文献类型：期刊文献

英文题名：Region-based Mixture Models for human action recognition in low-resolution videos

作者：Zhao, Ying[1,2,3];Di, Huijun[1];Zhang, Jian[3];Lu, Yao[1];Lv, Feng[1];Li, Yufang[2]

第一作者：Zhao, Ying;赵英

通讯作者：Lu, Y[1]

机构：[1]Beijing Inst Technol, Beijing Lab Intelligent Informat Technol, Sch Comp Sci, 5 South Zhongguancun St, Beijing 100081, Peoples R China;[2]Beijing Union Univ, Teachers Coll, 5 Waiguanxiejie St, Beijing 100011, Peoples R China;[3]Univ Technol Sydney, Adv Analyt Inst, Sydney, NSW, Australia

第一机构：Beijing Inst Technol, Beijing Lab Intelligent Informat Technol, Sch Comp Sci, 5 South Zhongguancun St, Beijing 100081, Peoples R China

通讯机构：[1]corresponding author), Beijing Inst Technol, Beijing Lab Intelligent Informat Technol, Sch Comp Sci, 5 South Zhongguancun St, Beijing 100081, Peoples R China.

年份：2017

卷号：247

起止页码：1-15

外文期刊名：NEUROCOMPUTING

收录：;EI(收录号：20171503558392);Scopus(收录号：2-s2.0-85017164854);WOS:【SCI-EXPANDED(收录号:WOS:000401392200001)】；

基金：This work was supported by the 2014 General/Senior Foreign Visiting Scholars Training Project for Excellent Young Teachers in Beijing Colleges Governed by the Beijing Municipal Commission of Education (No. 067145301400), by the National Natural Science Foundation of China (No.61273273).

语种：英文

外文关键词：Low-resolution; Action recognition; Elastic motion tracking; Mixture model; Expectation Maximization (EM) algorithm

摘要：State-of-the-art performance in human action recognition is achieved by the use of dense trajectories which are extracted by optical flow algorithms. However, optical flow algorithms are far from perfect in low-resolution (LR) videos. In addition, the spatial and temporal layout of features is a powerful cue for action discrimination. While, most existing methods encode the layout by previously segmenting body parts which is not feasible in LR videos. Addressing the problems, we adopt the Layered Elastic Motion Tracking (LEMT) method to extract a set of long-term motion trajectories and a long-term common shape from each video sequence, where the extracted trajectories are much denser than those of sparse interest points (SIPs); then we present a hybrid feature representation to integrate both of the shape and motion features; and finally we propose a Region-based Mixture Model (RMM) to be utilized for action classification. The RMM encodes the spatial layout of features without any needs of body parts segmentation. Experimental results show that the approach is effective and, more importantly, the approach is more general for LR recognition tasks. (C) 2017 Published by Elsevier B.V.