Localization of noncircular iris boundaries using morphology and arched Hough transform

Iris segmentation, including localization and noise removal, is a fundamental step in iris recognition systems as the performance of the system is highly depend on this step. The aim of localization is to detect inner (pupil) and outer (limbic) boundaries. Noise removal consists of eliminating eyelids and eyelashes from localized image. In this paper, we propose a new localization algorithm, in which, unlike the previously reported works, no assumption for the shape of the boundaries is supposed. Inner boundary is localized by use of a coarse-to-fine strategy. In so doing, a set of morphological operators and canny edge detector are applied to the square region, which surrounds the pupil. Outer boundary is divided into right and left sides in which they are detected by arched Hough transform and finally merged together. The proposed algorithm is tested on the CASIA and MMU databases and the localized image is evaluated using the ground truth method. The obtained results indicate that our algorithm improves the precision of the iris localization.