Template Protection for PCA-LDA-based 3D Face Recognition Systems

Abstract

Authentication based on biometrics became significantly important over the last years. Privacy and security concerns arise by the extensive deployment of bio- metrics. The used biometric features itself have to be secured. We propose a security mechanism that solves privacy related problems in 3D facial verification systems. Our solution combines PCA/LDA feature extraction with the Helper Data Scheme for tem- plate protection. The evaluation shows recognition rates at the same level for secured and unsecured templates, which leads to a win-win scenario for users and providers of the adapted systems.