噪声稳健的高分辨距离像雷达目标识别方法

项目名称： 噪声稳健的高分辨距离像雷达目标识别方法

项目编号： No.61271024

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 无线电电子学、电信技术

项目作者： 杜兰

作者单位： 西安电子科技大学

项目金额： 80万元

中文摘要： 高分辨距离像（HRRP）提供了目标散射点沿距离方向的分布特征，对目标识别非常关键。因此，基于HRRP的雷达自动目标识别（RATR）技术成为了雷达信号处理和应用领域的研究热点。经过近30年的研究积累，基于HRRP的雷达目标识别的理论和方法研究已经比较深入。然而，如何提高低信噪比条件下识别方法的稳健性成为制约HRRP目标识别工程应用的主要“瓶颈”之一。雷达对空观测时我们考虑的噪声主要是接收机热噪声，目标回波信噪比和目标与雷达之间的距离直接相关，因此，提高低信噪比条件下RATR的性能对提高雷达的识别距离意义重大。本项目以低信噪比条件下雷达目标HRRP样本的RATR技术为研究背景，从数据预处理、特征提取和特征选择、统计建模三个层次展开研究。主要研究内容包括：1）基于特征变换和参数化散射中心建模的HRRP特征增强算法；2）噪声稳健的HRRP特征提取方法；3）HRRP统计模型的噪声自适应修正方法。

中文关键词： 雷达自动目标识别；高分辨距离像；特征增强；特征提取；统计建模

英文摘要： A high-resolution range profile (HRRP) contains the target scatterer distribution signature, which is very important for target recognition. Therefore, radar automatic target recognition (RATR) using HRRP has received intensive attention from the radar signal processing and its application community. Although some important theories and methods have been studied for target recognition based on radar HRRP during the past nearly three decades, how to improve the robustness of recognition methods under the test condition of low signal-to-noise ratio (SNR) is still a open problem which restricts the recognition techniques from engineering application. Since the noise considered here mainly refers to the heat noise from the radar receiver for aircraft-like targets under a look-up scenario, the SNR of target echo directly relates to the distance between the target and radar for a given noise level and radar transmitted power. Therefore, it is important to improve the recognition performance under low SNR for increasing the recognition distance between target and radar in the real application. This project will focus on target recognition based on radar HRRP under the test condition of low SNR. The noise robust recognition methods are researched from three aspects, i.e., data pre-processing, feature extraction and feat

英文关键词： Radar automatic target recognition (RATR)；High resolution range profile (HRRP)；Feature enhancement；Feature extraction；Statistical modeling