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 target¡¦s 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 (¥þ¤åÀÉ) 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.