Returns image discriptor size
sz = extractor.descriptorSize()
This is basically size(Vocabulary,1) (i.e number of clusters).
size(Vocabulary,1)