Face Liveness Detection Using Eulerian Video Magnification and SIFT Algorithm

In this paper, face liveness detection method which is based on very short video is proposed. Firstly, spatial and temporal video processing is performed by using Eulerian video magnification. Next, the matching pairs between frames are extracted from the original video. Then, the paper constructs the feature histogram from the color difference between two pixels according to SIFT matching pairs. Finally, we use SVM classifier to determine the liveness of the subject. The experimental results show that the detection accuracy is 95% and the liveness detection can be completed in less than one second without any specific user action.