计算机科学 ›› 2015, Vol. 42 ›› Issue (3): 311-315.doi: 10.11896/j.issn.1002-137X.2015.03.064

• 图形图像与模式识别 • 上一篇    下一篇

基于可变形部件模型的内河多船舶跟踪算法研究

朱 琳,郭建明,刘 清,李 静   

  1. 武汉理工大学自动化学院 武汉430070,武汉理工大学自动化学院 武汉430070,武汉理工大学自动化学院 武汉430070,武汉理工大学自动化学院 武汉430070
  • 出版日期:2018-11-14 发布日期:2018-11-14
  • 基金资助:
    本文受国家自然科学基金项目(51279152),武汉理工大学自主创新研究基金项目(145211005)资助

Multiple Ship Tracking in Inland Waterway via Deformable Part Model

ZHU Lin, GUO Jian-ming, LIU Qing and LI Jing   

  • Online:2018-11-14 Published:2018-11-14

摘要: 内河航运事业的不断发展对电子巡航闭路电视(CCTV)监控系统的智能化水平提出了越来越高的要求。针对目前CCTV监控系统智能化水平较低、人工参与量较大的局限,提出了一种鲁棒的基于可变形部件模型的内河多船舶跟踪算法。该算法将每条运动目标船舶视为一个部分,通过最小生成树模型来建立各个部分之间的联系,并基于可变形部件模型的原理最终实现同时对多条目标船舶的鲁棒跟踪。为了得到精确的目标船舶外观模型,首先对目标区域进行梯度方向直方图(HOG)特征提取,然后利用模糊支持向量机(SVM)算法进行训练得到每条目标船舶的参数化的外观模型。其中,由于模糊SVM中模糊度的引入,对不同的输入训练样本赋予不同的重要性,因此将获得比线性SVM算法更加精确的目标外观模型。结构化的学习方法保证了在目标运动过程中该算法能即时更新目标间的空间相互关系参数,实现鲁棒的跟踪效果。实验结果表明,提出的算法适用于在内河环境下鲁棒、有效的多目标船舶跟踪。

关键词: 内河,闭路电视监控系统,多船舶跟踪,模糊SVM,可变形部件模型

Abstract: The closed-circuit television (CCTV) surveillance system is developing rapidly in recent years.But the intelligence level is relatively low.In this paper,a robust multiple ship tracking algorithm was proposed based on the defor-mable part model.The proposed algorithm treats every ship as a part.By incorporating the spatial constraints,the interrelations model between ships with the minimum spanning tree model can be effectively built.Then the robust multiple ship tracking is accomplished based on the deformable part model.Moreover,aiming at obtaining an accurate paramete-rized appearance model of ships,the HOG features combined with the fuzzy SVM is adopted to train those object regions.Especially,because of the ambiguity in the fuzzy SVM,every training samples are given different importance so as to obtain a more accurate appearance model.At the same time,structured learning can guarantee to update the interrelation parameters on time when ships move.Experimental results demonstrate that our proposed algorithm is suitable for inland waterway and can accomplish robust and effective multiple ship tracking.

Key words: Inland waterway,Closed-circuit television system (CCTV),Multiple ship tracking,Fuzzy SVM,Deformable part model

[1] 张国平.CCTV视频监控系统在内河重要河段中的应用[J].交通科技,2012(6):95-97
[2] 蒋恋华,甘朝晖,蒋旻.多目标跟踪综述[J].计算机系统应用,2010 (12):271-275
[3] 闫辉,许廷发,吴青青,等.多特征融合匹配的多目标跟踪[J].中国光学,2013,6(2):163-170
[4] 万琴,王耀南,袁小芳.改进联合概率数据关联的视频多目标快速跟踪[J].控制理论与应用,2012,28(10):1421-1430
[5] Osawa T,Sudo K,Arai H,et al.Monocular 3D tracking of multiple interacting targets[C]∥19th International Conference on Pattern Recognition,2008(ICPR 2008).IEEE,2008:1-4
[6] 杨伟,柴奇.基于mean-shift的多目标粒子滤波跟踪算法设计[J].航天电子对抗,2009,25(5):57-60
[7] 周维,许海霞,郑金华.基于RJMCMC的视觉多目标跟踪算法[J].计算机科学,2012,39(7):270-275
[8] Duan G,Ai H,Cao S,et al.Group tracking:exploring mutual relations for multiple object tracking [M].Computer Vision-ECCV 2012.Springer Berlin Heidelberg,2012:129-143
[9] Zhang L,Van Der Maaten L.Structure preserving object trac-king[C]∥2013 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Portland,OR,2013:1838-1845
[10] Dalal N,Triggs B.Histograms of oriented gradients for human detection[C]∥IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2005(CVPR 2005).San Diego,CA,USA,2005,1:886-893
[11] Lin Chun-fu,Wang Sheng-de.Fuzzy Support Vector Machines[J].IEEE Transactions on Neural Networks,2002,13(2):464-471
[12] Tsochantaridis I,Joachims T,Hofmann T,et al.Large marginmethods for structured and interdependent output variables[J].Journal of Machine Learning Research,2005,6(9):1453-1484
[13] Crammer K,Dekel O,Keshet J,et al.Online passive-aggressivealgorithms[J].The Journal of Machine Learning Research,2006,7:551-585
[14] Kalal Z,Mikolajczyk K,Matas J.Face-tld:Tracking-learning-detection applied to faces[C]∥2010 17th IEEE International Conference on Image Processing (ICIP).Hong Kong,2010:3789-3792
[15] 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

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!