Mean shift tracking combining SIFT

Abstract

A novel visual tracking algorithm to cope with occlusion and scale variation is proposed. This method combines mean shift and SIFT algorithm to track object. SIFT algorithm is invariant to rotation, translation and scale variation. But it is a time-consuming algorithm. The wasting time is related to image size. So the proposed algorithm first adopts mean shift to initially locate object position, then SIFT operator is used to detect features in object area and model area, lastly, the proposed method matches features in these two areas and calculates the relationship between them using affine transform. According to affine transform parameters, the state of object can be adjusted in time. In order to reduce process time, an improved feature matching algorithm is proposed in this paper. Experiments show that the proposed algorithm deals with occlusion successfully and can adjust object size in time.