ROBUST MULTIMODAL BIOMETRIC AUTHENTICATION INTEGRATING IRIS, FACE AND PALMPRINT

Abstract

Fusion of multiple biometric modalities for human authentication performance improvement has received considerable attention. This paper presents a robust multimodal biometric authentication scheme integrating iris, face and palmprint based on score level fusion. In order to overcome the limitation of the possible missing modalites, the multiple parallel support vector machines (SVMs) fusion strategy is applied, in which all possible modality combination cases are considered and each case has a corresponding SVM to combine the scores to generate a fused score for the final decision. Experimental results show that the proposed multimodal scheme is more robust and flexible, especially when some of the biometric modalities are unavailable.