吉林大学学报(工学版) ›› 2017, Vol. 47 ›› Issue (1): 281-287.doi: 10.13229/j.cnki.jdxbgxb201701041

• 论文 • 上一篇    下一篇

遮挡环境下多示例学习分块目标跟踪

才华1, 陈广秋1, 刘广文1, 程帅1, 于化东2   

  1. 1.长春理工大学 电子信息工程学院,长春 130022;
    2.长春理工大学 机电工程学院,长春 130022
  • 收稿日期:2016-02-21 出版日期:2017-01-20 发布日期:2017-01-20
  • 通讯作者: 陈广秋(1977-),男,副教授,博士.研究方向:图像(序列)配准与融合.E-mail:guangqiu_chen@126.com
  • 作者简介:才华(1977-),男,副教授,博士.研究方向:图像处理与机器视觉.E-mail:caihua@cust.edu.cn
  • 基金资助:
    吉林省科技发展计划项目(20130101179JC); 教育部留学基金委留学归国人员科研启动基金(教外师留1685).

Novelty fragments-based target tracking with multiple instance learning under occlusions

CAI Hua1, CHEN Guang-qiu1, LIU Guang-wen1, CHENG Shuai1, YU Hua-dong2   

  1. 1.School of Electronic and Information Engineering, Changchun University of Science and Technology, Changchun 130022,China;
    2.College of Mechanical and Electric Engineering, Changchun University of Science and Technology, Changchun 130022,China
  • Received:2016-02-21 Online:2017-01-20 Published:2017-01-20

RICH HTML

0   

PDF (PC)

370

摘要： 针对跟踪过程中遮挡导致错误累积而产生目标漂移甚至目标丢失的问题,提出多示例学习分块目标跟踪方法。该方法以随机蕨为基础检测器,通过多示例学习在线更新检测器,提高检测器对目标变化的适应能力及学习的准确性。将目标均匀分成多个子块,选取部分子块作为候选集合,每个候选块分配一个检测器,利用检测器完成每个候选块的跟踪,根据候选块的信息确定目标最终位置。实时检测候选块的有效性,替换无效的候选块,提高跟踪的鲁棒性。在目标被遮挡等复杂条件下,与目前主流跟踪算法进行了对比实验,结果表明该算法能够有效解决目标漂移甚至跟踪丢失问题,具有更高的跟踪精确度及更好的鲁棒性。

关键词: 信息处理技术, 随机蕨分类器, 多示例学习, 分块, 无效子块更换

Abstract: To solve the problem that tracking algorithm may lead to drift or failure due to the accumulated error under the occlusion environment, a Multiple instance learning based Fragment Tracker (MFT) is proposed. In this MFT, the random ferns is used as the basic detector. To improve the adaption of the target change and the precision of the learning, the multiple instance learning is introduced to online update the detector. The object is segmented into fragments and parts of them are selected as the candidate set. The candidate block is tracked by the corresponding detector. The object can be finally located by the selected blocks. A real-time valid detection is made for the candidate blocks and the invalid block is replaced with an appropriate block to improve the robustness of the tracking. Experiments on variant challenging image sequence in the occlusion environment were carried out. Results show that, compared with the state-of-art trackers, the proposed MFT solves the problem of target drift and failure efficiently and has higher accuracy and better robust.

Key words: information processing, random ferns detector, multiple instance learning, fragment, invalid block replacement

中图分类号: 

