Novel and Fast Approach for Iris Location

Abstract

Iris recognition is regarded as the most reliable and accurate biometric identification system. Speed and accuracy of iris location are required to achieve the success of iris recognition. This paper presents a new iris location algorithm to improve speed and accuracy. The first step is to detect a point inside the pupil using the method of minimum local block mean. Then the specialized boundary detection mask (SBDM) is employed to locate three points along the inner and outer iris boundaries. Finally, Thales' theorem is applied to find the hypotenuse that is a diameter of their circumcircle and then the circle parameters can be calculated. Based on the CASIA iris image database, experiment results show that the proposed method is superior to the existing approaches on the speed and precision of iris location.