基于涡旋电磁波雷达回波时频图像的动态手势识别  

Dynamic Gesture Recognition Based on Vortex Electromagnetic WaveRadar Echo Time-frequency Image

作　　者：王硕[1] 孙梦轩 杨志晓 王辉[1] 郑戍华[3] WANG Shuo;SUN Mengxuan;YANG Zhixiao;WANG Hui;ZHENG Shuhua(School of Energy and Intelligence Engineering,Henan University of Animal Husbandry and Economy,Zhengzhou 450044,China;Durham University,UK DH13LE;School of Automation,Beijing Institute of Technology,Beijing 100081,china)

机构地区：[1]河南牧业经济学院能源与智能工程学院,郑州450044 [2]英国杜伦大学,英国DH13LE [3]北京理工大学自动化学院,北京100081

出　　处：《火力与指挥控制》2022年第8期109-115,共7页Fire Control & Command Control

摘　　要：现有基于雷达的动态手势识别技术,通过分析传统电磁波回波中的多普勒效应获取目标特征,并利用深度学习模型获得分类结果。涡旋电磁波的波前相位呈螺旋状,其回波中含有线多普勒和角多普勒,蕴含更多的目标信息。基于涡旋电磁波的独特性质,提出了一种基于涡旋电磁波雷达回波时频图像的动态手势识别方法,将涡旋电磁回波时频图作为特征,输入到深度学习模型中,取得了较好的识别结果。In the existing radar based dynamic gesture recognition technology,target features are extracted by analyzing the Doppler effect in the traditional electromagnetic wave echo.And,the classification results are obtained using the deep learning model.Due to the spiral wave front phase of vortex electromagnetic wave radar,both of the linear Doppler and the angular Doppler are included in the echo,which contains more target information.Based on the unique properties of vortex electromagnetic wave,a dynamic gesture recognition method based on vortex electromagnetic wave radar echo time-frequency image is proposed in this paper.The vortex electromagnetic echo time-frequency image is input into the deep learning model as a feature,and the good recognition results are obtained.

关 键 词：涡旋电磁波雷达 深度学习 时频图像 动态手势识别 

分 类 号：TN957[电子电信—信号与信息处理]