Human action recognition based on multi-scale feature maps from depth video sequences

Human action recognition is an active research area in computer vision. Although great progress has been made, previous methods mostly recognize actions from depth video sequences at only one scale, and thus they often neglect multi-scale spatial changes that provide additional information practical applications. In this paper, we present a novel framework with mechanism to improve scale diversity of motion features. We propose feature map called Laplacian pyramid images(LP-DMI). First, employ images (DMI) as the templates generate static representation actions. Then, caculate LP-DMI enhance dynamic motions reduce redundant human bodies. further extract multi-granularity descriptor LP-DMI-HOG more discriminative Finally, utilize extreme learning machine (ELM) for classification. The proposed method yeilds accuracy 93.41%, 85.12%, 91.94% on public MSRAction3D, UTD-MHAD DHA dataset. Through extensive experiments, prove our outperforms state-of-the-art benchmarks.