Improving Mean Shift Tracking Stability by SGM Feedback

In this paper, a new approach to cope with fixed kernel bandwidth issue of mean shift is proposed. Our contribution lies in two aspects: 1) A theorem is proved that the sum of gradient mode (SGM) of image is monotone decreasing when scale parameter is increasing in Gaussian scale-space. 2) Establishes feedback mechanism between kernel bandwidth of mean shift and SGM. Then kernel bandwidth can be updated according to the SGMs of two successive frames. Experiment results show that the proposed method can satisfy real-time tracking demand, and outperforms the traditional algorithm.