Improved mean shift algorithm for moving object tracking

Abstract

Mean shift is a nonparametric estimator of density gradient. Traditional mean shift algorithm is rather sensitive to the influence of background. Therefore, an improved mean shift object tracking algorithm is proposed. Firstly, a novel weights given method is given, which improves the kernel function. The method is that the pixels which near the centre of object are given biggest weights, and the pixels which at the edge of the object are given by exponent distribution as a result of occlusion. In order to hand the occlusion, the occlusion detecting method based on sub-block detecting is also established. The novel sub-block detecting algorithm is that the tracking window is divided into two parts, including right and left, and the similarity measure is calculated respectively. The simulation results show that the improved mean shift algorithm can hand the occlusion effectively and track the moving object very well, and it can track moving object more powerful than the basic mean shift tracked.