摘要
视觉跟踪问题是当前计算机视觉领域中的热点问题,本文对这一问题进行了详细的介绍．首先,对视觉跟踪技术在视频监视、图像压缩和三维重构等三个主要方面的应用进行了论述．其次,详细阐述了该技术的研究现状,介绍了其中的一些常用方法,为清楚说明这些方法,先对视觉跟踪问题进行了分类,然后介绍了处理视觉跟踪问题的两种思路即自底向上和自顶向下的思路,最后将具体的视觉跟踪方法分为四类进行了介绍,这四类分别是基于区域的跟踪、基于特征的跟踪、基于变形模板的跟踪和基于模型的跟踪．最后,从控制论角度给出视觉跟踪算法所面临的难点,即算法要满足鲁棒性、准确性和快速性要求时所遇到的困难,并对视觉跟踪问题的研究前景进行了展望．
This paper introduces in detail the research of visual tracking which is a hot spot currently in the domain of computer vision. Firstly, the applications of visual tracking in three areas including visual surveillance, image compression, and 3-D reconstruction are discussed. Secondly, the state of the art about visual tracking is introduced, especially the main approaches of visual tracking are shown. In order to explain these methods clearly, the problems of visual tracking are classified. Then two ways to research the visual tracking problem are presented. And the visual tracking algorithms are classified into four classes： the area-based methods, feature-based methods, deformable-template-based methods and model-based methods. Finally, from the point of view of control theory, the difficulties of visual tracking are discussed that the algorithms should have robustness accuracy and be fast. Meanwhile, some future directions of visual tracking are also addressed shortly.
出处
《自动化学报》
EI
CSCD
北大核心
2006年第4期603-617,共15页
Acta Automatica Sinica
基金
国家973项目(2001CB309403)
国家自然科学基金(60574033)资助~~
