通信学报 ›› 2020, Vol. 41 ›› Issue (8): 43-54.doi: 10.11959/j.issn.1000-436x.2020141

• 学术论文 • 上一篇    下一篇

基于连续图像深度学习的Wi-Fi人体行为识别方法

周启臻1,邢建春1(),杨启亮1,韩德帅2   

  1. 1 陆军工程大学国防工程学院,江苏 南京 210007
    2 火箭军工程大学作战保障学院,陕西 西安 710025
  • 修回日期:2020-06-08 出版日期:2020-08-25 发布日期:2020-09-05
  • 作者简介:周启臻(1993- ),男,浙江温州人,陆军工程大学博士生,主要研究方向为无线感知和智能建筑|邢建春(1964- ),男,河北正定人,博士,陆军工程大学教授、博士生导师,主要研究方向为国防工程智能化和智能建筑|杨启亮(1975- ),男,河南信阳人,博士,陆军工程大学副教授、硕士生导师,主要研究方向为软件工程、信息物理融合系统和智能建筑|韩德帅(1990- ),男,山东聊城人,博士,火箭军工程大学讲师,主要研究方向为软件工程和国防工程智能化
  • 基金资助:
    国家重点研发计划基金资助项目(2017YFC0704100)

Sequential image deep learning-based Wi-Fi human activity recognition method

Qizhen ZHOU1,Jianchun XING1(),Qiliang YANG1,Deshuai HAN2   

  1. 1 College of Defense Engineering,Army Engineering University of PLA,Nanjing 210007,China
    2 College of Combat Support,Rocket Force University of Engineering,Xi’an 710025,China
  • Revised:2020-06-08 Online:2020-08-25 Published:2020-09-05
  • Supported by:
    The National Key Research and Development Program of China(2017YFC0704100)

摘要:

针对基于深度学习的Wi-Fi人体行为识别技术存在抗噪声能力弱、信号尺寸不兼容和特征提取不充分等问题,提出了一种基于连续图像深度学习的识别方法。首先把时变Wi-Fi信号重构为若干个连续图像帧,确保输入尺寸一致;进而设计低秩分解算法,对噪声湮没的关键运动信息进行分离;同时提出一种时间域和空间域信息融合的深度模型,自动捕捉变长图像序列的时空域特征,并在WiAR数据集和自主采集数据集上对所提方法进行验证。实验结果表明,所提方法平均识别精度分别为0.94和0.96,具备普适场景下的高精度和稳健性。

关键词: 行为识别, Wi-Fi信号, 深度学习, 图像识别, 低秩分解

Abstract:

For the problems existing in most of the researches,such as weak anti-noise ability,incompatible signal size and insufficient feature extraction of deep-learning-based Wi-Fi human activity recognition,a kind of sequential image deep learning-based recognition method was proposed.Based on the idea of sequential image deep learning,a series of image frames were reconstructed from time-varied Wi-Fi signal to ensure the consistency of input size.In addition,a low-rank decomposition method was innovatively designed to separate low-rank activity information merged in noises.Finally,a deep model combining temporal stream and spatial stream was proposed to automatically capture the spatiotemporal features from length-varied image sequences.The proposed method was extensively tested in WiAR dataset and self collected dataset.The experimental results show the proposed method could achieve the accuracy of 0.94 and 0.96,which indicate its high-accuracy performance and robustness in pervasive environments.

Key words: activity recognition, Wi-Fi signal, deep learning, image recognition, low-rank decomposition

中图分类号: 

No Suggested Reading articles found!