  • TP391.41
[1] 薛陈, 朱明, 刘春香. 遮挡情况下目标跟踪算法综述[J].中国光学与应用光学, 2009, 2(5):388-394.
Xue Chen, Zhu Ming, Liu Chun-xiang. Review of tracking algorithms under occlusions[J]. Chinese Journal of Optics and Applied Optics, 2009, 2(5): 388-394.
[2] 颜佳, 吴敏渊. 遮挡环境下采用在线Boosting的目标跟踪[J]. 光学精密工程, 2012,20(2):439-446.
Yan Jia, Wu Min-yuan. On-line boosting based target tracking under occlusion[J]. Optics and Precision Engineering, 2012,20(2):439-446.
[3] Alper Y, Omar J, Mubarak S. Object tracking: a survey[J]. ACM Computing Surveys, 2006, 38(4):112-123.
[4] Medeiros H, Park J, Kak A. A parallel color-based particle filter for object tracking[C]∥IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Anchorage, USA, 2008: 1-8.
[5] Zhang T Z, Ghanem B, Xu C, et al. Object tracking by occlusion detection via structured sparse learning[C]∥IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Portland, USA,2013: 1033-1040.
[6] Zhang S P,Wu J P, Tian Y, et al. Robust visual tracking based on occlusion detection and particle redistribution[C]∥The Second International Conference on Internet Multimedia Computing and Service, New York, USA, 2010: 159-162.
[7] Mei X, Ling H B, Wu Y, et al. Minimum error bounded efficient L1 tracker with occlusion detection[C]∥IEEE Computer Society Conference on Computer Vision and Pattern Recognition, New York, USA, 2011: 1257-1264.
[8] 陈东成, 朱明, 高文, 等. 在线加权多示例学习实时目标跟踪[J]. 光学精密工程, 2014, 22(6): 1661-1667.
Chen Dong-cheng, Zhu Ming, Gao Wen, et al. Real-time object tracking via online weighted multiple instance learning[J]. Optics and Precision Engineering, 2014, 22(6): 1661-1667.
[9] Andriluka M, Roth S, Schiele B. People-tracking-by-detection and people-detection-by-tracking[C]∥IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Anchorage,USA, 2008: 1-8.
[10] Gall J, Yao A, Razavi N, et al.. Hough forests for object detection, tracking, and action recognition[J]. IEEE Trans Pattern Analysis and Machine Intelligence, 2011, 33(11): 2188-2202.
[11] Kalal Z,Matas J,Mikolajczyk K.Online learning of robust object detectors during unstable tracking[C]∥IEEE International Conference on Computer Vision Workshops, Kyoto, Japan, 2009: 1417-1424.
[12] 贾松敏, 文林风, 王丽佳. 基于多模板回归加权均值漂移的人体目标跟踪[J]. 光学精密工程, 2016, 24(9): 2339-2346.
Jia Song-min, Wen Lin-feng, Wang Li-jia. Person tracking based on multi-template regression weighted mean shift[J]. Optics and Precision Engineering, 2016, 24(9): 2339-2346.
[13] Adam A, Rivlin E, Shimshoni I. Robust fragment-based tracking using the integral histogram[C]∥IEEE Computer Society Conference on Computer Vision and Pattern Recognition, New York,USA,2006: 798-805.
[14] Naik N, Patil S, Joshi M. A fragment based scale adaptive tracker with partial occlusion handling[C]∥IEEE Region 10 Conference, Singapore, 2009: 1-6.
