Nearest-Neighbour-Search-Neural-Networks-Product-Quantization/TEXMEX/PQ Library/AsymmetricDistanceComputation.m

function [ iXMinDist ] = AsymmetricDistanceComputation(CSubSetMat, iXQ, YSetMat, nRecall)
%UNTITLED Summary of this function goes here
%   Detailed explanation goes here

kk=size(CSubSetMat,2);
mm=size(CSubSetMat,3);
imgSz=size(YSetMat,1);
subImgSz=imgSz/mm;
Ntests=size(YSetMat,2);
DistMat=zeros(kk,mm);
iXMinDist=zeros(Ntests,nRecall);

for ii=1:Ntests
    for jj=1:mm
       DistMat(:,jj)=EuclideanDistancesMat(YSetMat((jj-1)*subImgSz+1:jj*subImgSz,ii),CSubSetMat(:,:,jj));
    end
    DistVec = SumCentroidsDistances(DistMat,iXQ);
    [~, iXMinDisttemp] = sort (DistVec);
    maxRec=size(iXMinDisttemp,2);
        if  maxRec >= nRecall
            iXMinDist(ii,:)=iXMinDisttemp(:, 1:nRecall);
        else
            ouch=size(iXMinDisttemp,1);
            iXMinDist(ii,:)=[iXMinDisttemp, zeros(1,nRecall-maxRec)];
        end
    pp=double(ii/Ntests*100);
    if(pp<10)
    %fprintf('\b\b\b\b\b\b %2.2f%%',pp);
    else
    %fprintf('\b\b\b\b\b\b\b %2.2f%%',pp);
    end
end

% for rr=1:Ntests
%     for ii=1:Ntr
%         SEucDist=0;
%                 
%         for jj=1:m
%             SEucDist=SEucDist+dEucXCent(iYQ(rr,jj),ii,jj)^2;
%         end
%         TEucDist(ii)=sqrt(SEucDist);
%     end
%     pp=rr/Ntests*100;
%     if(pp<10)
%         fprintf('\b\b\b\b\b\b%2.2f %%',pp);
%     else
%         fprintf('\b\b\b\b\b\b\b%2.2f %%',pp);
%     end
%     [~,indMin]=min(TEucDist);
%     iXMinDist(rr)=indMin;
% end

end