Human action recognition based on locality constrained linear coding and two-dimensional spatial-temporal templates

Abstract

Human action recognition is a challenging and active research area in computer vision. In this paper, we propose a simple yet effective method, called the locality-constrained linear coding (LLC) based two-dimensional spatial-temporal templates, to learn a discriminative representation for human action recognition. Our proposed method calculates twodimensional spatial-temporal templates from each human action sequence as the global features to describe the human action information. To describe the local detailed features better, we construct a multi-layer patches descriptor by spatial pyramid matching (SPM) method. And we encode the patches descriptor by using LLC algorithm to obtain a coding with underlying properties of better construction and local smooth sparsity for human action recognition. To evaluate the proposed method, we evaluate and compare our algorithm with some state-of-the-art methods on both Weizmann and DHA datasets. Experimental results show that our method outperforms some state-of-the-art methods.