Secure face template generation via local region hashing

Abstract

Security is an important aspect in the practical deployment of biometric authentication systems. Biometric data in its original form is irreplaceable and thus, must be protected. This often comes at the cost of reduced matching accuracy or loss of the true key-less convenience biometric authentication can offer. In this paper, we address the shortcomings of current face template protection schemes and show the advantages of a localized approach. We propose a framework that utilizes features from local regions of the face to achieve exact matching, and thus, enables the security offered by hash functions like SHA-256. We study the matching accuracy of different feature extractors, and propose measures to quantify the security offered by the scheme under reasonable real-world assumptions. The efficacy of our approach is demonstrated on the Multi-PIE face database.