Research On Video Object Tracking Based On Correlation Filter In Complex Scenes

Object tracking has always been a hotspot in the field of computer vision,involving many disciplines and theories.And the object tracking algorithm based on correlation filtering has attracted the attention of many scholars because of the high real-time performance and high tracking accuracy.However,when disturbed by illumination changes,object occlusion,object deformation and other issues,the performance of correlation filter tracking algorithm is not robust enough.Therefore,in order to improve the tracking performance of the correlation filter tracking algorithm in complex scenes,a thorough study is carried out,and the results are as follows:Aiming at tracking drift caused by complex illumination interference and object scale change,a scale-adaptive correlation filter tracking algorithm based on multiple features was proposed.Firstly,in the position prediction stage,the characteristics of Hue feature and histogram of oriented gradient are analyzed,and the object location is predicted by distributing the weight of each translation filter output response values of the two features.Then,in the scale prediction stage,a scale filter is trained independently by the multi-scale image sampling in the object position,and the scale of the object is estimated according to the scale filter response value of the sample,which made the tracking algorithm adapt to the scale change of the object.Finally,the difference between two frames was used to adjust the learning rate adaptively to update the model of the translation filter.The experimental results show that the tracking accuracy of the improved algorithm is higher than that of other classical algorithms.When the object scale changes greatly,the improved algorithm can still track stably.The boundary effect will lead to the inadequate robustness of the tracking algorithm to the interference of occlusion and fast motion.To solve this problem,a correlation filtering tracking algorithm based on sample enhancement and alternating direction multiplier method is proposed.Firstly,the influence of boundary effect is weakened by increasing the sampling frame of cyclically shifted samples and cutting them.Then the alternating direction multiplier method is introduced to optimize the model solution,and Sherman Morrison formula is used to further optimize the iterative solution of the model.Finally,an occlusion decision mechanism is set up to determine whether the model is updated or not,so as to improve the anti-occlusion ability.The experimental results show that,compared with the original correlation filtering tracking algorithm,the improved algorithm is still reliable in occlusion and fast motion scenarios,and the robustness of the algorithm is improved.