Authentication And Identification Of Individuals From The Iris Images

In this paper, we present an efficient method that allows us to authenticate and identify individuals by using iris images. In fact, the proposed method consists of three main steps. In the first step, we segment the image in order to define the upper and lower parts of the eyelids. We use two segments to exploit efficiently the region of interest of the iris and to extract only the interior half of the iris disc, which contains the most discriminate information. In the second step, the iris image is normalized by Daugman rubber sheet model, and then analyzed by bench of two 1D Log-Gabor filters to extract the texture characteristics. For the authentication and the similarity measurement between two irises, we use the Hamming distance with a threshold previously calculated. We then propose for the identification mode, a classification method based on the Multi-class SVM adopting the approach one against one. The proposed method has been tested on the Casia v1 database (756 iris images). For the authentication mode, we obtain very encouraging results: 1.39% for the global FAR, and 4.45% for the global FRR. For the identification mode, we obtain a rate recognition equals to 98.61%.