Fusion of face and voice for a multimodal biometric recognition system

Abstract

Biometric authentication system takes a primary role in the present modern society, computers are becoming a part of everyday life. It provides more security than the traditional systems. In traditional authentication systems password, pin-number, or signature is used for identification but these can be lost, stolen or subject to spoofing attacks. This paper introduces combination of two individual human traits, face and voice signal which are used for identification. The biometric authentication system with two traits supports more security and reliability than the single source of identification system. This paper presents a biometric recognition system integrating face and voice signal based on score level fusion. The features are extracted individually from the preprocessed traits and then classified the data using Gaussian mixture model. After classification, fuse the traits to make the training dataset. Test data is compared with the training dataset and then display the result whether the individual is genuine or an impostor. Performance measures like False Acceptance Rate (FAR), False Rejection Rate (FRR), Equal Error Rate (EER), and Failure To Capture (FTC) are calculated and performance evaluated. it is proved that the proposed biometric system overcomes the limitations of individual biometric systems and also meets the less response time as well as the good accuracy requirements.