Iris Recognition with Low Template Size

Abstract

Among all the biometric techniques known nowadays, Iris Recognition is taken as the most promising of all, due to its low error rates without being invasive and with low relation to police records. Based on Daugman’s work, the authors have developed their own Iris Recognition system, obtaining results that show the performance of the prototype and proves the excellences of the system initially developed by Daugman. A full coverage of the pre-processing and feature extraction blocks is given. Special efforts have been applied in order to obtain low template sizes and fast verification algorithms. This effort is intended to enable a human authentication in small embedded systems, such as an Integrated Circuit Card (smart cards). The final results show viability of this target, enabling a template size down to 256 bits. Future works will be focussed in new feature extraction algorithms, as well as optimising the pre-processing block.