Research On Moving Object Tracking Based On Particle Filter

Video target tracking is an important research direction in the field of computer vision fields, with a wide range of applications in military guidance, human-computer interaction, and robot visual navigation. Video target tracking refers to detecting, extraction, identification and tracking for the moving object in the video sequence, thus providing foundation for further video analysis and understanding. Although it has been proposed a number of effective video object tracking algorithm, but in practical applications, due to the target block, background change, the shadow interference and other factors, resulting in the effect of target tracking is less than ideal. Design suitable for any complex environment of the target tracking algorithm is still a challenging task.In terms of object detection, some common approaches in the field of object detecting are introduced. Focused on the frame difference method and the background difference method, and analyzed respective advantages and disadvantages through experimental results. Based on the analysis, a linear prediction of background updating algorithm is proposed by integrated use of the advantages of the frame-difference and the background-difference method, which made background can adaptive updates. The algorithm which improved the robustness of the background update can be accurate and reliable detection of the moving target.For the target tracking problem, firstly, a particle filter object tracking approach is proposed based on the new histograms by the fusion of histograms of oriented grads and color. The approach that can be more accurate characterization of the target can improve the accuracy and robustness of the video target tracking. Secondly, this article has conducted in-depth study of camshaft tracker, and verified its advantages and disadvantages through the experiment. Finally, this paper proposed a camshaft optimized particle filter tracking algorithm. The algorithm that combined the advantages of particle filter algorithm and Camshaft algorithm to a certain extent satisfy the requirements of real-time and accuracy. The algorithm also designed the discriminate function of similar color interference and target loss. When the target is in both cases, the corresponding measures to achieve the purposed of accurate tracking, that improve the robustness of the algorithm to a certain extent.Based on the above algorithm research, a video target tracking experiment platform is constructed. This article describes the hardware components and functions of the platform. The tracking test of video filer and real-time acquisition of image sequences by camera is accomplish by the platform.