vector<int> Recoloring::MatchGaussians(CvEM& source_model, CvEM& target_model) {
int num_g = source_model.get_nclusters();
Mat sMu(source_model.get_means());
Mat tMu(target_model.get_means());
const CvMat** target_covs = target_model.get_covs();
const CvMat** source_covs = source_model.get_covs();
double best_dist = std::numeric_limits<double>::max();
vector<int> best_res(num_g);
vector<int> prmt(num_g);
for(int itr = 0; itr < 10; itr++) {
for(int i=0;i<num_g;i++) prmt[i] = i; //make a permutation
randShuffle(Mat(prmt));
//Greedy selection
vector<int> res(num_g);
vector<bool> taken(num_g);
for(int sg = 0; sg < num_g; sg++) {
double min_dist = std::numeric_limits<double>::max();
int minv = -1;
for(int tg = 0; tg < num_g; tg++) {
if(taken[tg]) continue;
//TODO: can save on re-calculation of pairs - calculate affinity matrix ahead
//double d = norm(sMu(Range(prmt[sg],prmt[sg]+1),Range(0,3)), tMu(Range(tg,tg+1),Range(0,3)));
//symmetric kullback-leibler
Mat diff = Mat(sMu(Range(prmt[sg],prmt[sg]+1),Range(0,3)) - tMu(Range(tg,tg+1),Range(0,3)));
Mat d = diff * Mat(Mat(source_covs[prmt[sg]]).inv() + Mat(target_covs[tg]).inv()) * diff.t();
Scalar tr = trace(Mat(
Mat(Mat(source_covs[prmt[sg]])*Mat(target_covs[tg])) +
Mat(Mat(target_covs[tg])*Mat(source_covs[prmt[sg]]).inv()) +
Mat(Mat::eye(3,3,CV_64FC1)*2)
));
double kl_dist = ((double*)d.data)[0] + tr[0];
if(kl_dist<min_dist) {
min_dist = kl_dist;
minv = tg;
}
}
res[prmt[sg]] = minv;
taken[minv] = true;
}
double dist = 0;
for(int i=0;i<num_g;i++) {
dist += norm(sMu(Range(prmt[i],prmt[i]+1),Range(0,3)),
tMu(Range(res[prmt[i]],res[prmt[i]]+1),Range(0,3)));
}
if(dist < best_dist) {
best_dist = dist;
best_res = res;
}
}
return best_res;
}
bool RateAlgoDT::runAlgo(bool PhysicsBits[128],
L1Analysis::L1AnalysisRecoMuonDataFormat * mu,
L1Analysis::L1AnalysisGMTDataFormat * gmt,
L1Analysis::L1AnalysisDTTFDataFormat * dttf)
{
float PTmax = -1;
TriggeredMuon tMuPtMax(0,-1,0);
TriggeredMuons tMu(mu,gmt,dttf);
if ( PhysicsBits[55] ) // CB single mu open
{
tMu.findMuons(controlPlotter_) && tMu.runGmtDttfMatching(controlPlotter_) && tMu.runDTTriggerMatching(controlPlotter_);
triggeredMuonsIt tMuIt = tMu.dt_trigger_muons.begin();
triggeredMuonsIt tMuEnd = tMu.dt_trigger_muons.end();
for (; tMuIt != tMuEnd; ++tMuIt)
{
int bx = tMuIt->gmt()->Bxdt[tMuIt->igmt()];
if (bx != 0) continue;
float eta= tMuIt->gmt()->Etadt[tMuIt->igmt()];
if (fabs(eta) > trigEta_) continue;
float pt = tMuIt->gmt()->Ptdt[tMuIt->igmt()];
if ( pt > PTmax)
{
PTmax = pt;
tMuPtMax = (*tMuIt);
}
}
}
if (plotter_ && PTmax >= trigPt_)
plotter_->fill(gmt,-1,tMuPtMax.igmt(),-1,-1,-1,tMuPtMax);
return (PTmax >= trigPt_) ? increment() : false ;
