Lip-Sync Personal Authentication System Using Movement Feature of Lip

Biometric authentication using physiological feature has issue of forgery and theft. Furthermore it can´t change key in these case. In this paper, we propose new personal authentication system that uses behavioral feature of lip-movement among biometric features. Movement feature of lip is obtained by using visual camera. The proposal system is not dependent on voice information. Thus user can perform personal authentication by speaking in lip-sync in front of camera. Time series data is obtained by movement of lip. DP(Dynamic Programming) matching is used in personal authentication. Evaluation experiments were carried out in 43 person subjects. The system performance showed that false rejection rate(FRR) is 3.0%, false acceptance rate(FAR) is 14.5% in the case of using same words, 0.0% in the case of using different words, respectively. From experimental result, we found the effectiveness of the proposal system.