基于高效多示例学习的目标跟踪

doi:10.11772/j.issn.1001-9081.2015.02.0466

计算机应用 ›› 2015, Vol. 35 ›› Issue (2): 466-469.DOI: 10.11772/j.issn.1001-9081.2015.02.0466

基于高效多示例学习的目标跟踪

彭爽, 彭晓明

电子科技大学自动化工程学院, 成都 611731

收稿日期:2014-09-03 修回日期:2014-11-24 出版日期:2015-02-10 发布日期:2015-02-12
通讯作者: 彭爽
作者简介:彭爽(1989-),男,湖南湘潭人,硕士研究生,主要研究方向:图像处理、目标检测、目标跟踪; 彭晓明(1974-),男,湖北武汉人,副教授,博士,主要研究方向:目标检测、目标跟踪、目标识别、三维图像处理。
基金资助:
中央高校基本科研业务费专项(ZYGX2011J075)。

Object tracking with efficient multiple instance learning

PENG Shuang, PENG Xiaoming

School of Automation Engineering, University of Electronic Science and Technology of China, Chengdu Sichuan 611731, China

Received:2014-09-03 Revised:2014-11-24 Online:2015-02-10 Published:2015-02-12

摘要/Abstract

摘要：

基于多示例学习(MIL)的跟踪算法能在很大程度上缓解漂移问题。然而,该算法的运行效率相对较低,精度也有待提高,这是由于MIL算法采用的强分类器更新策略效率不高,以及分类器更新速度与目标外观变化速度不一致引起的。为此提出一种新的强分类器更新策略,以大幅提升MIL算法的运行效率;同时提出一种动态更新分类器学习率的机制,使更新后的分类器更符合目标的外观,提高跟踪算法的精度。通过实验将该算法和MIL算法以及基于加权多示例学习的跟踪算法(WMIL)进行对比,实验结果表明,所提出算法的运行效率和跟踪精度都是三者中最好的,在背景中没有与被跟踪目标外观相似的干扰物体存在时有较好的跟踪优势。

关键词: 目标跟踪, 多实例学习, 分类器, 漂移, 样本集

Abstract:

The method based on Multiple Instance Learning (MIL) can alleviate the drift problem to a certain extend. However, MIL method has relatively poor performance in running efficiency and accuracy, because the update strategy efficiency of the strong classifiers is low, and the update speed of the classifiers is not same with the appearance change speed of the targets. To solve this problem, a new update strategy for strong classifier was proposed to improve the running efficiency of MIL method. In addition, to improve the tracking accuracy of the MIL method, a new dynamic mechanisim for learning rate renewal of the classifier was given to make the updated classifier would more conform to the appearance of the target. The experimental results on comparison with MIL method and the Weighted Multiple Instance Learning (WMIL) method show that, the proposed method has the best performance in running efficiency and accuracy among the three methods, and has an advantage over tracking when there is no similar interference objects to target objects in background.

Key words: object tracking, Multiple Instance Learning (MIL), classifier, drift, sample set

中图分类号:

TP391.4

彭爽, 彭晓明. 基于高效多示例学习的目标跟踪[J]. 计算机应用, 2015, 35(2): 466-469.

PENG Shuang, PENG Xiaoming. Object tracking with efficient multiple instance learning[J]. Journal of Computer Applications, 2015, 35(2): 466-469.

