Action Recognition Based on Motion Representing and Reconstructed Phase Spaces Matching of 3D Joint Positions

Abstract:

This paper presents an efficient and novel framework for human action recognition based on representing the motion of human body-joints and the theory of nonlinear dynamical systems. Our work is motivated by the pictorial structures model and advances in human pose estimation. Intuitively, a collective understanding of human joints movements can lead to a better representation and understanding of any human action through quantization in the polar space. We use time-delay embedding on the time series resulting of the evolution of human body-joints variables along time to reconstruct phase portraits. Moreover, we train SVM models for action recognition by comparing the distances between trajectories of human body-joints variables within the reconstructed phase portraits. The proposed framework is evaluated on MSR-Action3D dataset and results compared against several state-of-the-art methods.

• Periodical: Applied Mechanics and Materials (Volumes 333-335)
• Pages: 675-679
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.333-335.675
• Online since: July 2013
• Authors: Yan Tao Zhao, Bo Zhang, Xu Guang Zhang, Xiao Li Li, Mei Ling Fu, Si Yuan Feng
• Keywords: Body-Joints Motion Representation, Human Action Recognition, Nonlinear Dynamical Systems, SVM Models, Time-Delay Embedding

References

[1] Takens F. Dynamical systems and turbulence. Springer Berlin Heidelberg, 1981: 366-381.

[2] Poppe R. Image and vision computing, 2010, 28(6): 976-990.

[3] Wang J, Liu Z, Wu Y, et al. Mining actionlet ensemble for action recognition with depth cameras[C]. Computer Vision and Pattern Recognition (CVPR), 2012 IEEE Conference on. IEEE, 2012: 1290-1297.

DOI: 10.1109/cvpr.2012.6247813

[4] Martens J, Sutskever I. Learning recurrent neural networks with Hessian-free optimization[C]. Proceedings of the 28th International Conference on Machine Learning. Madison, WI: Omnipress, 2011, 46: 68.

[5] Ali S, Basharat A, Shah M. Chaotic invariants for human action recognition[C]. Computer Vision, 2007. ICCV 2007. IEEE 11th International Conference on. IEEE, 2007: 1-8.

DOI: 10.1109/iccv.2007.4409046

[6] Lewandowski M, Martinez-del-Rincon J, Makris D, et al. Temporal extension of laplacian eigenmaps for unsupervised dimensionality reduction of time series[C]. Pattern Recognition (ICPR), 2010 20th International Conference on. IEEE, 2010: 161-164.

DOI: 10.1109/icpr.2010.48

[7] Raptis M, Wnuk K, Soatto S. Spike train driven dynamical models for human actions[C]. Computer Vision and Pattern Recognition (CVPR), 2010 IEEE Conference on. IEEE, 2010: 2077-(2084).

DOI: 10.1109/cvpr.2010.5539885

[8] Han L, Wu X, Liang W, et al. Image and Vision Computing, 2010, 28(5): 836-849.

[9] Fraser A M, Swinney H L. Physical review A, 1986, 33(2): 1134.

[10] Li W, Zhang Z, Liu Z. Action recognition based on a bag of 3d points[C]. Computer Vision and Pattern Recognition Workshops (CVPRW), 2010 IEEE Computer Society Conference on. IEEE, 2010: 9-14.

DOI: 10.1109/cvprw.2010.5543273

[11] Ali S, Shah M. Pattern Analysis and Machine Intelligence, IEEE Transactions on, 2010, 32(2): 288-303.

[12] Meng H, Pears N, Bailey C. A human action recognition system for embedded computer vision applica-tion[C]. Computer Vision and Pattern Recognition, 2007. CVPR'07. IEEE Conference on. IEEE, 2007: 1-6.

DOI: 10.1109/cvpr.2007.383420

[13] Cao L. Physica D: Nonlinear Phenomena, 1997, 110(1): 43-50.

[14] K.N. Tran, I.A. Kakadiaris, S.K. Shah. Pattern Recognition, 2012, 45: 2562-2572.