Faster secure computation for biometric identification using filtering

Several attempts have been made to apply Secure Multiparty Computation (SMC) to the context of biometric recognition. Such constructions are very bandwidth and time-consuming. In order to reduce both consumptions, we propose a new construction for biometric identification based on filtering techniques that has the same security properties as the other SMC-based schemes but runs significantly faster while keeping good biometric performances. We propose as an example a secure iris identification algorithm with a fast filtering phase based on Hamming distance. We moreover present implementation results for this proposal that confirm our improvements.