Research On Application Of Camshift And Kalman Filter Algorithm In Video Object Tracking

The video image processing is always an important research direction in the field of computer vision, the video object tracking is one of the most important research contents in video image processing research. At present, with the development of Urbanization, the expansion of the city and the increase of population, the administration of city public security has more and more severe situation and needs more and more people.But the increase is far less than the demand. The construction of city security video surveillance project has been put on the agenda and has been under construction. At the same time, this project is also related to the next stage of the Smart City, that is Safe City. The construction of Safe city in China began in 2004. After 10 years of construction, monitoring point has almost covered every area of people’s life. So what is the most important technology in video monitoring project? It must be the target tracking technologyFirstly,this paper studies some common target detection methods.For example, frame differential method, background subtraction method and optical flow method. Due to the limitation of the optical flow method, this paper don’t introduce it too much. But the frame differential method and background difference method have carried on the contrast experiment of programming.And this paper made a thorough research on Vi Be background subtraction method and hybrid Gauss method. Vi Be algorithm is used to detect target, hybrid Gauss method is used to detect and remove shadow. Then the two algorithms are combined to improve the accuracy of target detection.Secondly, it analyzed the advantages and disadvantages of continuous adaptive mean shift algorithm-it is usually called Cam Shift. At the same time it studies the prediction effect of Kalman filtering algorithm; But When the background is complex,or there is a lot color similar pixel interference, Camshift may lead to tracking failure. Aiming at these problems, this paper proposed a algorithm that combinating Camshift algorithm and Kalman filter, at the same time it limits the size of Search window, then it can improve the tracking accuracy effectively.In this paper, it usese Qt Creator5.1.1 tools environment in Linux system, combining the OPENCV library and usese C++ language to realize the simulation of the proposed algorithm. Through the verification of programming and experiment, this algorithm can achieve stable tracking of single target, and it can realize the goal of the correct tracking of multiple targets.