Research On Target Tracking Algorithm Based On Spatio-Temporal Context

In recent years,along with the rapid growth of science and technology,human society has become increasingly intelligent.Intelligent video surveillance system has been gradually applied in real life,which has brought a lot of convenience for human life,and has high value for use.As the key technology of the intelligent monitoring system,the target tracking technology has been paid more and more attention by the international experts.At the same time,a large number of target tracking algorithms have appeared.However,the current tracking algorithm has the problem of inaccurate tracking results during the light intensity changes,camera shaking,object size changes,occlusion and so on.Resulting in the practical application of the feasibility is not high.Therefore,it is necessary to study a highly robust target tracking algorithm that makes it possible to be applied in real life.In this paper,several common algorithms in the field of target tracking is described,focuses on the analysis of the spatio-temporal tracking algorithm.Then,two improved tracking algorithms are proposed to deal with the problem that the tracking results are not accurate when there is camera jitter or large range of occlusion.Firstly,the spatio-temporal context tracking algorithm is used to track,and then the similarity measure function based on the perceptual hash algorithm is used to determine the result of the tracking.When the video is jitter leading to the failure of the judgment result,the inclusion of the Mean Shift tracking algorithm to complete the tracking,so as to improve the accuracy and stability of the tracking.When the video target has a large range of occlusion lead to tracking failure,the least squares fitting algorithm and spatio-temporal context algorithm combined to achieve accurate tracking.Finally,the improved algorithm of this paper is simulated and verified.The results show that the improved algorithm can solve the above two problems well,and on the basis of preserving the advantages of the original algorithm,it can effectively improve the tracking accuracy,has a certain degree of robustness.