Abstract: Aiming at the difficulty that face recognition system could be easily deceived by face photos and face videos,a face anti-spoofing detection algorithm were proposed,which uses the fusion color texture features.At present,the main face anti-spoofing detection algorithms are divided into user-matched detection and silent detection.For hot online authentication system nowadays,silent detection has become popular because of its good user experience and accuracy of classification results.Different from the currently popular methods based on brightness characteristics and image quality analysis,the proposed method studies the effectiveness of color features and combined texture features,and then the method combining brightness features,color features and local texture features is proposed.Firstly,the seetaFace algorithm is used to get the coordinates of face and eyes.And then the images which only contain the face are extracted to reduce the interference of background.Secondly,the color information and the brightness information in the image are separated by converting color space and color channel separation.Finally,the method of extracting fusion local texture features is used to extract features from different channels and the feature vectors extracted by each channel are combined and stretched into one-D feature vector,and SVM(Support Vector Machine)is used to train the classifier.The algorithm was performed on the MSU,ASIA,ULU base-line spoofing face database.The experimental results show that the proposed method performs well on improving classification accuracy.

Key words: Color space, Face recognition, Liveness detection, Texture feature