Improving personal identification accuracy using multisensor fusion for building access control applications

Abstract

This paper discusses a multimodal biometric sensor fusion approach for controlling building access. The motivation behind using multimodal biometrics is to improve universality and accuracy of the system. A Bayesian framework is implemented to fuse the decisions received from multiple biometric sensors. The system accuracy improves for a subset of decision fusion rules. The optimal rule is a function of the error cost and a priori probability of an intruder. This Bayesian framework formalizes the design of a system that can adaptively increase or reduce the security level. This is important to systems designed for varying security needs and user access requirements. The additional biometric modes and variable error costs give the system adaptability improving system acceptability. This paper presents the framework using three different biometric systems: voice, face, and hand biometric systems.