Unsupervised Motion Pattern Mining for Crowded Scenes Analysis = Unsupervised Motion Pattern Mining for Crowded Scenes Analysis

다국어 초록 (Multilingual Abstract)

Crowded scenes analysis is a challenging topic in computer vision field. How to detect diverse motion patterns in crowded scenarios from videos is the critical yet hard part of this problem. In this paper, we propose a novel approach to mining motion patterns by utilizing motion information during both long-term period and short interval simultaneously. To capture long-term motions effectively, we introduce Motion History Image (MHI) representation to access to the global perspective about the crowd motion. The combination of MHI and optical flow, which is used to get instant motion information, gives rise to discriminative spatial-temporal motion features. Benefitting from the robustness and efficiency of the novel motion representation, the following motion pattern mining is implemented in a completely unsupervised way. The motion vectors are clustered hierarchically through automatic hierarchical clustering algorithm building on the basis of graphic model. This method overcomes the instability of optical flow in dealing with time continuity in crowded scenes. The results of clustering reveal the situations of motion pattern distribution in current crowded videos. To validate the performance of the proposed approach, we conduct experimental evaluations on some challenging videos including vehicles and pedestrians. The reliable detection results demonstrate the effectiveness of our approach.

참고문헌 (Reference)

1 G. Brostow, "Unsupervised bayesian detection of independent motion in crowds" CVPR2006 1 : 594-601, 2006

2 L. Kratz, "Tracking with local spatio-temporal motion patterns in extremely crowded scenes" 693-700, 2010

3 T. Zhao, "Tracking multiple humans in crowded environment" 2 : II–406-, 2004

4 A. Chan, "Modeling, clustering, and segmenting video with mixtures of dynamic textures" 30 (30): 909-926, 2008

5 X. Wang, "Learning semantic scene models by trajectory analysis" 110-123, 2006

6 A. Chan, "Layered dynamic textures" 31 (31): 1862-1879, 2009

7 A. Chan, "Layered dynamic textures" 18 : 2006

8 S. Ali, "Floor fields for tracking in high density crowd scenes" 2 : 1-14, 2008

9 S. Wu, "Chaotic invariants of lagrangian particle trajectories for anomaly detection in crowded scenes" 2054-2060,2010, 2010

10 L. Kratz, "Anomaly detection in extremely crowded scenes using spatio-temporal motion pattern models" 1446-1453, 2009

11 Z. Khan, "An mcmc-based particle filter for tracking multiple interacting targets" 279-290, 2004

12 N. Vaswani, "Activity recognition using the dynamics of the configuration of interacting objects" CVPR 2003 2 : II–633-, 2003

13 R. Mehran, "Abnormal crowd behavior detection using social force model" 935-942, 2009

14 R. Mehran, "A streakline representation of flow in crowded scenes" 439-452, 2006

15 S. Ali, "A lagrangian particle dynamics approach for crowd flow segmentation and stability analysis" 1-6, 2007