SAR目标鉴别中的变化特征提取算法研究

Study of Change Feature Extraction Algorithms in SAR Target Discrimination

合成孔径雷达自动目标识别(SAR ATR)算法一般分为三个步骤:预筛选、鉴别和分类,其中鉴别部分将基于预筛选提供的感兴趣区域(ROI)进行特征提取,根据提取的特征消除虚假目标,变化特征是用于消除固定强杂波形成虚假目标的重要特征。本文介绍了两种常用的变化检测算法,并对其在基于ROI变化特征提取中的适用性进行了分析,针对存在的问题,本文提出了三种适用的变化特征提取算法并进行了仿真试验,试验结果表明,本文提出的三种变化特征提取算法在不同检测条件下均保持了较好的稳健性,且滑窗平均相减法性能最优。

The synthetic aperture radar （SAR） automatic target recognition （ATR） algorithm is divided into three steps：a prescreener, a discriminator and a classifier. The discriminator extracts the features based on ROIs provided by the prescreener that are used to remove the false targets. The change feature is important to remove the false targets resulted from the strong stationary clutter. We introduce two change detection algorithms and analyze their inapplicabilities in change feature extraction based on ROI. In order to solve the inapplicabilities of above two algorithms,we propose three applicable algorithms of change feature extraction and make a simulation experiment based the proposed three algorithms. The experiment results indicate that the proposed algorithms are robust to extract the change feature in different experiment conditions where the subtraction algorithm of slip-window average.