boblsturm@0: function [ Y, PrunedFrames, KeepCorpusFrames ] = prune_corpus( target, corpus, reduction_coef )
boblsturm@0: 
boblsturm@0:     if reduction_coef == 1
boblsturm@0:         Y = corpus;
boblsturm@0:         PrunedFrames = [];
boblsturm@0:         KeepCorpusFrames = [];
boblsturm@0:         return;
boblsturm@0:     end
boblsturm@0:     [targetRows, targetCols]= size( target );
boblsturm@0:     [corpusRows, corpusCols]= size( corpus );
boblsturm@0:     
boblsturm@0:     % Matrix: distances of every target frame to every other target frame
boblsturm@0:     % i.e. Self-similarity matrix
boblsturm@0:     TargetSelfSimMat = zeros(targetCols);
boblsturm@0:     
boblsturm@0:     % Matrix: distances of every target frame to every corpus frame
boblsturm@0:     % Matrix( i, j ) = Similarity of corpus frame i to target frame j
boblsturm@0: %     TargetToCorpSimMat = -1*ones(WCols, targetCols);
boblsturm@0:     
boblsturm@0:     % Calculate self-similarity of target (euclidean dist)
boblsturm@0:     X2 = sum(target.^2,1);
boblsturm@0:     TargetSelfSimMat = bsxfun(@plus,X2,X2')-2*(target'*target);
boblsturm@0: %     TargetSelfSimMat = exp(-(1/10) * DistanceMat);
boblsturm@0:     
boblsturm@0:     % Calculate the distance between each corpus frame to first target
boblsturm@0:     % frame
boblsturm@0:     KeepCorpusFrames = [];
boblsturm@0:     RemainingTargetFrames = 1:targetCols;
boblsturm@0:     RemainingCorpusFrames = 1:corpusCols;
boblsturm@0:     
boblsturm@0:     while ~isempty(RemainingTargetFrames)
boblsturm@0:         Dist = repmat(target(:,RemainingTargetFrames(1)),1, ...
boblsturm@0:             length(RemainingCorpusFrames)) - corpus(:,RemainingCorpusFrames);
boblsturm@0:         Distances = sum(Dist.^2,1);
boblsturm@0:         % compute mean distances
boblsturm@0:         meanDistance = mean(Distances);
boblsturm@0:         
boblsturm@0:         % find those corpus frames closer than reduction_coef of mean
boblsturm@0:         idxcorpuskeep = Distances < reduction_coef*meanDistance;
boblsturm@0:         KeepCorpusFrames = [KeepCorpusFrames RemainingCorpusFrames(idxcorpuskeep)];
boblsturm@0:         
boblsturm@0:         % find those target frames within twice that to next frame
boblsturm@0:         distancetonext = TargetSelfSimMat(RemainingTargetFrames(1)+1,RemainingTargetFrames(1));
boblsturm@0:         idxtarget = TargetSelfSimMat(RemainingTargetFrames,RemainingTargetFrames(1)) ...
boblsturm@0:             < 2*distancetonext;
boblsturm@0:         
boblsturm@0:         if ~any( idxtarget ) && distancetonext == 0
boblsturm@0:             idxtarget = RemainingTargetFrames(2);
boblsturm@0:         end
boblsturm@0:         
boblsturm@0:         % shrink corpus
boblsturm@0:         RemainingCorpusFrames = RemainingCorpusFrames(~idxcorpuskeep);
boblsturm@0:         
boblsturm@0:         % shrink target frames index
boblsturm@0:         RemainingTargetFrames = RemainingTargetFrames(~idxtarget);
boblsturm@0:         
boblsturm@0:         fprintf( '%d remaining frames...\n', size( RemainingTargetFrames ) );
boblsturm@0:     end
boblsturm@0:     
boblsturm@0: KeepCorpusFrames = sort( KeepCorpusFrames );
boblsturm@0: Y = corpus( :, KeepCorpusFrames );
boblsturm@0: PrunedFrames = setdiff( 1:corpusCols, KeepCorpusFrames );
boblsturm@0: 
boblsturm@0: fprintf( 'Pruned %d frames...\n', size( PrunedFrames ) );
boblsturm@0: 
boblsturm@0: end