BIOFUSE: A framework for multi-biometric fusion on biocryptosystem level

Abstract

Biometric cryptosystems or biocryptoystems are gaining prominence for cryptographic key generation, encryption and biometric template protection. However, the most popular state-of-the-art biocryptosystems- fuzzy commitment and fuzzy vault are prone to multiple security attacks. Recently proposed multi-biometric cryptosystems improve security and enhance recognition performance. They perform the fusion of multi-biometric characteristics with either a single biocryptosystem or independently accessed, multiple biocryptosystems. An attack on any of the involved biocryptosystems can weaken the security of the whole system. In our paper, we propose a multi-biometric fusion framework- BIOFUSE, that combines fuzzy commitment and fuzzy vault using the format-preserving encryption scheme. BIOFUSE makes it improbable for an attacker to get unauthorized access to the system without impersonation of all the biometric inputs of the genuine user at the same instant. We present 4 most basic ways of constructing BIOFUSE and found only 1 named S-BIOFUSE (S3) as a secure design. We compare the recognition performance of the proposed scheme with existing multi-biometric cryptosystems on various databases. The results show 0.98 true match rate at 0.01 false match rate on a virtual IITD-DB1 database that indicates that our proposed work achieves significantly good recognition performance while providing high security.