Research And Realization Of Face Recognition System Based On Eye Tracking In Vitro Detection

With the development of biometric information extraction and recognition,face recognition technology has also been widely used.But with the popularity of face recognition technology,a variety of attacks against the technology are also evolving,and it brings challenges to the development of identity authentication system,especially the face recognition technology.Therefore,the development of a technology which can resist this kind of forged attack has become a hotspot in the field of information security.Among them,recognition of the living face has a very important practical significance to the application of face recognition technology.In this paper,the current living body detection technology will be briefly introduced and analyzed,so will the eye tracking technology.According to the method of the living body detection which is based on eye tracking,the judgment mechanism of eyeball tracking for distinguishing the detection of human face and photograph forged face will also be explained,and how the location algorithm of eyeball centering in eyeball tracing is improved;the time spent in detection positioning and robustness under different lighting conditions are analyzed.In addition,the SIFT-based face recognition technology is introduced.Finally,a live face detection system and a face recognition system are built to realize the integrated system of living body detection and face recognition.Aiming at the problem of low efficiency in positioning the center of the eye,this paper introduces a method to locate the center of the eyeball based on the combination of regional projection and gravity method.For the human eye detection,human face detection of the image is used to improve the correct rate of human eye detection and reduce the impact of image noise;in the center point positioning,the regional projection is firstly used to conduct the initial positioning,and then the gravity method is adopted to conduct precise positioning of the initial positioning point and the surrounding area.Compared with this method and simply using the center of gravity method or area enhancement method,it used with 10 character face images;each image runs 10 times in each method and the run time is faster than the other two methods.For instance,one portrait image,the detection time is 1.546 seconds through this method;and the detection time through gravity method and the areaenhancement method are 1.706 seconds and 1.681 seconds respectively.Moreover,under different lighting conditions,the method is still able to correctly detect,with good robustness.Aiming at how to use the eye movement to determine the living body,this paper gives a method based on relative displacement.Take the center point of the detected human face image as the reference point to judge the relative distance of the center point of the two eyes to the reference point.If the difference between the relative distance before and after the movement is within the range of the set threshold,the system considers that the lens is a living face;otherwise,the system thinks that the front of the camera is a fake photo face.At the same time,the detection effect of this method is carried out according to the experiment,the experimental samples are 100;the detection efficiency of forged photo face is 1% and the detection efficiency of living face is 100%,indicating that the method is feasible to a certain extent.The system designed in this paper uses VS and OpenCV framework to build a living detection system and face recognition system.The integrated system is mainly divided into two large modules: judgment and recognition of living human face,including the functions of positioning of eye ball center points of the static image,judgment of human face and five senses.In this paper,the various functional modules of the system are tested and the effect is good.