计算机工程与应用 ›› 2018, Vol. 54 ›› Issue (5): 14-23.DOI: 10.3778/j.issn.1002-8331.1711-0418

• 热点与综述 • 上一篇    下一篇

基于CSI的行为识别研究综述

王正杰,杨伟丽,王  喆,侯玉珊,郭银景   

  1. 山东科技大学 电子通信与物理学院,山东 青岛 266590
  • 出版日期:2018-03-01 发布日期:2018-03-13

Survey of behavior recognition based on CSI

WANG Zhengjie, YANG Weili, WANG Zhe, HOU Yushan, GUO Yinjing   

  1. College of Electronic, Communications and Physics, Shandong University of Science and Technology, Qingdao, Shandong 266590, China
  • Online:2018-03-01 Published:2018-03-13

摘要: 近年来,无线信号在行为识别、定位和目标检测等方面的应用取得了一系列突破性的研究成果,对智能家居、视频教学、虚拟现实、家庭娱乐等具有很强的理论研究意义和实用价值,是智能人机交互的基本研究内容。首先回顾了传统的动作识别研究成果,并介绍了CSI的基本特性与获取方法,然后对动作检测、特征提取和动作分类等相关算法进行了归纳和分析,总结并讨论了基于CSI的相关领域取得的最新研究成果,最后指出了目前研究的局限性与可能的发展方向。

关键词: 信道状态信息(CSI), 行为识别, 特征提取, 分类算法

Abstract: In recent years, the application of wireless signals has made a series of breakthrough research results in the fields of behavior recognition, location and target detection. It has strong theoretical research significance and practical value for smart home, video teaching, virtual reality, family entertainment and so on, which is also the basic research content of intelligent human-computer interaction. This paper gives a brief review towards the research results of traditional motion recognition firstly, and introduces the basic characteristics and acquisition method of CSI. Algorithms used in recent researches on action recognition using CSI are thoroughly summarized and analyzed in three aspects:motion detection, feature extraction and classification of the action. The latest research achievements in related fields based on CSI are summarized and discussed. Finally, the limitations and possible directions of current research are pointed out.

Key words: Channel State Information(CSI), behavior recognition, feature extraction, classification algorithm