An Efficient Approach to Generating Cryptographic Keys from Face and Iris Biometrics Fused at the Feature Level

Biometrics, defined as automated recognition of individuals based on their behavioral and biological characteristics, is beginning to gain acceptance as a legitimate authentication method and a practicable option to traditional identification methods in several application areas. Biometric cryptosystems, designed to generate a cryptographic key from a biometric trait, incorporate high level of security provided by cryptography and non-repudiation provided by biometry, as well as eliminating the need for a user to remember long passwords or carry tokens. Unlike unimodal biometric systems that employ single feature, multimodal biometric cryptosystems generate keys from two or more individual modalities typically fused at feature level. Fusing feature sets related to different modalities prevents possible spoof attacks and provides the system with higher level of overall security. This paper present an efficient approach to secure cryptographic key generation from iris and face biometric traits. Features extracted from preprocessed face and iris images are fused at the feature level and the multimodal biometric template is constructed from the Gabor filter and Principal Component Analysis outputs. This template is used to generate strong 256-bit cryptographic key. Experiments were performed using iris and face images from CASIA and ORL databases and the efficiency of the proposed approach is confirmed.