Fusing Binary Templates for Multi-biometric Cryptosystems

Biometric cryptosystem has been proven to be one of the promising approaches for template protection. Since most methods in this approach require binary input, to extend it for multiple modalities, binary template fusion is required. This paper addresses the issues of multi-biometrics' performance and security, and proposes a new binary template fusion method which could maximize the fused template discriminability and its entropy by reducing the bits dependency. Three publicly available datasets are used for experiments. Experimental results show that the proposed method outperforms the state of the art methods.