Research On Moving Object Tracking Algorithm Based On Mean Shift

With the rapid development of computer technology,video image processing and computational ability have been greatly improved,and computer vision technology has also made rapid development.Visual object tracking technology is a key problem in computer vision field.Although experts and scholars have studied for a long time and achieved fruitful research results,it is still faced with various challenges to track targets due to interference factors such as occlusion,illumination changes,rotation,etc.How to design efficient and robust object tracking algorithm to deal with all sorts of challenges is still an urgent problem to be solved.This paper mainly studies the target tracking algorithm based on MeanShift,which is widely used in the field of image processing.It is one of the basic algorithms of target tracking.MeanShift algorithm is semi automatic algorithm,need to manually select the target in the initial frame.Then,the pixel values of the target region are calculated,and the probability density of the subsequent frames iteration search is compared with the target template.When the selected similarity coefficients satisfy the threshold or the number of iterations reaches the upper limit,the position of the tracking target is determined.MeanShift algorithm has many advantages,such as low complexity,small amount of computation,good real-time performance,and not easy to be influenced by illumination,but the application of the algorithm is limited because of the fixed and lack of the necessary template update and prediction mechanism.This paper is based on one of the shortcomings of MeanShift algorithm that tracking window fixed to improve it.Tracking window fixed bring two problems: the target information will be lost when the target becoming big,meanwhile the target will be mixed with too much background information when it becoming small.Because the algorithm itself only relies on the pixel value of the target image as the feature,so thatboth lossing of information and mixed with the background information will be bring heavily influence to the result.On the basis of classical MeanShift,the log polar coordinate transformation of the target template and the candidate template is added,and the deformation of the object is translated into the translation on the polar coordinate axis.What's more,the Kalman filtering algorithm is added to MeanShift to predicate the centroid of the target,and the spatial location of the algorithm is more accurate.By comparing with the classical scaling adaptive algorithm and Camshift algorithm,the experimental results show that the proposed algorithm can effectively improve the problem of the deformation of the target,and the algorithm is robust.