Precise Pupil Boundary Detection Using Angular Integral Projection and Active Contour Model

Abstract

Precise pupil contour detection is of great importance to iris localization in iris-based biometric applications. A new algorithm for detecting the precise pupil contour in iris images is presented in this paper. The proposed algorithm detects the exact pupil contour rather than considering pupil boundary as a perfect circle which degrades the performance of later iris matching. It comprises two detection phases, the first for circular pupil contour and the second for the exact pupil contour. In the first phase, the approximate pupil center is determined as the center of mass of the binarized iris image, then the angular integral projection function (AIPF) is applied to detect a set of pupil boundary points, and finally the pupil circular contour is obtained by fitting a circle to the detected boundary points. While in the second phase, the accurate pupil contour is detected by employing an active contour model which is based on greedy optimization algorithm that is initialized on the pupil circle obtained form the first phase. Experimental results on three different iris databases indicate that the proposed algorithm can accurately detect pupil contour in less execution time as compared with other existing methods for iris segmentation, which confirms its performance.