Iris Recognition on Textured Contact Lens Using Proximal Support Vector Machine Classification Based Lens Detection Method

Iris is one of the most promising biometric modalities, and is in usual use in enormously large-scale applications. The presences of a contact lens, mainly a textured cosmetic lens, create a challenge to iris recognition as it obfuscates the natural iris patterns. Earlier work used novel lens detection algorithm with Modified Local Binary Pattern analysis (MLBP) features to produce feature values. Two databases, namely, the IIIT-D Iris Contact Lens database and the ND-Contact Lens database, are organized to analyze the variations caused due to contact lenses. However lower accuracy is obtained for various pairs of gallery probe pairs due to less accurate lens detection algorithm. To deal this problem, the present work proposes PSVM Based Lens detection method using two additional features namely iris edge sharpness and Iris-Texton features for characterizing visual primitives of Iris textures for textured and soft contact lens based iris. The problem of lens detection in an iris image is approached as a three class classification problem: no lens, soft lens, and textured lens. This classification is efficiently done by using Proximal Support Vector Machine Classifier. Proximal SVM classifier has comparable test set correctness to that of standard SVM classifiers, but with considerably faster computational time that can be an order of magnitude faster. Experimental result of proposed system provides better result when compared with existing system.