Score Level Fusion based Multimodal Biometric System using Symmetric-Sums

Abstract

Multimodal biometric systems, which incorporate data from several biometric sources, are being developed to enhance the accuracy of user authentication and overcome the inherent limitations of unimodal biometric systems. In this paper, an improved framework using binary functions called Symmetric-sums is used to develop an efficient multimodal biometric system. These S-sums are formulated using triangular norms. The experimental results are examined on a self-constructed database comprising fingerprint and face biometric modalities. The results exhibited by eight different S-sums are then compared with the results generated by their respective T-norms. The proposed Symmetric-sums approach outperformed state-of-the-art methodologies using T-norms in multimodal biometric systems, achieving an average accuracy rate of 98.06%.