Iris recognition using multiscale morphologic features

A new set of features for personal verification and identification based on iris image is proposed in this paper. The method consists of three major components: image pre-processing, feature extraction and classification. During image pre-processing, the iris segmentation is carried out using Restricted Circular Hough transformation (RCHT). Then only two disjoint quarters of the segmented iris pattern are normalized which is used to extract features for classification purposes. Here, method for feature extraction from iris pattern is based on multiscale morphologic operator. In this approach, the iris features are represented by the sum of dissimilarity residues obtained by applying morphologic top-hat transform. For classification purposes the multi-class problems is transformed to two-class problem using dichotomy method. The performance of the proposed system is tested on four benchmark iris databases UPOL, MMU1, IITD, and UBIRIS and is compared with well known existing methods.