Towards generating protected fingerprint templates based on bloom filters

Abstract

In order to satisfy the requirements for security and privacy of biometric enrolment data records, it is essential to protect this reference data by applying appropriate template protection schemes. Bloom filters have been applied successfully on iris biometrics and face biometrics and achieved good result in terms of irreversibility and biometric performance. In this paper we study, whether it is feasible to employ Bloom filters on fingerprint templates. In order to be resilient with fingerprint sample variations, a pre-alignment process is applied prior to binary template generation. After generating the binary template matrix, we propose to subdivide the matrix and achieve a variable size of the binary template. Experiments were conducted on public databases to confirm the proposed ideas. According to experimental results, applying Bloom filters on fingerprint template doesn't degrade the accuracy of the fingerprint recognition system. Therefore, we can conclude that it is feasible to apply Bloom filters on fingerprint biometrics.