CASIA OpenIR  > 毕业生  > 硕士学位论文
视觉单目标跟踪算法研究
Alternative TitleResearch on Visual Single Object Tracking Algorithm
彭曦
Subtype工学硕士
Thesis Advisor唐明
2011-05-23
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline模式识别与智能系统
Keyword目标跟踪 分类 K近邻 排序队列 多尺度 Object Tracking Classification K-nn Ranking List Multi-scale
Abstract随着计算机视觉及相关领域的发展,视觉目标跟踪展现出广泛的应用前景,因此视觉目标跟踪逐渐成为计算机视觉研究的热点之一。本文在对已有的方法以及相关技术进行分析和总结的基础上,提出基于判别性排序队列的视觉单目标跟踪方法。本文贡献主要有如下四个方面: 1. 对近年来视觉单目标跟踪方法,尤其是基于分类的视觉目标跟踪方法,进行了综述和讨论。同时,对目标跟踪问题所涉及的相关技术和方法进行了介绍; 2. 提出基于判别性排序队列的跟踪方法,该方法通过引入多尺度判别性排序队列建立目标模型,配合模型更新策略,有效的缓解了模型漂移问题。实验证明,该方法在一些较为困难环境中能够获得十分鲁棒的跟踪效果; 3. 针对理论分析和实验中遇到的实际问题,在基于判别性排序队列的跟踪方法的基础上,通过对判别性排序队列中元素加入权重以及引入部分背景模型进行改进,以提高跟踪器在困难环境中的跟踪性能; 4. 根据基于判别性排序队列视觉单目标跟踪算法及其相关改进,搭建了视觉目标跟踪系统平台。该软件平台通过合理的设计相关数据结构和进行算法优化,具备了较强的可维护性和可拓展性。
Other AbstractWith the development of computer vision and other related realms, visual object tracking has become a hotspot issue with universal concern, and it has shown a broad prospect in various applications. In this thesis, we propose a novel tracking algorithm based on the survey of existing methods and techniques. The contribution of the this thesis are fourfolded: Firstly, some visual object tracking algorithms published recently, especially the branch of tracking-by-classification, are reviewed in detail. And some related technics are also discussed to some extent; Secondly, in order to mitigate model drift problems, visual single object tracking algorithm with discriminative ranking lists is introduced, which is based on the concept of ranking list, purity and multi-scales; Thirdly, further work on this topic is also explored, including weighting different elements in the ranking lists and modeling confusing background, to enhance the robustness and adaptability of the tracker; Lastly, we build up a visual object tracking platform from the discriminative ranking lists based single object tracking algorithm and its related improvements. The proposed platform, equipped with proper system and data structure design and algorithm optimization, is maintainable and scalable.
shelfnumXWLW1644
Other Identifier200828014628049
Language中文
Document Type学位论文
Identifierhttp://ir.ia.ac.cn/handle/173211/7577
Collection毕业生_硕士学位论文
Recommended Citation
GB/T 7714
彭曦. 视觉单目标跟踪算法研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2011.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20082801462804(4054KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[彭曦]'s Articles
Baidu academic
Similar articles in Baidu academic
[彭曦]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[彭曦]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.