An approach to robust biometric key generation system design

Abstract

This paper presents a novel approach to the design of robust multimodal biometric cryptosystems. The design objectives behind the system are robustness, privacy of user’s biometric templates and stable cryptographic key generation. The framework presented in this paper employs two modalities and a look-up table. The hashes of cryptographic keys generated from a biometric template during the enrollment phase are stored in the look-up table with cancelable templates generated from the sample belonging to different modality of the same subject. During the operation phase, the system releases the key, only if the hash of the key generated from the provided biometric sample is found in the look-up table, and the similarity score between corresponding cancelable templates is less than a predefined threshold. The implementation of the proposed framework with iris and fingerprint biometrics is evaluated with the CASIA biometric template database.