4-D尺度空间中基于Mean-Shift的目标跟踪

王宇雄; 章毓晋; 王晓华

doi:10.3724/SP.J.1146.2009.01000

4-D尺度空间中基于Mean-Shift的目标跟踪

doi: 10.3724/SP.J.1146.2009.01000

王宇雄^{①;章毓晋,},
章毓晋,
王晓华

计量
- 文章访问数: 3991
- HTML全文浏览量: 138
- PDF下载量: 886
- 被引次数: 0
出版历程
- 收稿日期: 2009-07-14
- 修回日期: 2009-11-14
- 刊出日期: 2010-07-19

Mean-Shift Object tracking through 4-D Scale Space

摘要

摘要: 在基于Mean-Shift的目标跟踪算法中，尺度自适应机制是算法研究的一个重要方向。一种典型的方法采用Lindeberg的尺度空间理论以获取目标尺度信息。但现有算法中将尺度由2-D矢量压缩为1-D量，未能精细地刻画目标仿射变换时的尺度变化，从而限制了算法的适用范围。为此，该文将尺度维1-D滤波推广至2-D，构造得到了相应的4-D尺度空间，并利用空间维和尺度维的Mean-Shift交替迭代，实现了同时在空间位置和尺度方向对目标的有效跟踪，提高了算法在目标尺度变化时的自适应性，并扩大了算法的适用范围。
- 目标跟踪; 均值偏移; 尺度空间理论; 尺度自适应
Abstract: The scale self-adaptive mechanism is one of the promising research directions in the object tracking issue based on mean-shift. A typical method is to adopt Lindebergs scale-space theory to obtain the scale information of the target. However, the 2-D scale vector is reduced to 1-D in the existing algorithm. So it is not precise enough to portray the changing state of the object scale under affine transformation, which greatly limits the scope of application of the method. Hence, the 1-D filter in the scale dimension is extended to 2-D in this paper. Correspondingly, the 4-D scale space is constructed, then the problem is mapped into two interleaved mean-shift procedure in the spatial and scale dimension through scale space. The modified algorithm achieves the effective object tracking in spatial position and scale direction simultaneously, which enhances the self adaptability when the target scale is changing, and expands the scope of application of the algorithm.

HTML全文

参考文献(1)

Cheng Y. Mean shift, mode seeking, and clustering [J].IEEETransactions on Pattern Analysis and Machine Intelligence.1995, 17(8):790-799[2]Comaniciu D, Ramesh V, and Meer P. Real-time tracking ofnon-rigid objects using mean shift [C]. IEEE Conference onComputer Vision and Pattern Recognition, Hilton Head, SC,USA, 2000, II: 142-149.[3]Yilmaz A. Object tracking by asymmetric kernel mean shiftwith automatic scale and orientation selection [C]. IEEEConference on Computer Vision and Pattern Recognition,Minneapolis, MN, USA, 2007: 1-6.[4]Yao An-bang, Wang Gui-jin, Lin Xing-gang, and Wang Hao.Kernel based articulated object tracking with scaleadaptation and model update [C]. IEEE InternationalConference on Acoustics, Speech and Signal Processing, LasVegas, NV, USA, 2008: 945-948.Chen Xiao-peng, Zhou You-xue, Huang Xiao-san, and LiCheng-rong. Adaptive bandwidth mean shift object tracking[C]. IEEE Conference on Robotics, Automation andMechatronics, Chengdu, China, 2008: 1011-1017.Jiang Zhuo-lin, Li Shao-fa, Jia Xi-ping, and Zhu Hong-li. Animproved mean shift tracking method based onnonparametric clustering and adaptive bandwidth [C].Proceedings of the Seventh International Conference onMachine Learning and Cybernetics, Kunming, China, 2008:12-15.[5]Collins R T. Mean-shift blob tracking through scale space [C].IEEE Conference on Computer Vision and PatternRecognition, Baltimore, USA, 2003, 2: 234-240.[6]彭宁嵩, 杨杰, 刘志, 张凤超. Mean Shift 跟踪算法中核函数窗宽的自动选取 [J]. 软件学报, 2005, 16(9): 1542-1550.Peng Ning-song, Yang Jie, Liu Zhi, and Zhang Feng-chao.Automatic selection of kernel-bandwidth for Mean-Shiftobject tracking [J]. Journal of Software, 2005, 16(9):1542-1550.[7]章毓晋. 图像工程. 上册, 图像处理 [M]. 第2 版, 北京: 清华大学出版社, 2006: 62-71.Zhang Yu-jin. Image Engineering (I), Image Processing [M].Second Edition, Beijing: Tsinghua University Press, 2006:62-71.[8]Lindeberg T. Scale-Space Theory in Computer Vision [M].Netherlands: Kluwer Academic Publisher, 1994, Chapter 13.

施引文献

资源附件(0)

访问统计