Design And Implementation Of Face Liveness Detection System Based On Movement Information

With the rise of artificial intelligence,biological feature recognition method,as one of the most popular fields of artificial intelligence,certification system has been widely used in all walks of life,biometric identification methods including fingerprint authentication,iris certification,face authentication and certification of the gait,etc.Among them,face recognition is the most popular method of biometric identification,because face recognition verification is more direct,friendly and convenient than other methods.But the common face recognition system has one drawback,which is easy to attack,such as photo attack;In order to solve this problem,human face detection technology emerged,and gradually became the research focus of the researchers.In order to solve the problem of face authentication system,this paper designs and implements the face detection system based on movement information.The movement information refers to shaking your head,blinking and opening your mouth.These three movements are the most representative facial movement information.The main function of this system is to randomly generate two sequences of actions.The user can complete the action command according to the prompt,and both actions are passed,and the user can verify the system through the face living detection system.The system firstly carries out the face detection,determines whether the face enters the detection area,when has the face,starts the detection process;At the beginning of the detection,the scale normalization and histogram equalization of the extracted video frames were first pretreated.Then the face is detected and the key points are extracted and the regional information of the test site is extracted according to the key information.Testing for the action,the blink of an eye and mouth detection eventually converted to binary classification problems,using the trained SVM classification model for binary classification,shook his head test is the use of 3 d model of triangular projection to two-dimensional plane,the area ratio will change the characteristics of the designed algorithm implementation.Finally,the system is tested by different test cases,and the test results indicate that the system meets the requirements of face detection and face attack in both functional and accuracy...