Fusing Local and Global Features for Human Action Recognition

doi:10.16182/j.issn1004731x.joss.201807009

Abstract

Abstract: Recognizing human actions according to video features is an important research topic in a wide scope of applications. In this paper, we propose a robust human motion detection method that combines canny operator with the combination of local and global optic flow methods. Meanwhile, this paper presents a simple but efficient action recognition algorithm using fusion visual features. The mixed features fuse two action descriptors, namely centre distance-based space time interest point and curvature function-based Fourier descriptors. The frame-based human action classifier is developed using random forests algorithm. Experimental results show that the proposed method is accurate, efficient and robust compared with other supervised action recognition algorithms.

Key words: human action recognition, local features, global features, space-time interest points, Fourier descriptors, random forest

CLC Number:

TP391

Tang Chao, Zhang Miaohui, Li Wei, Cao Feng, Wang Xiaofeng, Tong Xiaohong. Fusing Local and Global Features for Human Action Recognition[J]. Journal of System Simulation, 2018, 30(7): 2497-2506.

References

[1] XIN M, ZHANG H, WANG H, et al.Arch: Adaptive Recurrent-Convolutional Hybrid Networks for Long-Term Action Recognition[J]. Neurocomputing (S0925-2312), 2016, 178: 87-102.
[2] MOHAMMADI E, WU Q M J, SAIF M. Human Activity Recognition Using an Ensemble of Support Vector Machines[C]//Proc of the 2016 International Conference on High Performance Computing & Simulation. NEW YORK, NY: IEEE, 2016: 549-554.
[3] YANG J-F, MA Z, XIE M.Multiscale Spatial Position Coding under Locality Constraint for Action Recognition[J]. Journal of Electrical Engineering & Technology (S1975-0102), 2015, 10(4): 1851-1863.
[4] LIU X, LI Y.Research on Human Action Recognition Based on Global and Local Mixed Features[C]// Proc of the 2014 International Conference on Mechatronics, Control and Electronic Engineering. NEW YORK, NY: IEEE, 2014, 113: 692-696.
[5] LI N, CHENG X, GUO H, et al.Human Action Recognition Based on Multi-Feature Fusion and Hierarchical BP-Ada Boost Algorithm[J]. Journal of Southeast University (Natural Science Edition)(S1003-7985), 2014, 44(3): 493-498.
[6] 郭利, 姬晓飞, 李平, 等. 基于混合特征的人体动作识别改进算法[J]. 计算机应用研究, 2013, 30(2): 601-604.
GUO L, JI X F, LI P, et al.Mixed Features Based Improved Human Action Recognition Algorithm[J]. Application Research of Computers, 2013, 30(2): 601-604.
[7] BRUHN A, WEICKERT J, SCHNORR C.Lucas/Kanade Meets Horn/Schunck: Combining Local and Global Optic Flow Methods[J]. International Journal of Computer Vision (S0920-5691), 2005, 61(3): 1-21.
[8] HARRIS C.A Combined Corner and Edge Detector[C]// Proc of the 4th Alvey Vision conference. 1988: 147-151.
[9] LAPTEV I, LINDEBERG T.Space-Time Interest Points[C]// Proc of the 9th IEEE International Conference on Computer Vision. NEW YORK, NY: IEEE, 2003: 432-439.
[10] DOLLAR P, RABAUD V, COTTRELL G, et al.Behavior Recognition Via Sparse Spatio-Temporal Features[C]// Proc of the 2nd Joint IEEE International Workshop on Visual Surveillance and Performance Evaluation of Tracking and Surveillance. NEW YORK, NY: IEEE, 2005: 65-72.
[11] ZAHN G T, ROSKIES R Z. Fourier Descriptors for Plane Closed Curves [J]. IEEE Transactions on Computers (S0018-9340), 1972, C-21(3): 269-281.
[12] YANG J, YANG J-Y, ZHANG D, et al.Feature Fusion: Parallel Strategy Vs. Serial Strategy[J]. Pattern Recognition (S0031-3203), 2003, 36(6): 1369-1381.
[13] LLINAS J, HALL D L.Introduction to Multi-Sensor Data Fusion[C]//Proc of the 1998 IEEE International Symposium on Circuits and Systems. NEW YORK, NY: IEEE, 1998, 6: 537-540.
[14] BREIMAN L.Random Forests[J]. Machine Learning (S0885-6125), 2001, 45(1): 5-32.
[15] LAPTEV I, CAPUTO B. Recognition of human actions [EB/OL].(2005-01-18)[2017-01-10]. http://www.nada.kth.se/cvap/actions/.
[16] GORELICK L, BLANK M, SHECHTMAN E, et al. Actions as Space-Time Shapes [EB/OL]. (2007-11-24) [2017-05-03]. http://www.wisdom.weizmann.ac.il/~vision/SpaceTimeActions.html.