Previous 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

@

Journal of Information Science and Engineering, Vol. 24 No. 3, pp. 673-690 (May 2008)

Dynamic Visual Tracking Using SDG-Like Matching

Ming-Yang Cheng, Mi-Ching Tsai1 and Chun-Jen Chen2
Department of Electrical Engineering
1Department of Mechanical Engineering
National Cheng Kung University
Tainan, 701 Taiwan
2Optical Storage BU
MediaTek Inc.
Hsinchu, 300 Taiwan

Motion detection/estimation plays a crucial role in dynamic visual tracking. Whether a dynamic visual tracking system can successfully track a moving target closely depends on the quality of motion detection/estimation results. In dynamic visual tracking, the camera used to capture images is not stationary, so any slight vibration of the camera motion or the target motion can lead to a blurry image causing the visual tracking performance to be deteriorated. To cope with this difficulty, a motion detection/estimation approach consisting of a region-based spatial distribution of Gaussians (SDG)-like matching algorithm, a template-update-with-memory algorithm, and a template mask is developed in this study. Moreover, linear interpolation on vision commands is performed to improve the tracking performance. A dynamic visual tracking system designed for locking the targets image in the center of the image plane is used as the test platform. Experimental results demonstrate the effectiveness of the proposed approach.

Keywords: dynamic visual tracking, region-based matching, spatial distribution of Gaussians, motion detection/estimation, template mask

Full Text () Retrieve PDF document (200805_02.pdf)

Received March 9, 2006; revised June 20, August 25 & October 27, 2006; accepted November 8, 2006.
Communicated by Jenq-Neng Hwang.
*This paper was partially supported by the National Science Council of Taiwan, R.O.C., under grant No. NSC 91-2213-E-006-122.