Fingerprint-fingervein multimodal biometric authentication system in field programmable gate array

Abstract

This paper discusses the System-on-Chip (SOC) Field Programmable Gate Array (FPGA) based implementation of multimodal biometric authentication. Multimodal biometric can solve several problems related to the unimodal biometric authentication such as accuracy problem due to noisy data acquisition, biometric spoofing, and non-universality of the biometric traits. Moreover, significant accuracy improvement is difficult to achieve on a unimodal biometric authentication system and the complexity of the improvement process is not suitable for low powered embedded system implementation. By adjusting certain part of each biometric system modal, the complex processing unit can be eliminated allowing for embedded system implementation. Although the accuracy of each biometric system modal will be affected, the fusion of the biometric information will overcome the problem. Furthermore, the need for portable personal authentication has led this research to explore the multimodal biometric authentication in embedded system in which the system is developed in a resource constrained environment. As the first step, fingerprint and fingervein are used as biometric traits and the whole process is implemented in SOC FPGA and executed by general purpose embedded processor in which the biometric information fusion strategy applies at the matching score level. The accuracy of the system is promising with an Error Equal Rate (EER) of 0.33%.