Ordinal Region-Based Representations for Iris Recognition

This paper presents local ordinal measures for iris recognition. It is inspired by the mechanism of response properties in neuroscience. The ordinal relationships can preserve the stability among a group of neighboring regions against noise in a given iris image. In this scheme, the iris texture is represented by a set of similarities and dissimilarities across neighboring regions. The experiment results show that proposed simple algorithm can achieve a reasonable recognition rate as compared to existing complex algorithms.