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