D-12-111 On Extracting Eye Positions for Dissimilarity Representation