ZeroBio - Evaluation and Development of Asymmetric Fingerprint Authentication System Using Oblivious Neural Network Evaluation Protocol

Abstract

We propose a cryptographic protocol for biometrics authentication without revealing personal biometrical data against malicious verifier. Our protocol uses a neural network and zero-knowledge interactive proof. In this paper, we developed a sample implementation system of our proposed protocol and we evaluate the performance and the accuracy of the proposed protocol. Especially, we study several algorithms for feature extraction of minutiae of fingerprint which is appropriate to our protocol. We examine false acceptance rates and rejection rates