A fast adaboosting based method for iris and pupil contour detection

Abstract

The iris localization plays a fundamental role in the recognition process because the speed and performance of the iris recognition system largely depends on the quality of the pupil and iris detection. This process includes the detection of inner (pupil) and outer (iris) boundaries. In this paper we present a new method for iris and pupil boundaries detection based on Adaboosting technique for localization of circular objects and an algorithm based on the elements of analytic geometry, in particular, the determination of the bounded circumference of a tangential square that encloses the pupil and iris. The proposed approach overcomes the limitations that had previous methods regarding the use of images obtained under not controlled conditions like specular light reflected in the pupil or in the iris. We experimented our approach comparing the results in detection with the results obtained by Daugman algorithm using images from two contrasting databases, CASIA and UBIRIS. © Springer-Verlag Berlin Heidelberg 2006.