Remote Multimodal Biometric Authentication using Visual Cryptography

Abstract

This work proposes an architecture for multimodal biometric recognition systems where user, recognition system, and template database are remotely located over a network. As the number of biometrics are limited and once lost they are compromised forever, it becomes imperative to design systems that optimize recognition rates and also address security and privacy issues for biometric-enabled authentication schemes. The proposed architecture provides revocability to multimodal biometric templates and secures their storage and transmission over a remote network with the help of visual cryptography technique. The proposed architecture gives a good matching performance and also fulfills four template protection criteria, i.e., security, diversity, revocability, and performance. Various attack scenarios such as phishing, replay, database, man-in-middle, and attack via record multiplicity are also addressed.