An Adaptive Tracking Algorithm Based on Mean Shift

Abstract:

Object tracking is to search the most similar parts to targets in video sequences. Among the various tracking algorithms, mean shift tracking algorithm has become popular due to its simplicity, efficiency and good performance. This paper focused on mean shift tracking algorithm, which is a modeling mechanism based on statistical probability density function. In practice, when the background of the tracking and characteristics of the target are similar, pixels of background occupy a large proportion in the histogram. The traditional mean shift cannot adapt to the mutative scene. Meanwhile, if there is block or disappearance, the result is not exact. Three algorithms were given for above difficulties. A weighted template background was established, that can highlight the features of target and improve real-time. Then this paper presented a selective mechanism to update the target model. Every component is updated based on the contribution to the target model. Finally, the Kalman filter was combined with mean shift algorithm. We saw the prediction points of Kalman filter as the initial point, carried out the mean shift iteration and then updated Kalman filter using the ultimate location. Extensive experimental results illustrated excellent agreement with these methods.

• Periodical: Advanced Materials Research (Volumes 538-541)
• Pages: 2607-2613
• DOI: https://doi.org/10.4028/www.scientific.net/AMR.538-541.2607
• Online since: June 2012
• Authors: Zheng Hong Deng, Ting Ting Li, Ting Ting Zhang
• Keywords: Kalman Filter (KF), Mean Shift Algorithm, Target Model, Tracking Algorithm

References

[1] Y.Z. Cheng: IEEE Trans Pattern Analysis and Machine Intelligence, vol. 17 (1995) No.8, p.790.

[2] D Comaniciu and P Meer: IEEE Trans Pattern Analysis and Machine Intelligence, vol. 25 (2003) No.5, p.564.

[3] D. Comaniciu., V Ramesh, and P Meer: 2000.Proceedings.IEEE Conference on Computer Vision and Pattern Recognition (Hilton Head Island, SC, USA,june 13-15,2000).vol.2,p.142.

[4] D. Comaniciu and P .Meer: The Proceeding of the Seventh IEEE International Conference on Computer Vision (Kerkyra , Greece,september 20-27,1999).vol.2, p.1197.

[5] F Aherne,N Thacker,and P Rockett: Kybernetika, vol.34(1998) No.4, p.363.

[6] T Kalath:IEEE Transactions on Communication Technology,vol. 15 (1967) No.1.,p.52.

[7] R.E. Kalman:Journal Of Basic Engineering,vol 82 (1960) Series D,p.35.

[8] Fred Daum:Nonlinear filters:IEEE A&E Systems Magazine,vol 20 (2005) No.8,p.57.