计算机科学 ›› 2014, Vol. 41 ›› Issue (10): 106-109.doi: 10.11896/j.issn.1002-137X.2014.10.024

• 2013’和谐人机环境联合学术会议 • 上一篇    下一篇

基于混合特征映射的密集场景运动模式分析

王冲鶄,赵旭,刘允才   

  1. 上海交通大学自动化系系统控制与信息处理教育部重点实验室 上海200240;上海交通大学自动化系系统控制与信息处理教育部重点实验室 上海200240;上海交通大学自动化系系统控制与信息处理教育部重点实验室 上海200240
  • 出版日期:2018-11-14 发布日期:2018-11-14
  • 基金资助:
    本文受973国家基础研究(2011CB302203),国家自然科学基金(61273285)资助

Motion Pattern Analysis in Crowded Scenes Based on Feature Maps

WANG Chong-jing,ZHAO Xu and LIU Yun-cai   

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

摘要: 密集场景分析是目前计算机视觉领域的热点和难点课题。提出了一种新的密集场景下集群目标运动模式的分析算法。该算法根据集群目标运动特有的规则获取集群目标的轨迹片段,对轨迹片段学习后验散度,得出产生式-判别式混合特征映射,该特征映射有效地将底层特征和运动模式的语义信息结合起来。通过对特征映射进行基于图模型算法的无监督分层聚类,挖掘出集群目标运动模式信息。实验结果准确地揭示了当前视频中运动模式的分布,证明了 该算法的有效性。

关键词: 密集场景分析,运动模式,轨迹片段,混合特征映射,自动聚类

Abstract: Crowded scene analysis is currently a hot and challenging topic in computer vision field.We proposed a novel approach to analyze motion patterns by clustering the hybrid generative-discriminative feature maps using unsupervised hierarchical clustering algorithm.The hybrid generative-discriminative feature maps are derived by posterior divergence based on the track-lets,which are captured by tracking dense points with three effective rules.The feature maps effectively associate low-level features with the semantically motion patterns by exploiting the hidden information in crowded scenes.Motion pattern analyzing is implemented in a completely unsupervised way and the feature maps are clustered automatically through hierarchical clustering algorithm building on the basis of graphic model.The experiment results precisely reveal the distributions of motion patterns in current crowded videos and demonstrate the effectiveness of our approach.

Key words: Crowded scene analysis,Motion pattern,Track-let,Feature maps,Automatic clustering

[1] Zhao T,Nevatia R.Tracking multiple humans in crowded environment [C]∥Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition.Princeton,NJ,USA,2004,2:II-406
[2] Khan Z,Balch T,Dellaert F.An MCMC-based particle filter for tracking multiple interacting targets [C]∥Computer Vision-ECCV 2004.Lecture Notes in Computer Science Volume 3024,2004:279-290
[3] Sugimura D,Kitani K M,Okabe T,et al.Using Individuality to Track Individuals:Clustering Individual Trajectories in Crowds Using Local Appearance and Frequency Trait [C]∥Proc.of IEEE Int’l Conf on Computer Vision.2009
[4] Ali S,Shah M.A lagrangian particle dynamics approach forcrowd flow segmentation and stability analysis [C]∥2007 IEEE Conference on Computer Vision and Pattern Recognition.IEEE,2007:1-6
[5] Ali S,Shah M.Floor fields for tracking in high density crowd scenes [C]∥European conference on computer vision.2008,2:1-14
[6] Ali S.Taming crowded visual scenes [D].University of Central Florida,2010
[7] Arandjelovic O,Zisserman A.Crowd detection from still images [C]∥British Machine Vision Conference.2008,1:523-532
[8] Mehran R,Moore B,Shah M.A streakline representation offlow in crowded scenes [C]∥Computer Vision-ECCV 2010.2010:439-452
[9] Wang X,Ma X,Grimson W E L.Unsupervised activity percep-tion in crowded and complicated scenes using hierarchical bayesian models [J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2009,31(3):539-555
[10] Wang C,Zhao X,Zou Y,et al.Detecting motion patterns in dynamic crowd scenes [C]∥2011 Sixth International Conference on Image and Graphics (ICIG).IEEE,2011:434-439
[11] Zhou B,Wang X,Tang X.Understanding collective crowd be-haviors:Learning a Mixture model of Dynamic pedestrian-Agents [C]∥2012 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).2012:2871-2878
[12] Saleemi I,Hartung L,Shah M.Scene understanding by statistical modeling of motion patterns [C]∥2010 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).IEEE,2010:2069-2076
[13] Shi J,Tomasi C.Good features to track [C]∥IEEE Computer Society Conference on Computer Vision and Pattern Recognition.IEEE,1994:593-600
[14] Sundaram N,Brox T,Keutzer K.Dense point trajectories byGPU-accelerated large displacement optical flow [C]∥2010 IEEE European conference on computer vision (ECCV).IEEE,2010:438-451
[15] Wang H,Klaser A,Schmid C,et al.Action recognition by dense trajectories [C]∥2011 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).IEEE,2011:3169-3176
[16] Martin R,Arandjelovic O.Multiple-object tracking in cluttered and crowded public spaces [C]∥Advances in Visual Computing.2010:89-98
[17] Arandjelovic O.Contextually Learnt Detection of Unusual Motion-Based Behaviour in Crowded Public Spaces [C]∥Computer and Information Sciences II.2012:403-410 (下转第143页)(上接第109页)
[18] Li X,Lee T S,Liu Y.Hybrid generative-discriminative classification using posterior divergence [C]∥2011 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).2011:2713-2720
[19] Wu Z,Li X,Zhao X,et al.Hybrid generative-discriminative recognition of human action in 3D joint space [C]∥Proceedings of the 20th ACM international conference on Multimedia.2012:1081-1084
[20] Rabiner L R.A tutorial on hidden Markov models and selected applications in speech recognition [J].Proceedings of the IEEE, 1989,77(2):257-286
[21] Hu M,Ali S,Shah M.Learning motion patterns in crowdedscenes using motion flow field [C]∥Proceedings of the 19th International Conference on Pattern Recognition (ICPR’08).Citeseer,2008
[22] Cho M,MuLee K.Authority-shift clustering:Hierarchical clustering by authority seeking on graphs [C]∥2010 IEEE Confe-rence on Computer Vision and Pattern Recognition (CVPR).IEEE,2010:3193-3200

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!