基于微多普勒特征的人体运动辨识  

Human Motion Identification Based on Micro-doppler Features

作　　者：张翼[1] 朱玉鹏[1] 刘峥[1] 黎湘[1] 

机构地区：[1]国防科技大学空间电子信息技术研究所,长沙410073

出　　处：《西南大学学报（自然科学版）》2010年第3期156-161,共6页Journal of Southwest University(Natural Science Edition)

摘　　要：基于雷达的人体微动研究是近年发展起来的新技术,雷达对人体的探测有着其他传感器不具备的优势,随着人体微动雷达特征研究的不断深入,对人体运动参数的估计和运动模式的分辨开始受到重视.针对人体规范行走和跑步的辨识问题展开研究,分析人体跑步与规范行走在时空特性上的差别,通过微动测量雷达得到人体行走与跑步的实测数据,着重分析了人体跑步的微多普勒特征及其与规范行走微多普勒特征的异同,然后介绍了参数提取方法并提取躯干的特征参数将其应用于识别,最后以支撑矢量机为分类器对人体运动模式进行辨识.Radar based human micro-motion study is a new technology developed in recent years. Radar has an advantage over other sensors in detecting human body. As the study on radar characters of human micro-motion gets deeper, the estimation of human motion parameters and distinction of motion mode have attracted the attention of the researchers. The present paper studies the distinguishing of human normal walking and human running. It first analyzes the time and space feature differences between human running and normal walking. Through the data of walking and running field-measured by the micro-motion measurement radar, the authors analyze the similarity and difference between the micro-doppler feature of human running and that of normal walking. Then the paper introduces the parameter extraction method and extracted the feature parameters of human body and uses it in identification. Finally, human motion modes are identified with the support vector machine as the classifier in this paper.

关 键 词：微多普勒 人体步态辨识 雷达测量 支撑矢量机 

分 类 号：TN95[电子电信—信号与信息处理]