基于组合带宽均值迁移的快速目标跟踪

周斌; 王军政; 沈伟

doi:10.3788/OPE.20101810.2297

您当前的位置：

首页 >

文章列表页 >

基于组合带宽均值迁移的快速目标跟踪

信息科学 | 更新时间：2020-08-12

- 基于组合带宽均值迁移的快速目标跟踪
- Fast object tracking with multi-bandwidth Mean Shift
- 光学精密工程 2010年18卷第10期页码：2297-2305
- 作者机构：
  
  北京理工大学自动化学院复杂系统智能控制与决策实验室北京,100081
- 作者简介：
- 基金信息：
  
  985"();工程学科建设投资项目(No.107008200400020)()
- DOI：10.3788/OPE.20101810.2297
  中图分类号： TP391.4
- 收稿日期：2010-01-29，
  
  修回日期：2010-03-25，
  
  网络出版日期：2010-10-28，
  
  纸质出版日期：2010-10-20
- 稿件说明：
移动端阅览
周斌, 王军政, 沈伟. 基于组合带宽均值迁移的快速目标跟踪[J]. 光学精密工程, 2010,18(10): 2297-2305

ZHOU Bin, WANG Jun-zheng, SHEN Wei. Fast object tracking with multi-bandwidth Mean Shift[J]. Editorial Office of Optics and Precision Engineering , 2010,18(10): 2297-2305
周斌, 王军政, 沈伟. 基于组合带宽均值迁移的快速目标跟踪[J]. 光学精密工程, 2010,18(10): 2297-2305 DOI： 10.3788/OPE.20101810.2297.

ZHOU Bin, WANG Jun-zheng, SHEN Wei. Fast object tracking with multi-bandwidth Mean Shift[J]. Editorial Office of Optics and Precision Engineering , 2010,18(10): 2297-2305 DOI： 10.3788/OPE.20101810.2297.

摘要

为了解决传统均值迁移(Mean shift)目标跟踪算法中跟踪窗口容易收敛至局部概率模式的问题

提出一种基于组合带宽Mean Shift的目标跟踪策略

并建立了一种自适应学习率的over-relaxed优化策略以加速收敛过程。根据目标尺度设定了一组从大到小排列的带宽序列

并依次根据每个带宽进行Mean Shift迭代收敛运算

利用大带宽的平滑作用避开局部概率模式的干扰;依靠小带宽进行精确定位

最终使其收敛到真实目标区域。由于组合带宽Mean Shift会造成一定的额外运算量

为此引入over-relaxed优化策略加速迭代过程。在边界优化算法的收敛条件约束下

根据采用over-relaxed策略前后相关系数的变化

自适应地调整学习率。实验结果表明

组合带宽Mean Shift能够有效地跟踪快速运动的目标

并且当目标短暂丢失时也有一定的恢复能力;实验采用over-relaxed策略后

收敛次数减少了30%~70%。

Abstract

An object tracking algorithm with multi-bandwidth and adaptive over-relaxed accelerated convergence was proposed to avoid the local probability mode in a Mean Shift tracking process. Firstly

a monotonically decreasing sequence of bandwidths was obtained according to the object scale. At the first bandwidth

a maximum probability could be found with the Mean Shift

and the next iteration loop started at the previous convergence location. Finally

the best density mode was obtained at the optimal bandwidth. In the convergence process

the compactness of the local probability mode was avoided with the smoothing effect of the large bandwidth

and the precise position of the object could be found with the optimal bandwidth

which was similar to the object scale. To speed up the convergence

an over-relaxed strategy was introduced to enlarge the step size. Under the convergence rule

the correlation coefficient was used to adjust the learning rate adaptively. The experimental results prove that the proposed tracker with multi-bandwidth Mean Shift is robust in high-speed object tracking

and performs well in occlusions. The experimental results also show that the adaptive over-relaxed strategy reduces the convergence iterations by 30%-70%.

关键词

Keywords

references

薛陈,朱明,陈爱华. 鲁棒的基于改进Mean-shift的目标跟踪[J]. 光学精密工程,2010,18(1):234-239. XUE CH, ZHU M, CHEN A H. Robust object tracking based on improved Mean-shift algorithm[J]. Opt. Precision Eng., 2010,18(1):234-239. (in Chinese)[2] 孟勃,朱明. MSMC跟踪算法在目标跟踪中的应用[J]. 光学精密工程,2008,16(1):122-127. MENG B, ZHU M. Application of MSMC algorithm to visual tracking[J]. Opt. Precision Eng., 2008,16(1):122-127.(in Chinese)[3] MAGGIO E, CAVALLARO A. Accurate appearance based bayesian tracking for maneuvering targets [J]. Computer Vision and Image Understanding,2009,113:544-555.[4] 王永忠, 梁彦, 赵春晖. 基于多特征自适应融合的核跟踪方法[J]. 自动化学报, 2008,34(1):393-399. WANG Y ZH, LIANG Y, ZHAO CH H. Kernel-based tracking based on adaptive fusion of multiple cues [J]. Acta Automatica Sinica, 2008,34(4):393-399.[5] ZHANG K, KWOK J T, TANG M. Accelerate convergence using dynamic mean shift . Proceedings of the 9th European Conference on Computer Vision, New York, 2006:257-268.[6] FASHING M, TOMASI C. Mean Shift is a bound optimization[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2005,27(3):471-474.[7] SHEN C, BROOKS M J. A fast global kernel density mode seeking with application to localization and tracking . Proceedings of IEEE International Conference on Computer Vision, Los Alamitos, 2005:1516-1523.[8] YIN Z Z, ROBERT T. Object tracking and detection after occlusion via numerical hybrid local and global mode-seeking . Proceedings of IEEE Conference on Computer Vision and Pattern Recognition(CVPR), Anchorage,2008:1-8.[9] ELGAMMAL A, DURAISWAMI R. Probabilistic tracking in joint feature-spatial spaces . Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. Washington, D.C,2004:790-797.[10] COMANICIU D, MEER P. Kernel-based object tracking [J]. IEEE Trans. On Pattern Analysis and Machine Intelligence, 2003,25(5):564-577.[11] CARREIRA PERPINAN M A. Acceleration strategies for Gaussian Mean Shift image segmentation . Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition,New York, 2006:543-549.

浏览量

410

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于知识蒸馏的Transformer视觉跟踪器

针对RGBT跟踪的特殊属性的跨模态交互融合网络

基于显著性区域加权的相关滤波目标跟踪

基于双模型的相关滤波跟踪算法

基于双模型的相关滤波跟踪