[15] Kalal Z, Mikolajczyk K, Matas J. Tracking-learning-detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(7): 1409-1422.
[16] Babenko B, Yang M H, Belongle S. Robust object tracking with online multiple instance learning[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(8): 1619-1632.
[17] Zhang K, Zhang L, Yang M H. Fast compressive tracking[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2014, 36(10): 2002-2015.
[18] Dinh T B, Vo N, Medion G. Context tracker: exploring supporters and distracters in unconstrained environments[C]∥IEEE Conference on Computer Vision and Pattern Recognition,Providence,USA,2011: 1177-1184.
[19] Ross D A, Lim J, Lin R S, et al. Incremental learning for robust visual tracking[J]. International Journal of Computer Vision, 2008, 77(1-3):125-141.
[20] Grabner H, Grabner M, Bischof H. Real-time tracking via on-line boosting[C]∥British Machine Vision Conference, Edinburgh, Britain, 2006:47-56.
[21] 罗艳,项俊, 严明君,等. 基于多示例学习和随机蕨丛检测的在线目标跟踪[J]. 电子与信息学报,2014,36(7):1605-1611.
Luo Yan, Xiang Jun, Yan Ming-jun, et al. Online target tracking based on mulitiple instance learning and random ferns detection[J]. Journal of Electronics & Information Technology, 2014,36(7):1605-1611.
[22] Wu Y, Lim J, Yang M H. Online object tracking:a benchmark[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 36(4): 2411-2418.
[1] 苏寒松,代志涛,刘高华,张倩芳. 结合吸收Markov链和流行排序的显著性区域检测[J]. 吉林大学学报(工学版), 2018, 48(6): 1887-1894.
[2] 徐岩,孙美双. 基于卷积神经网络的水下图像增强方法[J]. 吉林大学学报(工学版), 2018, 48(6): 1895-1903.
[3] 黄勇,杨德运,乔赛,慕振国. 高分辨合成孔径雷达图像的耦合传统恒虚警目标检测[J]. 吉林大学学报(工学版), 2018, 48(6): 1904-1909.
[4] 李居朋,张祖成,李墨羽,缪德芳. 基于Kalman滤波的电容屏触控轨迹平滑算法[J]. 吉林大学学报(工学版), 2018, 48(6): 1910-1916.
[5] 应欢,刘松华,唐博文,韩丽芳,周亮. 基于自适应释放策略的低开销确定性重放方法[J]. 吉林大学学报(工学版), 2018, 48(6): 1917-1924.
[6] 陆智俊,钟超,吴敬玉. 星载合成孔径雷达图像小特征的准确分割方法[J]. 吉林大学学报(工学版), 2018, 48(6): 1925-1930.
[7] 刘仲民,王阳,李战明,胡文瑾. 基于简单线性迭代聚类和快速最近邻区域合并的图像分割算法[J]. 吉林大学学报(工学版), 2018, 48(6): 1931-1937.
[8] 单泽彪,刘小松,史红伟,王春阳,石要武. 动态压缩感知波达方向跟踪算法[J]. 吉林大学学报(工学版), 2018, 48(6): 1938-1944.
[9] 姚海洋, 王海燕, 张之琛, 申晓红. 双Duffing振子逆向联合信号检测模型[J]. 吉林大学学报(工学版), 2018, 48(4): 1282-1290.
[10] 全薇, 郝晓明, 孙雅东, 柏葆华, 王禹亭. 基于实际眼结构的个性化投影式头盔物镜研制[J]. 吉林大学学报(工学版), 2018, 48(4): 1291-1297.
[11] 陈绵书, 苏越, 桑爱军, 李培鹏. 基于空间矢量模型的图像分类方法[J]. 吉林大学学报(工学版), 2018, 48(3): 943-951.
[12] 陈涛, 崔岳寒, 郭立民. 适用于单快拍的多重信号分类改进算法[J]. 吉林大学学报(工学版), 2018, 48(3): 952-956.
[13] 孟广伟, 李荣佳, 王欣, 周立明, 顾帅. 压电双材料界面裂纹的强度因子分析[J]. 吉林大学学报(工学版), 2018, 48(2): 500-506.
[14] 林金花, 王延杰, 孙宏海. 改进的自适应特征细分方法及其对Catmull-Clark曲面的实时绘制[J]. 吉林大学学报(工学版), 2018, 48(2): 625-632.
[15] 王柯, 刘富, 康冰, 霍彤彤, 周求湛. 基于沙蝎定位猎物的仿生震源定位方法[J]. 吉林大学学报(工学版), 2018, 48(2): 633-639.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
版权所有 © 《吉林大学学报(工学版)》编辑部
地址：长春市人民大街5988号　电话：+86-431-85095297　传真：+86-431-85094074　E-mail: xbgxb@jlu.edu.cn
本系统由北京玛格泰克科技发展有限公司设计开发