Studies of the Robustness of a Transformation-Based Multi-Biometric Template Schemes Protection

Abstract

As the number of applications that use multimodal biometrics grows, potential challenges to privacy and protection emerge. New multi-biometric technologies have been suggested in the context of privacy and confidentiality: template protection systems. For the relevant defensive technologies, very current solutions have been presented. Their decision on the optimal strategy is still being considered. This paper provides an evaluation of multi-biometric protection systems according to a transformation proposing some measures and metrics to test its performances and sturdiness faced against attacks. These measures enable evaluating the success of the approach in various scenarios. In this perspective, this article proposes a transformation-based multi-modal biometric protection system that focuses on BioHashing by merging fingerprint modalities. The fingerprint images are classified into three instances and represented by a feature vector using a 2D Log-Gabor filter. A biohashing is applied to its vectors to create Biocode vectors. Following that, a multi-instance score fusion formed by Biocodes of the fingerprints is merged at the matching level. Finally, to get a high score, multi-sample score fusion is used. Furthermore, a transformation-based protection model that focuses on BioHashing demonstrates how specific security and privacy may be validated.