Research On Algorithms Of Moving Target Detection And Location Based On Video

With the rapid economic development and social progress of China,more and more people are beginning to pay attention to their own security.Intelligent video surveillance systems have emerged and are widely used in transportation,prisons,airports,power systems and other places where safety performance is high.However,in the specific implementation process of the traditional video surveillance system,it is often necessary to rely on manpower to complete,resulting in a lot of unnecessary waste of resources.Therefore,the realization of the efficient moving target detection and localization algorithms are very practical.In this paper,the moving targets in the video sequence are taken as the research object,and the target detection,target tracking and target intrusion detection algorithms are analyzed.Aiming at the problems in traditional algorithms,the background subtraction method based on mixed Gaussian model and the traditional Camshift tracking algorithm are improved.The main work of this paper includes the following three points:In the aspect of moving target detection,the traditional three-frame difference method can only extract the approximate contour of the target.The specific implementation of the background subtraction method depends on the accurate establishment and real-time updating of the background model.The Canny operator was introduced to improve the traditional three-frame difference method,and the improved three-frame difference method is combined with the background subtraction method based on the mixed Gaussian model,and then the HSV color space is used to remove the shadow area in the obtained target.Finally the moving target detection was realized.In terms of moving target tracking,the traditional Camshift algorithm is susceptible to similar color interference in the background.The saturation component is introduced to improve the target histogram template in the algorithm to obtain more accurate target color information.Aiming at the problem that partial occlusion of the target will lead to tracking failure and that the algorithm needs to find the target area in the whole picture in each frame of the image in the specific implementation process,so that the calculation amount is too large.Kalman filter is added to predict and estimate the position of the next frame of the target,so as to quickly and accurately obtain the target information in the next frame of the image.In the aspect of target intrusion detection,for the false alarm phenomenon caused by the target on the warning line,the positional relationship between the warning line and the projection line is used to judge whether the target crosses thewarning line,and the projection line is obtained by projecting the lower edge of the target frame downward to the intersection with the ground.Experimental results show that compared with traditional algorithms,the improved target detection algorithm can extract the moving target completely and accurately in the complex environment.The tracking algorithm can improve tracking accuracy and reduce computational complexity.The target intrusion detection algorithm can effectively reduce false alarms caused by the angle of view.