Moving Target Tracking Method Based On Timed Motion History Image

With the development of computer vision theory,great progress has been made on the moving target analysis techniques.Among these techniques,the moving target tracking technique has great significance in computer vision research.It has been widely used in the field of video surveillance,visual navigation,human-computer interaction and so on.In the domestic and overseas,many researchers have conducted abundant researches on moving target tracking,and have proposed some effective methods.However,it is still a challenge to design a method which is able to ensure timeliness,accuracy,and robustness for moving target tracking under complex scenarios.In this article,researches are concentrated on the moving target tracking under complex visual environment based on the target's history moving information.The main work of this article is as follows:Firstly,the research status and development trend of moving target tracking are introduced.And the relevant theoretical basis of moving target tracking is expounded and analyzed.Secondly,since the traditional tracking algorithms are prone to drift even failed when the target's appearance is changed or targets under complex visual environment such as occlusion and illumination change.A real-time tracking by double templates matching based on timed motion history image with HSV feature method is proposed to sovle these problems.The target's timed motion history image(tMHI)information is obtained by u sing the motion history image segmentation method to segment the moving target in the image.This method analyzes and deals with multi-frame images,so it is more robust.Besides,the HSV color feature space is utilized to describe the target's appearance model with tMHI.Considering the target's appearance may change greatly during tracking,a double templates matching strategy is designed to locate the target.Different with the traditional methods which use the initial frame target as the target template,the online template is updated timely and the original template is kept and updated as needed to ensure accuracy and robustness of tracking.Finally,a double templates matching(DTM)tracking method,with the target described by the edge direction and the color feature based on the tMHI is proposed.This method utilizes the color histogram and edge direction histogram together to describe the target,and assign weights according to the similarit ies of the two features adaptively,and locate the target by calculating the fusion similatities.Experimental analysises demonstrate the effectiveness of the multi-feature representation of the target,especially in the case of some blurred background scenarios.T he proposed tracking method performs better than the original single color feature based method.