Abstract
Iris recognition system became a reliable system for authentication and verification tasks. It consists of five stages:
image acquisition, iris segmentation, iris normalization, feature encoding, and feature matching. Iris segmentation stage
is one of the most important stages. It plays an essential role to locate the iris efficiently and accurately. In this paper, we
present a new approach for iris segmentation using image processing technique. This approach is composed of four main
parts. (1) Eliminating reflections of light on the eye image based on inverting the color of the grayscale image, filling
holes in the intensity image, and inverting the color of the intensity image to get the original grayscale image without any
reflections. (2) Pupil boundary detection based on dividing an eye image to nine sub-images and finding the minimum
value of the mean intensity for each sub-image to get a suitable threshold value of pupil. (3) Enhancing the contrast of
outer iris boundary using exponential operator to have sharp variation. (4) Outer iris boundary localization based on
applying a gray threshold and morphological operations on the rectangular part of an eye image including the pupil and
the outer boundaries of iris to find the small radius of outer iris boundary from the center of pupil. The proposed
approach has been tested on CASIA v1.0 iris image database and other collected iris image database. The experimental
results show that the approach is able to detect pupil and outer iris boundary with high accuracy results approximately
100% and reduce time consuming.