Improved Fuzzy Hashing Technique for Biometric Template Protection

Biometrics provides a new dimension of security to modern automated applications since each user will need to prove his identity when attempting an access. However, if a stored biometric template is compromised, then the conventional biometric recognition system becomes vulnerable to privacy invasion. This invasion is a permanent one because the biometric template is not replaceable. In this paper, we introduce an improved FuzzyHashing technique for biometric template protection purpose. We demonstrate our implementation in the context of fingerprint biometrics. The experimental results and the security analysis on FVC 2004 DB1 and DB2 fingerprint datasets suggest that the technique is highly feasible in practice.