Security of Biometrics Using Multimodal Approach

Biometric based access control systems (BACS) are widely used for robust and reliable identity verification, but these systems also raise some serious privacy concerns. The biometric templates, used for access control process, are stored in the system database which is vulnerable to variety of attacks. This paper presents a two stage solution for securing human iris templates used for BACS. The proposed method first models the fuzziness of iris template bits, which is used to select a set of stable template bits. Channel coding based on low density parity check coding is then applied to the selected stable bits to protect template against adversary attacks. To enhance the security of the proposed method further, both human irises are used for the access control process. Security performance of the developed algorithm is evaluated using publicly available CASIA database. Simulation results show that the proposed method provides stronger computational security than existing iris template security methods.