One of the key problems of conventional iris recognition systems is that they all require rigorous constraints imposed on the subjects during eye image acquisition. With the increasing demands in public safety and security, more attention has been given to unconstrained iris recognition as it reflects better realistic scenarios. Unconstrained iris recognition is a faster process that is more convenient to the subjects undergoing identity verification. With the lifting of the constraints, caution should be taken to address potential degradation of the iris image quality. This research introduces a new iris segmentation approach which is much more tolerant to realistic noise effects than traditional approaches, with higher accuracy and faster processing speed.
