超宽带雷达人体姿态识别方法研究  

Research on human posture recognition method using ultra⁃wideband radar

作　　者：李俊侠 张秦[2] 郑桂妹[2] LI Junxia;ZHANG Qin;ZHENG Guimei(Postgraduate School,Air Force Engineering University,Xi’an 710051,China;Air and Missile Defense College,Air Force Engineering University,Xi’an 710051,China)

机构地区：[1]空军工程大学研究生院,陕西西安710051 [2]空军工程大学防空反导学院,陕西西安710051

出　　处：《现代电子技术》2021年第19期1-7,共7页Modern Electronics Technique

基　　金：国家自然科学基金面上项目(61971438);陕西省青年托举人才项目(20180109);陕西省自然科学基金面上项目(2019JM⁃155)。

摘　　要：利用传统摄像头进行人体姿态识别时,容易存在视觉盲区和泄露个人隐私等问题。针对这些问题,利用超宽带雷达具有高分辨率、强穿透性和抗多径干扰等特性,可实现全天时、全天候地对人体姿态进行识别,且对环境要求低、准确率高、保密性好。首先结合超宽带雷达系统的特性,对常见的超宽带脉冲信号进行具体分析;然后创新性地将超宽带人体姿态识别的研究方法分为传统机器学习方法和深度学习方法两大类进行综述,对具有代表性的支持向量机(SVM)和卷积神经网络(CNN)进行原理分析和讨论,并给出了基于高斯核的SVM识别流程和改进的CNN网络结构;最后划分了四类超宽带雷达回波信号特征提取算法,提出超宽带雷达人体姿态识别的通用模型,并指出超宽带雷达人体姿态识别亟需解决的问题。When using the traditional cameras to recognize human posture,it is prone to generating visual blind spots and revealing personal privacy.Therefore,the ultra⁃wideband(UWB)radar which has characteristics of high resolution,strong penetration and anti⁃multipath interference is used to realize the recognition of human posture all⁃time and all⁃weather.It has low environmental requirements,high accuracy and good confidentiality.The common UWB pulse signals are analyzed in detail in combination with the characteristics of the UWB radar system.The research methods of human posture recognition by UWB radar are innovatively divided into the traditional machine learning methods and deep learning methods for review.The representative support vector machine(SVM)and convolutional neural network(CNN)are analyzed and discussed,and the SVM recognition process based on Gaussian kernel and the improved CNN network structure are given.Finally,four types of UWB radar echo signal feature extraction algorithms are divided,a general model of human posture recognition by UWB radar is proposed,and the problems that need to be solved urgently are pointed out.

关 键 词：超宽带雷达 人体姿态识别 特征提取 深度学习 支持向量机 卷积神经网络 最大池化 

分 类 号：TN951-34[电子电信—信号与信息处理]