基于微多普勒特征的单人与小分队分类技术

人体行走是典型的非刚体运动，通常情况下每个人行走时的摆动周期不可能完全相同。通过提取目标频谱归一化幅度和、多普勒谱线数和谱宽的标准差3个典型特征，采用支持向量机( SVM)分类器，实现了短驻留时间条件下单人与多人的有效鉴别，平均识别率大于90%。雷达实测数据表明所提特征是有效并且稳健的。

Technology for classifying an individual soldier and a small group based on micro-Doppler features

Human walking is typical non-rigid motion and the swing periods for different men are usually not same.Through exacting three effective features,including the sum of the normalized magnitude,Doppler spectral line number and the standard deviation of spectrum width,a classifier of support vector machine(SVM) is used to distinguish an individual soldier and a small group for short dwell time.The average recognition rate is more than 90%.The experiments show that the proposed features are effective and robust.