Abnormal Behavior Recognition Based on Spatio-temporal Context
2
조회
0
다운로드
2
인용
0
참고문헌
키워드
초록
This paper presents a new approach for detecting abnormal behaviors in complex surveillance scenes whereanomalies are subtle and difficult to distinguish due to the intricate correlations among multiple objects’behaviors. Specifically, a cascaded probabilistic topic model was put forward for learning the spatial contextof local behavior and the temporal context of global behavior in two different stages. In the first stage of topicmodeling, unlike the existing approaches using either optical flows or complete trajectories, spatio-temporalcorrelations between the trajectory fragments in video clips were modeled by the latent Dirichlet allocation(LDA) topic model based on Markov random fields to obtain the spatial context of local behavior in each videoclip. The local behavior topic categories were then obtained by exploiting the spectral clustering algorithm. Based on the construction of a dictionary through the process of local behavior topic clustering, the secondphase of the LDA topic model learns the correlations of global behaviors and temporal context. In particular,an abnormal behavior recognition method was developed based on the learned spatio-temporal context ofbehaviors. The specific identification method adopts a top-down strategy and consists of two stages: anomalyrecognition of video clip and anomalous behavior recognition within each video clip. Evaluation was performedusing the validity of spatio-temporal context learning for local behavior topics and abnormal behaviorrecognition. Furthermore, the performance of the proposed approach in abnormal behavior recognitionimproved effectively and significantly in complex surveillance scenes.
Abstract
This paper presents a new approach for detecting abnormal behaviors in complex surveillance scenes whereanomalies are subtle and difficult to distinguish due to the intricate correlations among multiple objects’behaviors. Specifically, a cascaded probabilistic topic model was put forward for learning the spatial contextof local behavior and the temporal context of global behavior in two different stages. In the first stage of topicmodeling, unlike the existing approaches using either optical flows or complete trajectories, spatio-temporalcorrelations between the trajectory fragments in video clips were modeled by the latent Dirichlet allocation(LDA) topic model based on Markov random fields to obtain the spatial context of local behavior in each videoclip. The local behavior topic categories were then obtained by exploiting the spectral clustering algorithm. Based on the construction of a dictionary through the process of local behavior topic clustering, the secondphase of the LDA topic model learns the correlations of global behaviors and temporal context. In particular,an abnormal behavior recognition method was developed based on the learned spatio-temporal context ofbehaviors. The specific identification method adopts a top-down strategy and consists of two stages: anomalyrecognition of video clip and anomalous behavior recognition within each video clip. Evaluation was performedusing the validity of spatio-temporal context learning for local behavior topics and abnormal behaviorrecognition. Furthermore, the performance of the proposed approach in abnormal behavior recognitionimproved effectively and significantly in complex surveillance scenes.
참고문헌 (27)
- C. Piciarelli, "On-line trajectory clustering for anomalous events detection", Pattern Recognition Letters, 27(15), pp.1835~1842 (2006)
- S. Kamijo, "Traffic monitoring and accident detection at intersections", IEEE Transactions on Intelligent Transportation Systems, 1(2), pp.108~118 (2000)
- Y. Sun, "Activity anomaly detection based on vehicle trajectory of automatic number plate recognition system", Journal of Computer Research and Development, 52(8), pp.1921~1929 (2015)
- C. Micheloni, "Exploiting temporal statistics for events analysis and understanding", Image and Vision Computing, 27(10), pp.1459~1469 (2009)
- C. Piciarelli, "Trajectory-based anomalous event detection", IEEE Transactions on Circuits and Systems for Video Technology, 18(11), pp.1544~1554 (2008)
- H. Y. Hu, "Spatial pattern recognition and abnormal traffic behavior detection of moving object", Journal of Jilin University(Engineering and Technology Edition), 41(6), pp.1598~1602 (2011)
- L. Zelnik-Manor, "Event-based analysis of video", Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition(CVPR), pp.123~130 (2001)
- H. Zhong, "Detecting unusual activity in video", Proceedings of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition(CVPR), pp.819~826 (2004)
- T. Xiang, "Video behaviour profiling and abnormality detection without manual labelling", Proceedings of the 10th IEEE International Conference on Computer Vision(ICCV), pp.1238~1245 (2005)
- Y. Wang, "Unsupervised discovery of action classes", Proceedings of 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition(CVPR), pp.1654~1661 (2006)
- T. Xiang, "Beyond tracking : modelling activity and understanding behavior", International Journal of Computer Vision, 67(1), pp.21~51 (2006)
- T. Xiang, "Video behavior profiling for anomaly detection", IEEE Transactions on Pattern Analysis and Machine Intelligence, 30(5), pp.893~908 (2008)
- M. Brand, "Discovery and segmentation of activities in video", IEEE Transactions on Pattern Analysis and Machine Intelligence, 22(8), pp.844~851 (2000)
- X. Wang, "Unsupervised activity perception by hierarchical Bayesian models", Proceedings of 2007 IEEE Conference on Computer Vision and Pattern Recognition, pp.1~8 (2007)
- M. Y. Yang, "Video event recognition and anomaly detection by combining Gaussian process and hierarchical Dirichlet process models", Photogrammetric Engineering &Remote Sensing, 84(4), pp.203~214 (2018)
- X. Wang, "Trajectory analysis and semantic region modeling using a nonparametric Bayesian model", Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition(CVRP), pp.1~8 (2008)
- X. Wang, "Trajectory analysis and semantic region modeling using nonparametric hierarchical Bayesian models", International Journal of Computer Vision, 95(3), pp.287~312 (2011)
- B. Zhou, "Random field topic model for semantic region analysis in crowded scenes from tracklets", Proceedings of 2011 IEEE Conference on Computer Vision and Pattern Recognition(CVPR), pp.3441~3448 (2011)
- R. Kaviani, "Incorporating fully sparse topic models for abnormality detection in traffic videos", Proceedings of 2014 4th International Conference on Computer and Knowledge Engineering(ICCKE), pp.586~591 (2014)
- L. Song, "Understanding dynamic scenes by hierarchical motion pattern mining", Proceedings of 2011 IEEE International Conference on Multimedia and Expo, pp.1~6 (2011)
- J. Li, "Scene segmentation for behaviour correlation", Computer Vision – ECCV 2008, pp.383~395 (2008), Springer
- J. Li, "Global behaviour inference using probabilistic latent semantic analysis", Proceedings of the British Machine Vision Conference (2008)
- C. C. Loy, "Detecting and discriminating behavioural anomalies", Pattern Recognition, 44(1), pp.117~132 (2011)
- D. M. Blei, "Dynamic topic models", Proceedings of the 23rd International Conference on Machine Learning, pp.113~120 (2006)
- D. M. Blei, "Latent Dirichlet allocation", Journal of Machine Learning Research, 3, pp.993~1022 (2003)
- T. Hospedales, "Video behaviour mining using a dynamic topic model", International Journal of Computer Vision, 98(3), pp.303~323 (2012)
- J. Li, "Learning behavioural context", International Journal of Computer Vision, 97(3), pp.276~304 (2012)