基于改进霍夫森林框架的多目标跟踪算法

doi:10.11772/j.issn.1001-9081.2016.08.2311

计算机应用 ›› 2016, Vol. 36 ›› Issue (8): 2311-2315.DOI: 10.11772/j.issn.1001-9081.2016.08.2311

基于改进霍夫森林框架的多目标跟踪算法

高庆吉, 霍璐, 牛国臣

中国民航大学机器人研究所, 天津 300300

收稿日期:2016-01-19 修回日期:2016-03-23 出版日期:2016-08-10 发布日期:2016-08-10
通讯作者: 高庆吉
作者简介:高庆吉(1966-),男,黑龙江桦川人,教授,博士,主要研究方向:模式识别、人工智能、机器人自主导航与控制;霍璐(1987-),女,山西晋城人,硕士研究生主要研究方向:机器视觉、机器学习;牛国臣(1981-),男,山东菏泽人,副教授,博士研究生,主要研究方向:机器人自主导航与控制、智能控制。

Multi-target tracking algorithm based on improved Hough forest framework

GAO Qingji, HUO Lu, NIU Guochen

Robotics Institute, Civil Aviation University of China, Tianjin 300300, China

Received:2016-01-19 Revised:2016-03-23 Online:2016-08-10 Published:2016-08-10

摘要/Abstract

摘要： 针对单目视觉对多个相似的目标跟踪因遮挡等因素影响而失效的问题，提出一种基于改进霍夫森林框架的多目标跟踪算法。在将多目标跟踪问题归结为基于目标检测的轨迹关联过程基础上，通过引入在线学习霍夫森林框架将轨迹关联计算转化为最大后验概率（MAP）问题。通过在线采集多目标样本、提取目标外观和运动特征构建霍夫森林，进行森林训练得到轨迹关联概率，从而关联多目标轨迹；而引入低秩逼近Hankel矩阵进行轨迹校验，修复了误匹配的轨迹，改进了在线更新训练样本算法的效能。实验表明，轨迹误匹配率显著改善，能有效提高单目摄像机对多个相似目标有遮挡情况下跟踪的准确性和鲁棒性。

关键词: 多目标跟踪, 在线学习, 霍夫森林, 轨迹关联, Hankel矩阵, 相似目标

Abstract: For the failure of similar multi-target tracking with monocular vision caused by influence factors such as occlusion, a multi-target tracking algorithm based on improved online Hough forest tracking framework was proposed. Based on that, the tracking problem could be formulated as a detection-based trajectories association process, and the association calculation was formulated as a Maximum A Posteriori (MAP) problem with online learning Hough forest framework. Through online multi-objective samples collection and appearance and motion information extraction, a Hough forest was constructed to associate multi-target trajectories by training for track association probability. Low-rank approximation Hankel Matrix was employed to correct the trajectories, which modified associated errors and improved the efficiency of online update of the training set. Experimental results show that the trajectory miss match ratio is significantly decreased by the proposed method, and tracking accuracy and robustness of the monocular vision are effectively improved for similar or inter-occlusion targets.

Key words: multi-target tracking, online learning, Hough forest, trajectories association, Hankel matrix, similar-target

中图分类号:

TP391.41

高庆吉, 霍璐, 牛国臣. 基于改进霍夫森林框架的多目标跟踪算法[J]. 计算机应用, 2016, 36(8): 2311-2315.

GAO Qingji, HUO Lu, NIU Guochen. Multi-target tracking algorithm based on improved Hough forest framework[J]. Journal of Computer Applications, 2016, 36(8): 2311-2315.

