Design of a dual biometric authentication system

Present research is on the combination of two biometric identification and recognition, i.e., iris pattern based and contours of fingerprint based, for identification and hence authentication. From the pre processed iris images of any person and fingerprint, the contours and countourlets are obtained. Localization followed by a normalization of an iris image is given for feature extraction using Gabor convolution, while minutia extraction of an enhanced finger print image is done using Crossing Number method. These features of contours and countourlets obtained from a sampled stream and in order to get the matching scores, they are estimated with samples in database. Comparison techniques used are hamming distance and Euclidean distance respectively for iris and fingerprint. Pattern of similarity and difference of scores are obtained after the passing of matching to the combination module. The score obtained finally is used to inform whether the person is authenticated or a bluffer. For a critical situation both the scores need to be matched i.e., for high security related applications both the scores need to be matched individually so that the combination will give better results. For some non critical application any one of the traits need to be matched so that the genuine persons would not miss a chance.