基于多核融合的目标大面积遮挡处理方法

肖鹏; 段洣毅; 赵琪

基于多核融合的目标大面积遮挡处理方法

1.
北京航空航天大学, 计算机学院, 北京 100191
2. 北京图形研究所, 北京 100029

基金项目: 国家自然科学基金资助项目(61005084)

详细信息

中图分类号: TP391
计量
- 文章访问数: 4312
- HTML全文浏览量: 158
- PDF下载量: 588
- 被引次数: 0
出版历程
- 收稿日期: 2011-03-22
- 网络出版日期: 2012-06-30

Occlusion handling approach in visual tracking based on multiple-kernel fusion

1.
School of Computer Science and Technology, Beijing University of Aeronautics and Astronautics, Beijing 100191, China
2. Beijing Institute of Graphics, Beijing 100029, China

摘要

摘要: 提出了一种基于多核融合的目标遮挡处理方法,用于提高大面积遮挡情况下视觉目标跟踪算法的鲁棒性和准确性.与现有基于单个对称核加权直方图的mean shift跟踪算法不同,该方法以目标区域内的多个非中心位置为核函数中心,构建多个非对称核加权直方图.由于这些直方图对目标的不同区域赋予了不同的权重,使得在遮挡发生时总存在一些直方图受影响较小.依据各个直方图分别进行mean shift迭代获得一组目标位置估计后,利用D-S证据理论融合判定最终的目标位置.实验结果表明,该方法在目标被大面积遮挡时仍能够获得准确的跟踪.
- 目标跟踪 /
- 视觉跟踪 /
- 均值漂移 /
- 遮挡处理 /
- 多核 /
- 证据理论
Abstract: A novel visual tracking approach based on multiple-kernel fusion was proposed to improve robustness and accuracy of tracking under large-area occlusion. Unlike traditional single symmetric kernel weighted histogram used in mean shift tracking, this approach adopted several asymmetric kernel functions centered at different positions within target region to build a set of asymmetric kernel weighted histograms. Because these histograms weighted each part of the target region differently, there must be some less influenced histograms during occlusion. Based on each histogram, a set of target location estimations were provided respectively by mean shift iteration, and the target location was obtained by fusing these estimations using Dempster-Shafer evidence theory. The experimental results demonstrate the effectiveness of the proposed approach under large-area occlusion.
- target tracking /
- visual tracking /
- mean shift /
- occlusion handling /
- multiple kernels /
- evidence theory

HTML全文

参考文献(11)

[1]	Comaniciu D,Ramesh V,Meer P.Kernel-based object tracking[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2003,25(5):564-577
[2]	Fukunaga K,Hostetler L.The estimation of the gradient of a density function,with applications in pattern recognition[J].IEEE Trans on Inform Theory,1975,21(1):32
[3]	朱胜利.Mean Shift及相关算法在视频跟踪中的研究[D].杭州:浙江大学电气工程学院,2006 Zhu Shengli.A study of mean shift and correlative algorithm in Visual Tracking[D].Hangzhou:College of Electrical Engineering,Zhejiang University,2006(in Chinese)
[4]	Wang Jiangtao,Yang Jingyu.Object tracking based on Kalman-mean shift in occlusions[J].Journal of System Simulation,2007(9):4216-4220
[5]	刘少华,张茂军,熊志辉,等.一种鲁棒高效的视频运动目标检测与跟踪算法[J].自动化学报,2009,35(8):1055-1062 Liu Shaohua,Zhang Maojun,Xiong Zhihui,et al.A robust and efficient video moving object detection and tracking algorithm[J].Acta Automatica Sinica,2009,35(8):1055-1062(in Chinese)
[6]	贾慧星,章毓晋.基于梯度方向直方图特征的多核跟踪[J].自动化学报,2009,35(10):1283-1289 Jia Huixing,Zhang Yujin.Multiple kernels based object tracking using histograms of oriented gradients[J].Acta AutoMatica Sinica,2009,35(10):1283-1289(in Chinese)
[7]	颜佳,吴敏渊,陈淑珍,等.应用Mean Shift和分块的抗遮挡跟踪[J].光学精密工程,2010,18(6):1413-1419 Yan Jia,Wu Minyuan,Chen Shuzhen,et al.Anti-occlusion tracking algorithm based on Mean Shift and fragments[J].Optics and Precision Engineering,2010,18(6):1413-1419(in Chinese)
[8]	Comaniciu D,Meer P.Robust analysis of feature space:color image segmentation[C]//Proc 1997 IEEE Conf Computer Vision and Pattern Recognition.Puerto Rico:IEEE Computer Society,1997
[9]	Dempster A P.Upper and lower probabilities induced by a multi-valued mapping[J].Annual Mathematical Statistics,1967,38(4):325-339
[10]	Shafer G.A mathematical theory of evidence[M].Princeton,NJ:Princeton University Press,1976
[11]	韩崇昭,朱洪艳,段战胜.多源信息融合[M].北京:清华大学出版社,2006:86-88 Han Chongzhao,Zhu Hongyan,Duan Zhansheng.Multi-source information fusion[M].Beijing:Tsinghua University Press,2006:86-88(in Chinese)