增强目标模型鲁棒性的Mean-shift算法
PDF下载 (657)郭运艳,陈 恳,宋康康,刘 哲,黄小霞.增强目标模型鲁棒性的Mean-shift算法[J].宁波大学学报(理工版),2012,25(04):25-28.DOI:
GUO Yun-yan,CHEN Ken,SONG Kang-kang,LIU Zhe,HUANG Xiao-xia.Mean-shift Tracking with Enhanced Robustness of Target Model[J].Journal of Ningbo University(Natural Science & Engineering Edition),2012,25(04):25-28.DOI:
| Title: | Mean-shift Tracking with Enhanced Robustness of Target Model |
| 作者: | 郭运艳, 陈 恳, 宋康康, 刘 哲, 黄小霞 |
| Author(s): | GUO Yun-yan, CHEN Ken, SONG Kang-kang, LIU Zhe, HUANG Xiao-xia |
| 关键词: | Harris检测算子; 旋转不变性; 联合直方图 |
| Keywords: | Mean-shift; Harris; rotation-invariance; color and gradient orientation histogram distribution |
| 分类号: | TN919.8 |
| 文献标识码: | A |
| 摘要: | 针对传统Mean-shift算法中颜色核函数直方图对目标特征描述较弱的缺点, 提出了一种联合目标特征点的二维结构信息和颜色信息的Mean-shift改进算法. 改进算法细化了Harris检测算子的角点响应阈值, 提取出更多的目标特征点计算其方向分布, 并以方向与部分颜色特征的直方图构建目标模型, 该模型能显著区分目标与背景. 实验对不同算法进行了仿真及性能比较, 结果表明: 提出的改进算法在一定的复杂场景中提高了跟踪精度, 且具有较好的鲁棒性. |
| Abstract: | The traditional Mean-shift cannot represent accurately the color distribution of the object. To tackle this problem, a target motion-tracking algorithm is presented based on two-dimensional structural information of the target feature points and color histogram. The improved algorithm refines the Harris operator threshold to seek more feature points, with which their orientation distribution can also be sought. The target model built by combining the orientation histogram and the partial color histogram is found to be able to separate target from background to a satisfying extent. The application results from various video sequences suggest good localization precision in object tracking, and algorithmic robustness in complex scenes. |
| 参考文献 /References: | [1] Li P. An adaptive binning color model for mean shift tracking[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2008, 18(9):1293-1299. [2] Comaniciu D, Ramesh V, Meer P. Kernel-based object tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2003, 25(5):564-577. [3] 柳伟, 罗以宁, 孙南. 基于背景优化的Mean-shift目标跟踪算法[J]. 计算机应用, 2009, 29(4):1015-1017. [4] 冯祖仁, 吕娜, 李良福. 基于最大后验概率的图像匹配相似性指标研究[J]. 自动化学报, 2007, 33(1):1-8. [5] Wang Yongming, Wang Guijin. Image local invariant features and descriptors[M]. Beijing: National Defense Industry Press, 2010. [6] Ning Jifeng, Jiang Guang, Li Peifei. Mean shift tracking algorithm based on corner points[J]. Application Research of Computers, 2009, 26(1):4348-4350. [7] Shen Huan, Li Shunming, Bo Fangchao, et al. On road vehicles detection approach using multi-cues fuse[J]. Journal of Optoelectronics Laser, 2010, 21(1):74-77. [8] 郑玉凤, 马秀荣. 基于颜色与边缘特征的均值迁移目标跟踪算法[J]. 光电子激光, 2011, 22(8):1231-1235. [9] Harris C, Stephens M J. A combined corner and edge detector[C]. Alley Vision Conference, 1988:147-152. [10] 荆其诚. 色度学[M]. 北京: 科学出版社, 1991. |
| 备注/Memo: | 收稿日期: 2012-06-02. 基金项目: 国家自然科学基金（61071120）; 浙江省教育厅科研项目（Y200907622）; 宁波市自然科学基金（210A610109）. 第一作者: 郭运艳（1987－）, 女, 山东临沂人, 在读硕士研究生, 主要研究方向: 视频目标跟踪及图像处理. E-mail: yunyan_509@163.com *通讯作者: 陈 恳（1962－）, 男, 重庆人, 博士/副教授, 主要研究方向: 图像与视频分析处理及智能控制. E-mail: chenken@nbu.edu.cn 宁波大学学报（理工版）网址：http://journallg.nbu.edu.cn/ |
