Research And Implementation Of Target Person Tracking System Based On Video Surveillance

With the development of human civilization,as an important part of social public security protection,video surveillance has been paid more and more attention by researchers,especially in recent years,the continuous breakthrough of artificial intelligence technology,and now the society has put forward higher requirements for intelligent video surveillance.The research and application of face analysis technology in the field of computer vision and pattern recognition has been very successful,such as the face clock in system,mobile face unlocking and intelligent access control system,which have been widely used at present.However,the application of face analysis technology in video surveillance is still not mature.Considering its great significance for the criminal investigation of the public security department and the maintenance of social public order,combined with my own work,this paper studies face analysis technology and its application in video surveillance.The main work is as follows:Based on AdaBoost face detection algorithm,this paper studies and designs an improved AdaBoost face detection algorithm which integrates skin color segmentation technology.This algorithm improves the accuracy and speed of face detection.When AdaBoost algorithm trains the classifier,it may lead to the degradation of the overall performance of the integrated strong classifier due to the distortion of the sample weight distribution caused by the excessive noise sample weight.In this paper,the weight updating method of AdaBoost algorithm is improved.At the same time,in view of the slow detection speed of the trained AdaBoost classifier,this paper proposes to combine the skin color segmentation technology with the improved AdaBoost algorithm,use the skin color segmentation technology to pre select the face area,and then use the improved AdaBoost classifier to detect the face.Based on the CamShift face tracking algorithm,this paper studies and designs an improved CamShift face tracking algorithm,which improves the accuracy of face tracking.CamShift face tracking algorithm is easily affected by the background color when tracking,and when the target is seriously occluded,CamShift algorithm can not deal with it,resulting in tracking failure.In view of the above problems,this paper introduces the tracking drift coefficient to evaluate the search window of CamShift algorithm when tracking.When the above problems occur,combined with skin color segmentation technology and LBP feature histogram to relocate the face,the continuous and stable tracking of the face is realized.Finally,based on the improved face detection and face tracking algorithm,combined with face recognition technology,it is applied to the actual video surveillance,design and implement a target person tracking system based on video surveillance.The system has the functions of face detection,face recognition and face tracking,which can detect and recognize the faces in the video surveillance,and can track the recognized target face continuously and stably.