The human iris is one of the most reliable biometric features. Since it is a live organ, variations caused by pupil contraction/dilation will degrade the performance of a biometric system based on iris. This paper presents a method for generating images of irises with a specific dilation coefficient. The approach presented here uses a mathematical model that aims to emulate the dynamic of the iris with respect to the pupil dilation/contraction. The estimated images are generated using image an image warping technique. The iris image is approximated by remapping the radius of the polar coordinates using a nonlinear function. The proposed method benefits of low complexity and provides better results with respect to the photorealistic aspect and to the performance of iris biometric systems. The performance of the proposed model applied to the iris as a biometric system is tested using a commercial iris recognition system.
展开▼