一种改进的雷达高分辨距离像统计识别方法  

A Modified Method for Statistical Identification of High Resolution Range Profiles

作　　者：王彩云[1] 许小剑[1] 毛士艺[1] 

机构地区：[1]北京航空航天大学电子信息工程学院,北京100083

出　　处：《信号处理》2009年第4期571-574,共4页Journal of Signal Processing

摘　　要：基于雷达目标一维高分辨距离像的统计目标识别中,需解决两大问题:其一是如何处理距离像对姿态敏感和平移敏感,其二是如何准确地描述距离像的统计特征。直接将一维距离像用于目标识别通常很难取得好的识别效果。本文将高斯混合模型(GMM)应用到空中目标高分辨一维距离像统计建模中,提出了一种改进的高斯混合模型模糊聚类分析方法并用于目标识别。与传统的k-means聚类算法的实验结果比较表明,该方法是有效、稳健的,在低信噪比条件下具有较好的识别效果。In the statistical target recognition based on radar high resolution range profiles （HRRP）, two challenging task are how to deal with the target-aspect and time-shift variation sensitivity of HRRP and how to accurately describe HRRPs statistical charac- teristics. It is usually very hard to obtain a satisfactory result by applying range profile to target identification directly. In this work, an improved fuzzy clustering analysis method based on Gaussian mixture model （GMM） is proposed and applied to the statistical modeling of HRRP. Compared with conventional k-means clustering algorithm, it is found that the current method can extract features independ- ent of target orientation. Simulation results demonstrate the effectiveness and robustness of the proposed method, especially when the SNR is low.

关 键 词：雷达目标识别 一维距离像 高斯混合模型 模糊聚类 

分 类 号：TN958[电子电信—信号与信息处理]