Adaptive Margin Based Liveness Detection for Face Recognition

Face recognition is currently is becoming the hotspot in the area of deep learning, pattern recognition, and computer vision where it has been broadly utilized in many fields. Facial feature extraction is a key link in the face recognition system. The texture features of human faces are highly discriminative, so extracting the texture features of face images can often get a good classification and recognition effect. Image texture feature extraction methods can generally be classified into four categories: statistical methods, model methods, structural methods, and signal processing methods. Recently, face recognition based person authentication systems have been popular among other biometrics. However, hacking methods are also developed with this methodology. In this paper, we present a margin based liveness detection method (MLDM) for the face recognition system based on texture feature analysis. The fake images captured from a video that has edges generated by differences among different face images of real and fake people images. Moreover, we exploit a convolutional neural network to extract these features and differentiate real and fake face images. Experimental results show that our model has higher accuracy and can efficiently classify real faces and spoofed faces compare with the existing model. The outcome shows that our approach is better than the existing work which is experimentally proven.