Abnormal behavior analysis based on latent topic model

doi:10.7523/j.issn.1002-1175.2013.03.017

Abstract

Abstract:

Considering that most of abnormal behavior analysis methods do not consider the scene, we propose a method of abnormal behavior analysis based on latent topic model. Features of the scene are extracted and clustered to visual vocabulary by K-means. The visual vocabulary is divided into semantic topics to describe the scene by pLSA model. The descriptions for the scene and trajectory are combined to form feature vector which is modeled by CRF. Parameters of the CRF model are estimated by training, and abnormal behavior is analyzed by inference. The experiments show that abnormal behavior in a particular scene is accurately analyzed by this method.

Key words: latent topic model, abnormal behavior analysis, pLSA, CRF, global behavior

CLC Number:

TP391.4

ZHAO Long, GUO Li, XIE Jin-Sheng, LIU Hao, LU Hai-Xian. Abnormal behavior analysis based on latent topic model[J]. , 2013, 30(3): 387-393.

References

[1] Davis J W, Bobick A F. The representation and recognition of human movement using temporal templates[C]//IEEE International Conference on Computer Vision and Pattern Recognition. Puerto Rico: San Juan, 1997: 928-934.

[2] Yamato J, Ohya J, Ishii K. Recognizing human action in time-sequential images using hidden Markov model[C]//IEEE International Conference on Computer Vison and Pattern Recognition. USA: Champaign, 1992: 379-385.

[3] Zhong P, Wang R. Learning sparse CRFs for feature selection and classification of hyperspectral imagery[J]. IEEE Transactions on Geoscience and Remote Sensing,2008,46: 4186-4197.

[4] Xu J, Ye G T, Wang Y, et al. Online learning for PLSA-based visual recognition[J]. Lecture Notes in Computer Science, 2011,6493: 95-108.

[5] Neagoe V E, Mugioiu A C, Stanculescu I A. Face recognition using PCA versus ICA versus LDA cascaded with the neural classifier of concurrent self-organizing maps[C]//Proceedings of the 8th International Conference on Communications. 2010: 225-228.

[6] Wang Y, Mori. Human action recognition by semilatent topic models[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009, 31(10): 1762-1774.

[7] Mémin E, Pérez P. Dense estimation and object-based segmentation of the optical flow with robust techniques[J]. IEEE Transactions on Image Processing,1998,7: 703-719.

[8] Jain A K, Farrokhnia F. Unsupervised texture segmentation using Gabor filters[J]. Pattern Recognition,1991,24: 1167-1186.

[9] Hartigan J A, Wong M A. Algorithm AS 136: A k-means clustering algorithm[J]. Journal of the Royal Statistical Society, Series C (Applied Statistics),1979,28: 100-108.

[10] Bosch A, Zisserman A, Munoz X. Scene classification via pLSA[C]//Proceedings of the ECCV. Austria:Graz, 2006: 517-530.

[11] Aggarwal J, Ryoo M S. Human activity analysis: a review[J]. ACM Computing Surveys(CSUR), 2011, 43(3):16.

[12] Lafferty J, McCallum A, Pereira F C N. Conditional random fields: Probabilistic models for segmenting and labeling sequence data[C]//18th International Conference on Machine Learning 2001(ICML 2001). Massachusetts, USA, 2001.

[13] Akaike H. Information theory and an extension of the maximum likelihood principle[C]//Second International Symposium on Information Theory. USSR: Armenia, 1973: 267-281.

[14] Murphy K P, Weiss Y, Jordan M I. Loopy belief propagation for approximate inference: An empirical study[C]//Proceedings of the Fifteenth Conference on Uncertainty in Artificial Intelligence. USA: San Francisco, 1999: 467-475.

[15] Jian L, Shaogang G, Tao X. On-the-fly global activity prediction and anomaly detection[C]//Computer Vision Workshops (ICCV Workshops), 2009 IEEE 12th International Conference on. 2009: 1330-1337.