The Research Of Moving Objects Tracking Algorithm Based On The Color Information

With the development of hardware and software of computer, computer vision technology is attracted more and more attention by people. As the important projects of computer vision, Moving Objects Tracking has been widely applied in military affairs, aviation spaceflight, CAD, Intelligent Robot and so on. The stability and accuracy of tracking effects depend on the design of algorithm to a great extent. The research of the dissertation involves region extraction, moving target detection, tracking and so on.First, whether we can accurately detect motion objects from the video sequences is directly relation to the next processes, for example Motion Estimation, Target Recognition, Action Understanding and so on. To achieve the effective detection of moving targets, many scholars have doing interrelated investigation and provide a lot of algorithm. Firstly the thesis analysis and comparison of the Time Difference Method, Background Modeling Method and the Optical flow, points out its advantages, disadvantages and application; secondly, study a segmentation algorithms for moving objects which is based on the Optical Flow and Wavelet Fusion; thirdly, the experimental result proves the stability and real-time of the algorithm.On the research of motion tracking, the paper analysis the main algorithm of object tracking, the CamShift algorithm mainly discussed in this paper. When the object is moving, the color characteristics will not significant change like the shape characteristics, such as rotation or zoom, etc. CamShift algorithm is a tracking method based on the object's color information, because of the parameter-less and fast calculation the algorithm is widely used in many domain. In this paper, add the tracking template of Optical Flow into the CamShift. on the one hand, the algorithm combines of motion feature and color feature so it could solve the failure cause by illumination changes or feature points have been blocked which is the disadvantages of Optical Flow; on the other hand, when the color of background is similar to the target, the tracking will fail with the method of CamShift, the new algorithm will solve this problem, because the Optical Flow could reduce correlation between the target and background. The final experiment test and verify the validity and real-time of the algorithm.