EI能够检索到Accession number: 203Title: Action recognition based on motion representing and reconstructed phase spaces matching of 3D joint positions Authors: Zhao, Yan Tao1 ; Zhang, Bo1 ; Zhang, Xu Guang1; Li, Xiao Li1; Fu, Mei Ling1; Feng, Si Yuan1 Author affiliation: 1 Institute of Electrical Engineering of YanShan University, Qinhuangdao, 066004, China Source title: Applied Mechanics and Materials Abbreviated source title: Appl. Mech. Mater. Volume: 333-335 Monograph title: Measurement Technology and Engineering Researches in Industry Issue date: 2013 Publication year: 2013 Pages: 675-679 Language: English ISSN: 16609336 E-ISSN: 16627482 ISBN-13: 9783037857502 Document type: Conference article (CA) Conference name: 2013 2nd International Conference on Measurement, Instrumentation and Automation, ICMIA 2013 Conference date: April 23, 2013 - April 24, 2013 Conference location: Guilin, China Conference code: 99005 Sponsor: Korea Maritime University; Hong Kong Industrial Technology Research Centre; Inha University Publisher: Trans Tech Publications Ltd, Kreuzstrasse 10, Zurich-Durnten, CH-8635, Switzerland Abstract: This paper presents an efficient and novel framework for human action recognition based on representing the motion of human body-joints and the theory of nonlinear dynamical systems. Our work is motivated by the pictorial structures model and advances in human pose estimation. Intuitively, a collective understanding of human joints movements can lead to a better representation and understanding of any human action through quantization in the polar space. We use time-delay embedding on the time series resulting of the evolution of human body-joints variables along time to reconstruct phase portraits. Moreover, we train SVM models for action recognition by comparing the distances between trajectories of human body-joints variables within the reconstructed phase portraits. The proposed framework is evaluated on MSR-Action3D dataset and results compared against several state-of-the-art methods. © (2013) Trans Tech Publications, Switzerland. Number of references: 14 Main heading: Motion estimation Controlled terms: Gesture recognition - Image recognition - Nonlinear dynamical systems - Phase space methods - Time delay Uncontrolled terms: Human pose estimations - Human-action recognition - Motion representation - Pictorial structures - Reconstructed phase space - State-of-the-art methods - SVM model - Time delay embedding Classification code: 713 Electronic Circuits - 716 Telecommunication; Radar, Radio and Television - Information Theory and Signal Processing - 731 Automatic Control Principles and Applications - 921 Mathematics DOI: Database: Compendex Compilation and indexing terms, © 2013 Elsevier Inc.