参考文献

[1] BENFOLD B,REID I.Stable multi-target tracking in real-time surveillance video[C]//CVPR'11:Proceedings of the 2011 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2011:3457-3464.
[2] COLLINS R T.Multitarget data association with higher-order motion models[C]//CVPR'12:Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2012:1744-1751.
[3] HUANG C,WU B,NEVATIA R.Robust object tracking by hierarchical association of detection responses[C]//ECCV'08:Proceedings of the 10th European Conference on Computer Vision,LNCS 5303.Berlin:Springer-Verlag,2008:788-801.
[4] 胡昭华,邢卫国,何军,等.多通道核相关滤波的实时跟踪方法[J].计算机应用,2015,35(12):3544-3549.(HU Z H,XING W G,HE J,et al.Real-time object tracking method based on multi-channel kernel correlation filter[J].Journal of Computer Applications,2015,35(12):3544-3549.)
[5] 余礼杨,范春晓,明悦.改进的核相关滤波器目标跟踪算法[J].计算机应用,2015,35(12):3550-3554.(YU L Y,FAN C X,MING Y.Improved target tracking algorithm based on kernelized correlation filter[J].Journal of Computer Applications,2015,35(12):3550-3554.)
[6] KUO C-H,HUANG C,NEVATIA R.Multi-target tracking by on-line learned discriminative appearance models[C]//CVPR'10:Proceedings of the 2010 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2010:685-692.
[7] XIANG J,SANG N,HOU J.An online learned Hough forest model for multi-target tracking[C]//ICIP 2014:Proceedings of the 2014 IEEE International Conference on Image Processing.Piscataway,NJ:IEEE,2014:2398-2402.
[8] DICLE C,CAMPS O I,SZNAIER M.The way they move:tracking multiple targets with similar appearance[C]//ICCV'13:Proceedings of the 2013 IEEE International Conference on Computer Vision.Washington,DC:IEEE Computer Society,2013:2304-2311
[9] GALL J,LEMPITSKY V.Class-specific Hough forests for object detection[C]//CVPR 2009:Proceedings of the 2009 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2009:1022-1029.
[10] NEVATIA R.An online learned CRF model for multi-target tracking[C]//CVPR'12:Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2012:2034-2041.
[11] DING T,SZNAIER M,CAMPS O.Fast track matching and event detection[C]//CVPR'08:Proceedings of the 2008 IEEE Conference on Computer Vision and Pattern Recognition.Washington,DC:IEEE Computer Society,2008:1-8.
[12] DOLLÁR P,WOJEK C,SCHIELE B,et al.Pedestrian detection:an evaluation of the state of the art[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2012,34(4):743-761.

基于改进霍夫森林框架的多目标跟踪算法

Multi-target tracking algorithm based on improved Hough forest framework

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 11

编辑推荐

Metrics

[1]	郭一村, 陈华辉. 在线哈希算法研究综述[J]. 计算机应用, 2021, 41(4): 1106-1112.
[2]	何祥宇, 于斌, 夏玉杰. 粒子概率假设密度平滑器异常平滑问题的解决方法[J]. 计算机应用, 2020, 40(1): 299-303.
[3]	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.
[4]	吴婉婷, 朱燕, 黄定江. 在线投资组合选择的半指数梯度策略及实证分析[J]. 计算机应用, 2019, 39(8): 2462-2467.
[5]	郁顺昌, 黄定江. 基于自回归移动平均反转的在线投资组合选择[J]. 计算机应用, 2018, 38(5): 1505-1511.
[6]	杜启亮, 黎浩正, 田联房. 基于Adaboost和码本模型的手扶电梯出入口视频监控方法[J]. 计算机应用, 2017, 37(9): 2610-2616.
[7]	许浩锋, 凌青. 分布式在线交替方向乘子法[J]. 计算机应用, 2015, 35(6): 1595-1599.
[8]	朱红运, 王长龙, 王建斌, 马晓琳. 基于奇异值分解和Savitzky-Golay滤波器的信号降噪方法[J]. 计算机应用, 2015, 35(10): 3004-3007.
[9]	李唯实毛晓光谢建文. 学习者知识模型的在线学习算法[J]. 计算机应用, 2012, 32(02): 436-439.
[10]	孟繁锋屈桢深曾庆双李莉. isMHI结合CAMSHIFT的多目标跟踪[J]. 计算机应用, 2009, 29(06): 1673-1676.
[11]	王孝刚吴晓娟周鑫张小燕. 共享存储并行多目标跟踪[J]. 计算机应用, 2008, 28(9): 2303-2305.