An Improved Iris Recognition System with Template Security using CT and SVD

Abstract

The Iris biometric system is the most prominent method for identification of individual. Many researchers have been presented iris recognition methods from decade but a fully suitable solution for real world scenario is not implemented yet. The two major issues are responsible for it. First is no accurate method to operate on non-ideal iris images with high recognition rate. Second one is deployment of system with high security on the existing real world situations. In this Paper, the above mentioned problems are solved to an extent. An accurate and secured iris template encoding method is used for generate highly secured encoded binary pattern for iris template. Contourlet transform and Singular Value decomposition is used for this purpose. Beside this security feature, the proposed method used best combinations of algorithm for provide high accuracy as compared to conventional system of iris recognition. In Our approach IIT Delhi iris database is used as input image. Iris region from eye image is extracted by canny edge detection and Hough transforms to achieve high recognition rate. Daugman’s rubber sheet model is used for normalization. Security for normalized template is provided by Contourlet transform and Singular Value Decomposition. At last stage the combination of Hamming Distance and Normalized Correlation coefficient is used to achieve high recognition rate. So at each stage of iris recognition system all methods and algorithms are performed very well and provide higher accuracy as compared to existing iris recognition system.