[OPTICAL REVIEW Vol. 21, No. 3 (2014) 226-236]
© 2014 The Japan Society of Applied Physics

Spatially Regularized and Locality-Constrained Linear Coding for Human Action Recognition

Bin WANG1*, Wen GAI1, Shouchun GUO1, Yu LIU2, Wei WANG2, and Maojun ZHANG2

1Facility Design and Instrumentation Institute, China Aerodynamics Research and Development Center, Mianyang, 621000, P. R. China
2College of Information System and Management, National University of Defense Technology, Changsha, 410073, P. R. China

(Received May 9, 2013; Accepted February 8, 2014)

To reduce quantization error, preserve the manifold of local features, distinguish the ambiguous features, and model the spatial configuration of features for Bag-of-Features (BoF) model- based human action recognition, a novel feature coding method called spatially regularized and locality-constrained linear coding (SLLC) is proposed. The spatial regularization and locality constraint are involved in the feature coding phase to model the spatial configuration of features and preserve their nonlinear manifold. The action recognition experimental results on benchmark datasets show that SLLC achieves better performance than the state-of-the-art feature coding methods such as soft vector quantization, sparse coding, and locality-constrained linear coding.

Key words: image processing, video context analysis, action recognition, Bag-of-Features, local spatiotemporal feature, feature coding

*E-mail address: nudtwangbin@163.com