Cancelable biometrics for bimodal cryptosystems

摘要

Biometric-based techniques have recently emerged as a trustworthy and effective approach of user authentication; however, unlike conventional authentication methods such as passwords and tokens, if an enrolled biometric template is compromised, usually it cannot be revoked or re-issued. In this paper, four naive cancelable techniques, namely, shifting, password-dependent shifting, XOR and adding, for a bimodal biometric cryptosystem are presented. The proposed cancelable techniques are designed to be embedded into any bimodal biometric cryptosystem. The bimodal biometric cryptosystem uses speech and electrocardiogram signals as biometric information. The biometric cryptosystem implements an error-correction layer using the Hadamard code. The performance of the four cancelable techniques is assessed using ECG signals from MIT-BIH database and speech signals from a speech database created for testing purposes. The results show that the best performance in terms of FAR and FRR metrics is achieved with XOR and adding techniques.