International Journal of Computer Applications (0975 – 888) Volume 47– No.16, June 2012 40 Iris Segmentation using Geodesic Active Contour for Improved Texture Extraction in Recognition Minal K. Pawar P.G.Student, Department of E & Tc, Sinhgad College of Engineering, Pune. Sunita S. Lokhande Asst. Professor, Department of E & Tc, Sinhgad College of Engineering, Pune. V. N. Bapat GITM Jhajjav ABSTRACT Automatic identification/verification of a person through biometrics has been getting extensive attention due to an increasing importance of security. The most popular biometric authentication scheme employed for the last few years is Iris Recognition. The performance of iris recognition system highly depends on segmentation. For instance, even an effective feature extraction method would not be able to obtain useful information from an iris image that is not segmented accurately. The iris proposed recognition module consists of the preprocessing system, segmentation, feature extraction and recognition. Mainly it focuses on image segmentation using Geodesic Active Contours and comparison with traditional methods of segmentation. As active contours can 1) assume any shape and 2) segment multiple objects at the same time, they lessen some of the concerns related with conventional iris segmentation models. The iris texture is extracted in an iterative fashion by considering both local and global properties of the image. The matching accuracy of an iris recognition system is observed to improve upon application of the proposed segmentation algorithm. Experimental results on the CASIA (Institute of Automation, Chinese Academy of Sciences) Interval version3 iris databases implemented in MATLAB shows the efficiency of the proposed technique application. General Terms Pattern Recognition, Computer vision, Biometrics, Image processing. Keywords Iris recognition, iris segmentation, level sets, snakes, geodesic active contours (GACs), iriscodes. 1. INTRODUCTION Biometrics is the science which deals in identification of person based on his physiological and/or behavioral characteristics. The two categories of biometric identifiers include physiological and behavioral characteristics. Physiological characteristics are those which a person physically owns like fingerprints, iris patterns, face, ear shape, hand geometry, retina patterns, palm prints etc. and behavioral characteristics are the attitude of a person like gait, voice, signature etc. The requirements for a characteristic to be a biometric are universality, uniqueness, permanence and collectability. Applications of these systems include computer systems security, e-banking, credit card, access to buildings in a secure way. Iris recognition is recognizing a person by analyzing random pattern of iris. Iris recognition is the most trustworthy method of person authentication due to unique textures, non- invasiveness, and stability throughout the human life time, public acceptance, and availability of user friendly capturing Devices. The function of an iris recognition system is to extract, represent and compare the textural complexity present on the surface of the iris. Such a system comprises of iris segmentation, feature extraction (encoding) and feature matching. Each algorithm of iris recognition system starts with iris segmentation. Iris segmentation is to locate the valid part of the iris for iris biometrics, including finding the pupillary and limbic boundaries of the iris, localizing its upper and lower eyelids if they occlude and detecting and excluding any superimposed occlusions of eyelashes, shadows or reflections. It is reported that most failures to match in iris recognition system result from inaccurate iris segmentation. So for the better performance of the iris recognition system correct segmentation method plays vital role. 2. RELATED WORK In literature different methods used for segmentation, focused on finding parameters that best fit the iris. Several iris segmentation algorithms have been proposed. Daugman et al. [21] used an integro differential operator to segment the iris. Wildes [20] employed the binary edge map and the Hough transform to detect the iris and pupil boundaries. Ma et al. [18] used the Hough transform to detect the inner and outer boundaries of the iris. Recently, researchers have focused on processing of unideal iris images, which are defined to account for the off angle, occluded, blurred, and noisy images. For iris segmentation, most of the researchers assume that iris is circular or elliptical. However, in the case of unideal iris images such as the images that are affected by deviated gaze, eyelid and eyelash occlusions, non-uniform intensity, motion blur, reflections, etc., an iris may appear a s non-circular or non-elliptical.[1][2] Abhyankar and Schuckers et al. [7] proposed active shape model. Zuo et al.[8] used Randomized Elliptical Hough Transform Weighted Integro-differential operator. Kennell et al.[10] made use of Integro-differential operator, morphological operator for Image Binarization based on pixels and neighbourhood varaiance form fitting. Proenca and Alexandre [11] applied a clustering algorithm along with canny edge detector and circular Hough transform to separate the iris region from an unideal iris image. Arvacheh & Tizhoosh et al.[12] used near circular active contour model (snakes), interpolation process to improve performance; Integro- differential operator. Ann A. J., K.Wang , Ghassan J. M.et al.[3] used angular integral projection function(AIPF) for pupil boundary localization and for the localization of limbus boundary, the AIPF is applied again within two rectangles on both iris sides. Zheng et al. [16] applied Integro- differential operator; Iterative shift shrink & expand circumference process to minimize average intensity. Xu & Shi et al.[13] used Sobel