Research and Application of Face Anti-spoofing Based on Depth Camera
Jie Zhou, Chenyang Ge, Jiaqi Yang, Huimin Yao, Xin Qiao, Pengchao Deng · 2019
With the wide application of face recognition technology in many scenarios, some cases of property loss and privacy leakage caused by face spoofing attacks have occurred, which has aroused extensive discussion on the security and reliability of face recognition technology. Then research on face anti-spoofing has recently gained a well-deserved popularity. In this paper, we use low-cost infrared depth camera to achieve real-time face anti-spoofing. The proposed method (DMF) combines depth information and motion information of human face. An improved RANSAC is used to eliminate face photos and videos spoofing. Then a new blink detection method is used to eliminate 3D print head spoofing. In addition, previous face spoofing databases generally used 2D videos or photos which lacked depth information, we used infrared depth camera to collect a new 3D face spoofing database (3DFSD), which gathered both depth images and infrared images. Abundant comparative experiments on 3DFSD show that DMF can efficiently eliminate face spoofing attacks.