mat& mask, GMM& bgGMM, GMM& fgGMM )
{
const int kMeansNum = 10;
const int kMeansType = KMEANS_PP_CENTERS;
Mat bgClustered;
Mat fgClustered;
vector<Vec3f> bgSamples;
vector<Vec3f> fgSamples;
//for each pixel in the image, put those BG in bgsample, FG in fgsample
for( int x = 0; x < img.rows; x++ )
{
for( int y = 0; y < img.cols; y++ )
{
if( mask.at<uchar>(x,y) == BG || mask.at<uchar>(x,y) == PBG )
bgSamples.push_back( (Vec3f)img.at<Vec3b>(x,y) );
else // GC_FGD | GC_PR_FGD
fgSamples.push_back( (Vec3f)img.at<Vec3b>(x,y) );
}
}
Mat bgdSamplesMat( (int)bgSamples.size(), 3, CV_32FC1, &bgSamples[0][0] );
kmeans( bgdSamplesMat, GMM::modelNumInGMM, bgClustered, TermCriteria( CV_TERMCRIT_ITER, kMeansNum, 0.0), 0, kMeansType );
Mat fgdSamplesMat( (int)fgSamples.size(), 3, CV_32FC1, &fgSamples[0][0] );
kmeans( fgdSamplesMat, GMM::modelNumInGMM, fgClustered, TermCriteria( CV_TERMCRIT_ITER, kMeansNum, 0.0), 0, kMeansType );
bgGMM.initParameter();
for( int i = 0; i < (int)bgSamples.size(); i++ )
bgGMM.calcParameter( bgClustered.at<int>(i,0), bgSamples[i] );
bgGMM.endLearning();
fgGMM.initParameter();
for( int i = 0; i < (int)fgSamples.size(); i++ )
fgGMM.calcParameter( fgClustered.at<int>(i,0), fgSamples[i] );
fgGMM.endLearning();
}
/*
Initialize GMM background and foreground models using kmeans algorithm.
*/
void initGMMs( const Mat& img, const Mat& mask, GMM& bgdGMM, GMM& fgdGMM )
{
const int kMeansItCount = 10;
const int kMeansType = KMEANS_PP_CENTERS;
Mat bgdLabels, fgdLabels;
vector<Vec3f> bgdSamples, fgdSamples;
Point p;
for( p.y = 0; p.y < img.rows; p.y++ )
{
for( p.x = 0; p.x < img.cols; p.x++ )
{
if( mask.at<uchar>(p) == GC_BGD || mask.at<uchar>(p) == GC_PR_BGD )
bgdSamples.push_back( (Vec3f)img.at<Vec3b>(p) );
else // GC_FGD | GC_PR_FGD
fgdSamples.push_back( (Vec3f)img.at<Vec3b>(p) );
}
}
CV_Assert( !bgdSamples.empty() && !fgdSamples.empty() );
Mat _bgdSamples( (int)bgdSamples.size(), 3, CV_32FC1, &bgdSamples[0][0] );
kmeans( _bgdSamples, GMM::componentsCount, bgdLabels,
TermCriteria( CV_TERMCRIT_ITER, kMeansItCount, 0.0), 0, kMeansType, 0 );
Mat _fgdSamples( (int)fgdSamples.size(), 3, CV_32FC1, &fgdSamples[0][0] );
kmeans( _fgdSamples, GMM::componentsCount, fgdLabels,
TermCriteria( CV_TERMCRIT_ITER, kMeansItCount, 0.0), 0, kMeansType, 0 );
bgdGMM.initLearning();
for( int i = 0; i < (int)bgdSamples.size(); i++ )
bgdGMM.addSample( bgdLabels.at<int>(i,0), bgdSamples[i] );
bgdGMM.endLearning();
fgdGMM.initLearning();
for( int i = 0; i < (int)fgdSamples.size(); i++ )
fgdGMM.addSample( fgdLabels.at<int>(i,0), fgdSamples[i] );
fgdGMM.endLearning();
}
