A Study Of Target Tracking Algorithm Based On Emplate Updating

In recent years,with the rapid development of computer technology and the improvement of people's safety awareness,visual tracking algorithm has drawn more and more attention from researchers and gradually become a hot research topic.A lot of new tracking algorithms and important theory have been proposed.However,the visual target tracking will be affected by background changes,light changes,target rotation,occlusion and target appearance deformation,resulting in the overall update of the target template is not enough,so that tracking becomes very difficult,while there are still many deficiencies in the theory,the application is also very scarce and simple.In this paper,the template representation ability of the mean shift tracking algorithm is insufficient and the template updating is insufficient in the complex environment.The mean shift algorithm of multi-feature fusion is proposed.The target histogram model is built by using the target color features and texture features.By using the temporal and spatial characteristics of feature information,selective sub-feature classification and updating are improved,which improves the traditional global updating strategy.Tracking caused by over-updating is inaccurate,improving the accuracy and robustness of the tracking algorithm.Finally,the algorithm is applied to the video surveillance system to achieve the goal of tracking and monitoring.Based on the research of visual tracking algorithm,this paper focuses on the mean shift target tracking algorithm.The main contents and contribution of the paper are as follows:(1)Compared with the popular target tracking algorithm,combining with the color texture features,Feature Blending Mean Shift Algorithm.(2)According to the different target application scenarios,in order to solve the target can be correctly tracked in a variety of complex scenarios to classify the target features as a whole.(3)Analyze the relationship between the target features in time and space,adjust and update the target template accordingly,and realize the algorithm of mean shift based on the template update.(4)Realize a video surveillance system based on target tracking algorithm.