Object tracking by color distribution fields with adaptive hierarchical structure

详细信息    查看全文

• 作者：Yawen Wang ; Hongchang Chen ; Shaomei Li ; Jianpeng Zhang ; Chao Gao
• 关键词：Object tracking ; Distribution field ; Appearance modeling ; Simulated annealing
• 刊名：The Visual Computer
• 出版年：2017
• 出版时间：February 2017
• 年：2017
• 卷：33
• 期：2
• 页码：235-247
• 全文大小：
• 刊物类别：Computer Science
• 刊物主题：Computer Graphics; Computer Science, general; Artificial Intelligence (incl. Robotics); Image Processing and Computer Vision;
• 出版者：Springer Berlin Heidelberg
• ISSN：1432-2315
• 卷排序：33

文摘

The essence of visual tracking is to distinguish the target from background, so how to describe the difference between target and background is a key problem. In this paper, tracking algorithm by color distribution fields with adaptive hierarchical structure is presented to solve this problem. First, multichannel color distribution fields are presented for appearance modeling, which represents color distinction between the target and background. Second, in order to adapt to the individuality of each target, the hierarchical structure of its color distribution fields are generated via k-means cluster. Third, weighted multichannel \(L_1 \) distance is used to measure the similarity between the candidate region and the template; the weight of each channel is adjusted online according to its discrimination. Finally, a search strategy based on simulated annealing is proposed to improve the search efficiency and reduce the probability of falling into the local optimum. Experimental results demonstrate that the proposed algorithm outperforms the state-of-the-art tracking algorithms.