Face and Lip-Reading Authentication System Based on Android Smart Phones

摘要

In all biometrics, face features are widely used because of their advantages. Compared with other biological features, the recognition accuracy based on face features is not very high, and the system is easy to be deceived by photos. Identity authentication based on multi-modal biometric fusion has opened up a new field of development, which has higher technical requirements and better application performance than single biometric recognition and authentication system. As a dynamic feature, lip reading feature is not easy to imitate, and can be collected with face information at the same time, so it has high convenience. Fusion of lip reading and face information can improve the accuracy of the authentication system, at the same time, greatly enhance the anti-counterfeiting ability of the recognition system. In this paper, Linear Preserving Projection (LPP)is used to extract face features, and Projection Local Spatiotemporal Descriptor (PLSD)is used to extract lip reading features. In Android smart phone platform, face and lip reading are fused by Extreme Learning Machine, which improves the applicability of multi-biometric identity authentication system. The multimode authentication system improves the information security of mobile phone users.