Human action recognition based on chaotic invariants

Abstract

A new human action recognition approach was presented based on chaotic invariants and relevance vector machines (RVM). The trajectories of reference joints estimated by skeleton graph matching were adopted for representing the nonlinear dynamical system of human action. The C-C method was used for estimating delay time and embedding dimension of a phase space which was reconstructed by each trajectory. Then, some chaotic invariants representing action can be captured in the reconstructed phase space. Finally, RVM was used to recognize action. Experiments were performed on the KTH, Weizmann and Ballet human action datasets to test and evaluate the proposed method. The experiment results show that the average recognition accuracy is over 91.2%, which validates its effectiveness.

Keywords

chaotic system / action recognition / chaotic invariants / dynamic time wrapping (DTW) / relevance vector machines (RVM)

References

[1] RamassoE, PanagiotakisC, RombautM, PellerinD, Tziritasg. Human shape-motion analysis in athletics videos for coarse to fine action-activity recognition using transferable belief model [J]. Electronic Letters on Computer Vision and Image Analysis, 2009, 7(4): 32-50

[2] XiaL-m, WangQ, WuL-shi. Vision based behavior prediction of ball carrier in basketball matches [J]. Journal of Central South University, 2012, 19(8): 2142-2151

[3] AggarwalJ K, RyooM S. Human activity analysis: A review [J]. ACM Computing Surveys, 2011, 43(3): 1-43

[4] RonaldP. A survey on vision-based human action recognition [J]. Image and Vision Computing, 2010, 28(6): 976-990

[5] RodriguezM D, AhmedJ, ShahM. Action MACH: A spatio-temporal maximum average correlation height filter for action recognition [C]. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2008Anchorage, Alaska, USAIEEE Press1-8

[6] MaryamZ, HosseinE. Hierarchical human action recognition by normalized-polar histogram [C]. 2010 International Conference on Pattern Recognition, 2010Istanbul, TurkeyIEEE Press3720-3723

[7] AhmadM, LeeS. Human action recognition using shape and CLG-motion flow from multi-view image sequences [J]. Pattern Recognition, 2008, 41(7): 2237-2252

[8] AliS, MubarakS. Human action recognition in videos using kine-matic features and multiple instance learning [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2010, 32(2): 288-303

[9] SheikhY, SheikgM, ShahM. Exploring the space of a human action [C]. Proceedings of IEEE International Conference on Computer Vision, 2005Beijing, ChinaIEEE Press144-149

[10] BissaccoA, ChiusoA, MaY, SoattoS. Recognition of human gaits [C]. Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2001Kauai, HI, USAIEEE PressII-52-II-57

[11] BreglerC. Learning and recognizing human dynamics in video sequences [C]. Proceedings of IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 1997San Juan, Puerto RicoIEEE Press568-574

[12] RalaivolaL, d’Alche-BucF, PierreU, CurieM. Dynamical modeling with kernels for non-linear time series prediction [C]. Neural Information Processing Systems, 2003Vancouver, CanadaNIPS

[13] AliS, BasharatA, ShahM. Chaotic invariants for human action recognition [C]. International Conference on Computer Vision, 2007Rio de Janeiro, BrazilICCV1-8

[14] GongW-j, AndrewD B, XavierR F, GonzaleaJ. Automatic key pose selection for 3D human action recognition [J]. Computer Science, 2010, 61(6): 290-299

[15] XuF, KinM L, DaiQ-hai. Video-object segmentation and 3D-trajectory estimation for monocular video sequences [J]. Image and Vision Computing, 2011, 29(1): 190-205

[16] XiangB, LonginJ L. Path similarity skeleton graph matching [J]. Pattern Analysis and Machine Intelligence, 2008, 30(7): 1282-1292

[17] XiangB, LonginJ L, LiuW Y. Skeleton pruning by contour partitioning with discrete curve evolution [J]. IEEE Trans Pattern Analysis and Machine Intelligence, 2007, 29(3): 449-462

[18] KimH S, EykholtR, SalasJ D. Nonlinear dynamics, delay times, and embedding windows [J]. Physica D: Nonlinear Phenomena, 1999, 127(1): 48-60

[19] ParthapratimB, HironoriS, LawrenceR. MEAD. Lyapunov exponents and the natural invariant density determination of chaotic maps: An iterative maximum entropy ansatz [J]. Journal of Physics A: Mathematical and Theoretical, 2010, 43(12): 1-12

[20] RosensteinM T, CollinsJ J, CollinsC J, LucaD. A practical method for calculating largest lyapunov exponents from small data sets [J]. Physica D, 1993, 65(1): 117-134

[21] FathiA, MoriG. Action recognition by learning mid-level motion features [C]. In IEEE Conference on Computer Vision and Pattern Recognition, 2008Anchorage, Alaska, USAIEEE Press1-8