Research On Algorithm Moving Object Detection And Tracking In Intelligent Video Surveillance

With the rapid development of computer application technology,Internet of things technology,and video image processing technology,intelligent video surveillance systems have also been more widely used..Intelligent video surveillance is the automatic detection and feature extraction of the target through the computer system.Based on this,it completes the positioning and tracking of the target.Detection and tracking of moving targets is the most critical technology in intelligent video surveillance,and it is also the basis for advanced processing such as target classification,behavior understanding,and event detection.The specific research content of this paper is as follows:1.An Improved Target Detection Algorithm Based on Fusion of Three-Frame Difference Method and Background Difference Method.First of all,in order to maintain a good detection effect of video surveillance,this paper analyzes the basic theory of optical flow method,background difference method,inter-frame difference method and detection experiment,and compares the advantages and disadvantages of these three methods,and gives three An improved algorithm that combines the frame difference method and the background difference method detects moving targets in video surveillance.The fusion algorithm makes full use of the three-frame difference method's high real-time performance,strong adaptability to the environment,and background difference method to detect the advantages of the complete contour of the moving target.The adaptive frame update is performed using the results obtained by the three-frame difference method.2.An Improved Target Tracking Algorithm Based on Mean Shift.For the Mean shift algorithm,when the moving target encounters occlusion,there is a target-to-lose situation.An improved tracking algorithm based on the target position feature Mean shift is given.The position of the next frame of the moving target is predicted by the position of the moving target,and the target position predicted by the position feature of the moving target at any time and the target position obtained by the Mean shift algorithm are comprehensively considered,and finally the actual position of the moving target is obtained3.Multi-feature Particle Filter and Improved Mean shift Fusion Algorithm in Complex Environments.Under a complex background,the single tracking algorithm can not meet the target's accurate tracking under a series of situations such as the speed of the target,the environment change,and the obstruction of the object.In order to solve the above problems,this paper presents multi-feature particle filtering and improved Mean shift fusion algorithm.Based on particle filtering,the algorithm uses colors and edge features as the observation model to express the target.The Mean shift algorithm clusters the particles that are closer to the true posterior distribution,calculates the new particle weight,and uses the target position feature.Based on the position,speed,acceleration,etc.of the target at the previous time,the possible position of the target at the next time is predicted,and the tracking error is caused by reducing the target too quickly or blocking.The experimental results show that the proposedalgorithm is more robust than the color,edge and multi-feature particle filter tracking algorithms,and the average weight of the particles is improved,which enables stable tracking of targets in a complex environment.