Object Tracking Based On Particle Filter And Multiple Features

Study of image tracking is a hot issue in computer vision and image processing. It is widely used in video surveillance, intelligent household, interpersonal interaction, and other fields. How to track object accurately and quickly in video is the important research content in the field of computer vision. This paper studies object tracking based on particle filter and multiple features, the main research contributions are listed as follow:Firstly, a novel target tracking algorithm for target’s localization with Mean shift based on multiple features and least squares method prediction was proposed to solve background interference, problems of partial occlusion and other issues. Aimed at the problems of background interference, the improved Gaussian mixture model(IGMM) was proposed to build background model and to extracted foreground moving object. The motion cue, color cue and texture cue were fused to represent the target, according to the change of the target and backgrounds, combined the motion feature to remove noise interference of background and to locate the object. In the occlusion process, when the Mean-Shift integrating multi-feature failed to track the target, it would use least squares method to predict the location of the target. Experimental results demonstrate that the method can track the target accurately and have better Robustness in complex backgrounds or occlusion situations.Secondly, in order to solve the problem of complex environmental impact and serious occlusion during the object tracking in the sequence images, a hybrid particle filter tracking method based on the global and local information was proposed. The block texture histogram was imported into the traditional particle filter algorithm and made the algorithm be able to improve the robustness of the tracking algorithm which included the spatial information of target. The adaptability and occlusion resistance of the tracking algorithm was improved since adjusting adaptively the contribution of the global and local information by the degree of the object occlusion. Experimental results show that the proposed tracking algorithm exhibited good result in the presence of partial occlusion and serious occlusion.Finally, considering the two methods can solve the impact of complex environment and target occlusion, but they couldn’t solve the influence of target appearance changes. So in order to solve the problem of complex environmental impact like illumination variation,appearance change and partial occlusion during the object tracking in the sequence images, a hybrid particle filter tracking method based on the global and local information was proposed.The LBP textual feature was imported into the particle filter algorithm which uses local information of the target via sparse coding on local patches and combines the global information to determine the tracking object. In the procedure, the robustness of the tracking algorithm was improved since the template is updated on the time. Experimental results showthat the proposed tracking algorithm exhibited good result in the presence of complex background and partial occlusion.Through the proposed three kinds of algorithms for target tracking in image sequence, it can overcome the deformation, target appearance change, complex environmental impact and occlusion, so it can realize the robustness tracking.