Providing multimodal biometric authentication using five competent traits

Abstract

Recognition accuracy of a single biometric authentication system is often much limited; hence, a multimodal biometric approach for identity verification is proposed. A new way of person authentication based on five-competent traits, namely, iris, ear, palm print, fingerprint and retina, is proposed in this paper. Each metric is analysed individually to get the matching scores from the corresponding feature sets. These scores are then combined using weighted sum fusion rule. In order to provide liveness verification for our authentication system, we employ the retinal blood vessel pattern recognition. To validate our approach, several experiments were conducted on the images obtained from five different datasets. The experimental results reveal that this multimodal biometric verification system is much more reliable and precise than the single biometric approaches.

Keywords:

• fusion
• matching score
• multimodal
• weighted sum rule

The authors wish to thank the National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Science, CASIA, for providing the iris and palm print databases, VARPA retinal images for authentication, VARIA for providing retinal images, and FVC2002 for fingerprint dataset and USTB for providing the ear database.