Abstract
Human identification via multibiometrics is a very promising approach to improve the overall system's accuracy and recognition performance. In recent years, several approaches toward studying the fusion strategies of different biometric evidence have been proposed. However, there are a number of major problems detected on some of those approaches such as weakness against spoofing attacks and higher acceptable error rate. In this paper, a novel multibiometrics fusion strategy based on dual iris, visible and thermal face traits is proposed. Initially, the features of related biometrics (dual iris, visible with thermal faces) are fused in feature level. Then, the matching scores of iris and face traits are fused via triangular norm. The proposed multibiometrics fusion achieves higher identification performance as well as immune to spoofing attacks. All the simulation are performed based on a virtual multibiometrics database, which merges the challenging CASIA-Iris-Thousand database with noisy samples and the NVIE face database with visible and thermal face images. The results show that the proposed fusion strategy outperforms the state-of-the-art approaches in the literature.