Human Actions Recognition Using Improved MHI and 2-D Gabor Filter Based on Energy Blocks

DOI

10.2991/icaita-18.2018.2How to use a DOI?

Keywords

human actions recognition; motion-history image (MHI); 2-D Gabor feature; adaboost

Abstract

The concept of motion-history image (MHI) is widely adopted by many researchers to solve problems of human actions recognition. An improved MHI with only one parameter is proposed in this paper, it is easier to be implemented and can retain more effective movement information compared with the original method. Furthermore, two-dimensional (2-D) Gabor feature based on energy blocks (EB-Gabor) is proposed to encode the texture information of MHI. The 2-D Gabor feature with high dimension is divided into multiple energy blocks and then the energy features of these energy blocks can be obtained. The energy features are served as the input of Adaboost. Experimental results on public benchmark KTH video database demonstrate the superiority of the proposed method over the state-of-the-art action recognition approaches.

Copyright

© 2018, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Junfeng Sun
AU  - Hongji Xu
AU  - Yingming Zhou
AU  - Lingling Pan
AU  - Feifei Li
AU  - Min Chen
PY  - 2018/03
DA  - 2018/03
TI  - Human Actions Recognition Using Improved MHI and 2-D Gabor Filter Based on Energy Blocks
BT  - Proceedings of the 2018 2nd International Conference on Artificial Intelligence: Technologies and Applications (ICAITA 2018)
PB  - Atlantis Press
SP  - 5
EP  - 8
SN  - 1951-6851
UR  - https://doi.org/10.2991/icaita-18.2018.2
DO  - 10.2991/icaita-18.2018.2
ID  - Sun2018/03
ER  -