Secure biometric template protection via randomized dynamic quantization transformation

Fuzzy Commitment Scheme is one of the biometric encryption approaches for biometric template protection. The idea is to bind an identifier with a biometric template in binary format called difference vector during enrollment. Ideally, a difference vector is infeasible to recover either the biometric template or the identifier without any knowledge of the user's biometric data. Yet, this is only valid if the biometric template is uniformly random, but this is not the case in reality. In this paper, we propose a method known as randomized dynamic quantization transformation (RDQT) to binarize biometric data but still highly distinctive among the users and highly random. We demonstrate the implementation in the context of fingerprint biometrics. The experiment results and the security analysis in DB1 (FVC 2002) dataset suggest that the technique is feasible in practical usage.