A novel visual tracking algorithm based on random sampling and color gradient

In this paper, we propose a novel robust vehicle tracking method based on random sampling. Firstly, random sampling and prediction to the random sampling points are performed. Secondly, color gradient feature of object contour combined with template match is used for weighting. Finally the centroid of target is located through summing to the weighted random sampling points. This method needn't too many restrictions, for example a certain hypothesis or requirement to background et al. With the calculating to object contour, too much time is saved. Then, real-time and accurate tracking is achieved in this paper. Experimental results show an important performance gain and better tracking accuracy in favor of our approach.