MATV-feature-based face living-body detection method

Abstract

The invention relates to an MATV-feature-based face living-body detection method. The method comprises: a training set video is read and decomposed, a face region of each frame is extracted and is processed by normalization for inputting, a magnitude-angle-total-variation (MATV) optical flow filed feature based on a magnitude-angle variational method is employed, the feature is coded by using a statistical histogram to obtain an MATV histogram feature, and then the MATV histogram feature is inputted into an SVM classifier to carry out training and an SVM model is established; and for a face video frame sequence of a test set, face detection is carried out on each frame, the face region of each frame is extracted and normalization is carried out to obtain a test set face frame sample, an MATV histogram feature of the face frame sample is calculated and then is inputted into a trained SVM model, and a face living-body detection result is predicted. According to the invention, on the basis of the traditional magnitude-feature-based optical flow method, the angle feature is added and the direction characteristic of the optical flow field is considered, thereby obtaining more living-body and non-living-body face motion details. Therefore, the accuracy of the face living-body detection is improved.

Description

technical field [0001] The invention relates to the research field of human face living body detection, in particular to a human face living body detection method based on MATV (Magnitude-AngleTotal Variation) features. Background technique [0002] With the popularization of automatic face recognition technology, it brings a series of potential security threats. Criminals can gain benefits by attacking and deceiving the automatic face recognition system. There are two common forms of attack, one is to print the legitimate user's face photo to attack, and the other is to record the legitimate user's face dynamic video through the playback medium to attack. If the criminals succeed in their attempt, the consequences will be disastrous. [0003] In order to deal with this potential spoofing attack, face liveness detection technology came into being. The current live detection technology is mainly divided into two categories: one is based on image texture features, such as L...

AI Technical Summary

Problems solved by technology

However, the traditional optical flow method is not good enough to estimate the optical flow of the motion information of living and non-living faces, so it cannot distinguish living and non-living faces well. At the same time, these methods often only consider the amplitude characteristics of optical flow and ignore Directional characteristics of optical flow field

Therefore, traditional optical flow methods do not achieve very effective liveness detection performance.

Images