Improve Anti-spoofing Performance of Face Recognition-based Identity Authentication System by Real-time Facial Expression Recognition

Abstract

Face recognition technology has been applied extensively in many flelds, such as biometric authentication. However, its application is not widely popularized since its anti-spooflng performance is very deflcient. A real-time facial expression recognition algorithm is proposed to realize face liveness detection in this paper, and the anti-spooflng performance can be efiectively enhanced. After the face and the preset facial expression are properly recognized, the algorithm requires the user to change to another randomly designated expression. According to the second facial expression recognition result, it can judge the face image whether is captured from a live face. The four facial expressions, which have high recognition rate and the user is willing to accept and easily represent, are selected to verify the validity of the algorithm. The experiment shows the algorithm can efiectively prevent the spoof attacks using photo, 3D model and video.