Fast Iris Segmentation by Rotation Average Analysis of Intensity-Inversed Image.

Iris recognition is a reliable and accurate biometric technique used in modern personnel identification system. Segmentation of the effective iris region is the base of iris feature encoding and recognition. In this paper, a novel method is presented for fast iris segmentation. There are two steps to finish the iris segmentation. The first step is iris location, which is based on rotation average analysis of intensity-inversed image and non-linear circular regression. The second step is eyelid detection. A new method to detect the eyelids utilizing a simplified mathematical model of arc with three free parameters is implemented for quick fitting. Comparatively, the conventional model with four parameters is less optimal. Experiments were carried out on both self-collected images and CASIA database. The results show that our method is fast and robust in segmenting the effective iris region with high tolerance of noise and scaling. [ABSTRACT FROM AUTHOR]