Multifactor authentication system based on color visual cryptography, facial recognition, and dragonfly optimization

Abstract

ABSTRACT Facial recognition as an authentication factor requires that facial images and features are tamper-free. Visual cryptography (VC) is commonly used for this purpose but leads to additional computational overhead and lower recognition accuracy. To address these issues, we propose a multifactor authentication (MFA) system based on facial recognition that uses VC to secure biometric data, and also as a second authentication factor. The shares generated by VC are used as authentication tokens. These tokens are verified by the same facial recognition algorithm used to recognize a live facial image of the user. This simple albeit novel approach leads to lower computational cost because the same algorithm can be used to verify both authentication factors. To maximize verification accuracy, the binary dragonfly optimization algorithm is used to maximize the quality of the recovered image from VC, as well as the accuracy of the facial recognition algorithm itself through feature selection. The combination of these separate ideas leads to a novel MFA system that is efficient and highly accurate. Experimental verification based on various face image databases depicts the proposed system’s security, efficiency, and near-ideal recognition accuracy of up to 99.81%.