}
void Recoloring::Recolor(Mat& _source, Mat& source_mask, Mat& _target, Mat& target_mask) {
Mat source; _source.convertTo(source,CV_32F,1.0/255.0);
Mat target; _target.convertTo(target,CV_32F,1.0/255.0);
CvEM source_model,target_model;
//cvtColor(target,target,CV_BGR2Lab);
//cvtColor(source,source,CV_BGR2Lab);
TrainGMM(source_model,source,source_mask);
TrainGMM(target_model,target,target_mask);
vector<int> match = MatchGaussians(source_model,target_model);
Mat target_32f;
//if(target.type() != CV_32F)
// target.convertTo(target_32f,CV_32F,1.0/255.0);
//else
target.copyTo(target_32f);
const CvMat** target_covs = target_model.get_covs();
const CvMat** source_covs = source_model.get_covs();
Mat sMu(source_model.get_means()); Mat sMu_64f; sMu.convertTo(sMu_64f,CV_64F);
Mat tMu(target_model.get_means()); Mat tMu_64f; tMu.convertTo(tMu_64f,CV_64F);
int num_g = target_model.get_nclusters();
Mat pr; Mat samp(1,3,CV_32FC1);
for(int y=0;y<target.rows;y++) {
Vec3f* row = target_32f.ptr<Vec3f>(y);
uchar* mask_row = target_mask.ptr<uchar>(y);
for(int x=0;x<target.cols;x++) {
if(mask_row[x] > 0) {
memcpy(samp.data,&(row[x][0]),3*sizeof(float));
float res = target_model.predict(samp,&pr);
//cout << res << ":" << ((float*)pr.data)[0] << "," <<
// ((float*)pr.data)[1] << "," <<
// ((float*)pr.data)[2] << "," <<
// ((float*)pr.data)[3] << "," <<
// ((float*)pr.data)[4] << "," << endl;
Mat samp_64f; samp.convertTo(samp_64f,CV_64F);
//From Shapira09: Xnew = Sum_i { pr(i) * Sigma_source_i * (Sigma_target_i)^-1 * (x - mu_target) + mu_source }
Mat Xnew(1,3,CV_64FC1,Scalar(0));
for(int i=0;i<num_g;i++) {
if(((float*)pr.data)[i] <= 0) continue;
Mat A = /*Mat(*/
//Mat(target_covs[i]) *
//Mat(source_covs[match[i]]).inv() *
Mat(samp_64f - tMu_64f(Range(i,i+1),Range(0,3))).t();
Mat B = sMu_64f(Range(match[i],match[i]+1),Range(0,3)).t();
Xnew += Mat((A + B) * (double)(((float*)pr.data)[i])).t();
}
Mat _tmp; Xnew.convertTo(_tmp,CV_32F);
//allow for mask with alpha
float* _d = ((float*)_tmp.data);
float alpha = mask_row[x] / 255.0f;
for(int cl=0;cl<3;cl++)
_d[cl] = _d[cl] * (alpha) + row[x][cl] * (1-alpha);
memcpy(&(row[x][0]),_tmp.data,sizeof(float)*3);
}
}
}
//cvtColor(target,target,CV_Lab2BGR);
//cvtColor(source,source,CV_Lab2BGR);
//cvtColor(target_32f,target_32f,CV_Lab2BGR);
if(!this->m_p.no_gui) {
namedWindow("orig target");
imshow("orig target",target);
namedWindow("source orig");
imshow("source orig",source);
namedWindow("source masked");
Mat _tmp_S; source.copyTo(_tmp_S,source_mask);
imshow("source masked",_tmp_S);
namedWindow("dest target");
imshow("dest target",target_32f);
waitKey(this->m_p.wait_time);
}
target_32f.convertTo(_target,CV_8UC3,255.0);
}