参考文献

[1] CAI R, WU Y, WANG M, et al. Survey of visual object tracking algorithms[J]. Video Engineering, 2010, 34(12): 135-142. (蔡荣太, 吴元昊, 王明佳,等. 视频目标跟踪算法综述[J]. 电视技术, 2010, 34(12): 135-142.)
[2] ROSS D A, LIM J, LIN R-S, et al. Incremental learning for robust visual tracking [J]. International Journal of Computer Vision, 2009, 77(1/2/3): 125-141.
[3] ZHANG L, ZHANG K, YANG M. Real-time compressive tracking [C]//ECCV 2012: Proceedings of the 12th European Conference on Computer Vision, LNCS 7574. Berlin: Springer, 2012: 866-879.
[4] KWON J, LEE K M. Visual tracking decomposition [C]//CVPR 2010: Proceedings of the 2010 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2010: 1269-1276.
[5] MEI X, LING H. Robust visual tracking using l1 minimization [C]//ICCV 2009: Proceedings of the 2009 IEEE 12th International Conference on Computer Vision. Piscataway: IEEE, 2009: 1436-1443.
[6] KALAL Z, MATAS J, MIKOLAJCZYK K. P-N learning: bootstrapping binary classifiers by structural constraints [C]//CVPR 2010: Proceedings of the 23rd IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2010: 49-56.
[7] ZHOU Q-H, LU H, YANG M-H. Online multiple support instance tracking [C]//FG 2011: Proceedings of the 2011 IEEE International Conference on on Automatic Face and Gesture Recognition. Piscataway: IEEE, 2011: 545-552.
[8] LEISTINER C, GODEC M, SAFFARI A, et al. Online multi-view forests for tracking [C]//Proceedings of the 32nd DAGM Symposium on Pattern Recognition, LNCS 6376. Berlin: Springer, 2010:493-502.
[9] BABENKO B, YANG M-H, BELONGIE S. Robust object tracking with online multiple instance learning [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(8): 1619-1632.
[10] NING J, SHI W, YANG S, et al. Improved appearance updating method in multiple instance learning tracking [J]. IET Computer Vision, 2014, 8(2): 118-130.
[11] NING J, ZHAO Y, SHI W. Multiple instance learning based object tracking with multi-channel Haar-like feature [J]. Journal of Image and Graphics, 2014, 19(7): 1038-1045. (宁纪锋,赵耀博,石武祯.多通道Haar-like特征多示例学习目标跟踪[J]. 中国图象图形学报,2014,19(7):1038-1045.)
[12] ZHANG K, SONG H. Real-time visual tracking via online weighted multiple instance learning [J]. Pattern Recognition, 2013, 46(1): 397-411.
[13] GRABNER H, GRABNER M, BISCHOF H. Real-time tracking via online boosting [C]//BMVC 2006: Proceedings of the 17th British Machine Vision Conference. Nottingham: British Machine Vision Association Press, 2006: 6.1-6.10.
[14] GRABNER H, LEISTNER C, BISCHOF H. Semi-supervised on-line boosting for robust tracking [C]//ECCV 2008: Proceedings of the 10th European Conference on Computer Vision, LNCS 5302. Berlin: Springer, 2008: 234-247.

[1]	姬张建, 任兴旺. 带旋转与尺度估计的全卷积孪生网络目标跟踪算法[J]. 计算机应用, 2021, 41(9): 2705-2711.
[2]	尹春勇, 张帼杰. 面向分布式漂移数据流的集成分类模型[J]. 计算机应用, 2021, 41(7): 1947-1955.
[3]	严爱军, 魏志远. 案例推理分类器的权重分配及案例库维护方法[J]. 计算机应用, 2021, 41(4): 1071-1077.
[4]	李萍, 汪芬, 陈祺东, 孙俊. 求解多目标社区发现问题的离散化随机漂移粒子群优化算法[J]. 计算机应用, 2021, 41(3): 803-811.
[5]	钟莎, 黄玉清. 基于孪生区域候选网络的无人机指定目标跟踪[J]. 计算机应用, 2021, 41(2): 523-529.
[6]	闫钧华, 侯平, 张寅, 吕向阳, 马越, 王高飞. 基于多尺度多分类器卷积神经网络的混合失真类型判定方法[J]. 《计算机应用》唯一官方网站, 2021, 41(11): 3178-3184.
[7]	尹春勇, 朱宇航. 基于垂直集成Tri-training的虚假评论检测模型[J]. 计算机应用, 2020, 40(8): 2194-2201.
[8]	肖跃雷, 张云娇. 基于特征选择和超参数优化的恐怖袭击组织预测方法[J]. 计算机应用, 2020, 40(8): 2262-2267.
[9]	张杰, 常天庆, 戴文君, 郭理彬, 张雷. 基于相关滤波与颜色概率模型的目标跟踪算法[J]. 计算机应用, 2020, 40(6): 1774-1782.
[10]	王留洋, 俞扬信, 陈伯伦, 章慧. 基于共识和分类改善文档聚类的识别信息方法[J]. 计算机应用, 2020, 40(4): 1069-1073.
[11]	何祥宇, 李静, 杨数强, 夏玉杰. 基于ET-PHD滤波器和变分贝叶斯近似的扩展目标跟踪算法[J]. 计算机应用, 2020, 40(12): 3701-3706.
[12]	邓天民, 方芳, 周臻浩. 基于改进空间金字塔池化卷积神经网络的交通标志识别[J]. 计算机应用, 2020, 40(10): 2872-2880.
[13]	石国强, 赵霞. 基于联合优化的强耦合孪生区域推荐网络的目标跟踪算法[J]. 计算机应用, 2020, 40(10): 2822-2830.
[14]	杨博, 蔺素珍, 禄晓飞, 李大威, 秦品乐, 左健宏. 基于多特征融合与分层数据关联的空中红外多目标跟踪方法[J]. 计算机应用, 2020, 40(10): 3075-3080.
[15]	俞皓芳, 孙力帆, 付主木. 基于改进K-means++聚类的多扩展目标跟踪算法[J]. 计算机应用, 2020, 40(1): 271-277.

基于高效多示例学习的目标跟踪

Object tracking with efficient multiple instance learning

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics