摘要
复杂背景下运动目标跟踪一直是计算机视觉的重要研究方向之一。针对这类问题,提出了一种基于显著性加权的Mean Shift跟踪算法。首先根据目标上不同区域和背景的差别大小,给目标每个区域赋予不同的权值。然后将权值与Mean Shift算法结合起来对目标进行跟踪。实验表明,在复杂背景下,新算法仍然可以有效、准确地跟踪运动目标。
To track the moving target in the cluttering background is one of the most important research fields in computer vision. A new significance weighted mean shift tracking method is proposed. This new tracking algorithm sets the different weights to every parts of the target according to the variance between the target and the background, and tracking the target by the significance weighted mean shift algorithm （SWMS）. Experimental results show that the new method can effectively and accurately track moving target in the cluttering background.
出处
《光学技术》
CAS
CSCD
北大核心
2008年第3期404-407,共4页
Optical Technique
