Score Level Fusion for Face-Iris Multimodal Biometric System

Abstract

A high performance face-iris multimodal biometric system based on score level fusion techniques is presented in this paper. The aim of multimodal biometric systems is to improve the recognition performance by fusing information from more than one physical and/or behavioral characteristics of a person. This paper focuses on combining the strengths of face and iris modalities by employing the optimization method particle swarm optimization (PSO) to select facial features and the well known matching score level fusion technique, Weighted-Sum Rule, in order to obtain better recognition accuracy. Prior to performing fusion of face and iris modalities, standard feature extraction methods on face and iris images are employed separately. The unimodal and multimodal systems are experimented on different subsets of FERET, BANCA, and UBIRIS databases. Evaluation of the overall results based on recognition performance and ROC analysis demonstrates that the proposed multimodal biometric system achieves improved results compared to unimodal and other multimodal systems.