IJCSNS International Journal of Computer Science and Network Security, VOL.11 No.7, July 2011 93 Manuscript received July 5, 2011 Manuscript revised July 20, 2011 Automatic Localization of Iris Using Region Properties Haider Ali1 † and Ahmad Ali2 † and Riaz Ul Husnain3 †† and Roman Khan4 †† and Mohsin Khan5 †† and Ihsan Ullah Khan$6 †† , CIIT Abbottabad Campus, Engineering Department KPK Pakistan Summary Iris localization is one of the most important step in iris based recognition systems. Iris localization means locating the inner boundary (pupil localization) and outer boundary of iris. Both boundaries of the iris are nearly circular which are surrounded by pupil, sclera, eyelashes and eyelids. In the proposed algorithm inner boundary (Pupil) is localized by using two region properties Eccentricity and Area without using any iterative method. On the other hand the outer boundary of iris is located using Gaussian derivatives. The proposed algorithm is tested on CASIA-1 Iris database. Experimental results show that the proposed method is quite fast efficient and accurate method. Key words: Iris Recognition, Iris Segmentation, Pupil Localization. Introduction A lot of question arises to a person’s identity in variety of forms and in different contexts. Is this person is entitled for using these facilities? Is this person a wanted criminal? Or is this person is allowed to enter here? And the list may goes on and on. But with this discipline of biometrics, we can reduce misrepresentation; fraud etc in all the above listed lists and also in all others disciplines as well. This is why biometric based person identification is getting popular due to its accuracy and high reliability. For this purpose different biometrics are used like, Face recognition, Fingerprint identification, Voice recognition, Signature verification and Iris recognition. Among these biometrics iris is believed to be more foolproof as it can’t be artificially copied and also it remains same during a person’s life time. Generally an iris recognition system is based on four steps or blocks which are given in Fig1[1]. Fig. 1 Major blocks of Iris recognition system Iris localization is one of the fundamental step of any Iris recognition system, because the speed and accuracy of the system is totally solely depends on the segmented Iris. Segmentation of Iris is normally considered as Iris segmentation is considered as first step in any Iris recognition system [2]. In many machine vision applications, such as Pupil Tracking [3], Iris Recognition, Pupil Size estimation [5], Ocular Torsion Measurement [4], Pupillometry [6] and Point of Gaze Extraction [7] etc. localization of the inner boundary (Pupil Localization) of Iris is consider as the most important preprocessing step. As, performance and accuracy of any pupil based system depend on pupil localization. Thus, it is very important to develop an accurate and fast pupil localization method for these systems. Pupil is nearly a circular region which is located in the center of the Iris of the eye. The basic function of pupil is to control the amount of light that enters in the eye [8]. Pupil absorbs most of the light that enters inside the eye. Due to this reason it appears black. Most of the algorithms use two methods to locate the pupil region either by finding its edges using circular mask since pupil is nearly a circular region or by using thresholding as it is the darkest region in an eye image. A circular based algorithm proposed by J. Daughman [9][10][11] is very much popular for iris localization. In it circular edge detector are used for Iris segmentation. Wildes [12] proposed a two stage algorithm. It constructs an edge map based on gradient in stage-1. In stage-2 Hough transform is used for Pupil and Iris are segmentation. Lui et al [13] uses an improve Hough Transform for the segmentation of Pupil and Iris region. A. Basit et al [14] uses adaptively binarization and centroid of the region utilization for pupil localization and for iris localization gradient of pupil’s central horizontal line is calculated. Comparison and Recognition Image Acquisition Segmentation Normalization and Iris code generation