Iris recognition on edge maps

Iris recognition is a potential tool in secure personal identification and authentication due to its desirable properties such as uniqueness, non-invasiveness and stability of human iris patterns. In this paper, a new approach based on the Hausdorff distance measure is proposed for iris recognition. A new measure, namely local partial Hausdorff, is computed directly on the binary edge features of the normalized iris images. The proposed edge dissimilarity measure is tested on a set of well-segmented UPOL iris images. For different values of parameters such as block size and partialness, the results indicate high recognition performance of the proposed method.