An improved mean shift algorithm for object tracking

MeanShift algorithm is a popular method for searching for local extreme value in the density distribution of a set of data. Traditional MeanShift object tracking algorithm mainly uses a single histogram to describe the color characteristics of an object, and the detection precision and stability are not good enough in a complex background due to its lacking of spatial information of pixel colors. As for this defect, this paper presents a new method combined with distribution information of space to reduce the effect of image flaws by setting a weight to pixels, on the basis of the distance from the center point of target to the current point. The experiment results show that our method promotes the tracking accuracy of moving object under a complicated environment and has better stability.