基于方向性粗糙度特征的SAR目标检测算法  

Method for SAR Target Detection Using Directional Roughness Feature

作　　者：胡风明[1,2] 杨汝良[1] 

机构地区：[1]中国科学院电子学研究所,北京100190 [2]中国科学院研究生院,北京100190

出　　处：《系统仿真学报》2010年第1期236-240,共5页Journal of System Simulation

摘　　要：针对扩展分形(EF)特征检测SAR目标虚警率高的不足,提出了基于方向性粗糙度特征(Directional Roughness Feature,DRF)对SAR图像目标检测的算法。该算法用指数小波在一个尺度和任意一个方向θ(0 0<θ<900)上对SAR图像滤波,对滤波后图像应用能量关系函数求各像素点的DRF进行目标检测。针对X波段和Ka波段的SAR图像,确定了用该算法检测目标的最优参数。分别用该算法和EF特征方法对不同波段SAR图像进行目标检测,结果表明该算法具有检测虚警率低和目标空间可分辨性高的优点。As detected results false alarm rate high using Extended Fractal （EF） feature for SAR target detection, an efficient method was proposed based on Directional Roughness Feature （DRF） for SAR image target detection. The proposed method used exponential wavelet to filter SAR image on a scale and an arbitrary angle 0 （ 0~ 〈 0 〈 900 ）, and the DRF of the filtered image pixel obtained from the energy relation function was used to detect target. The optimal parameters of the method were determined for X-band and Ka-band SAR images. The comparison of detection results respectively using DRF and EF feature for different band SAR images shows the lower false alarm rate and higher target spatial resolution using the proposed DRF method.

关 键 词：合成孔径雷达(sAR) 目标检测 指数小波 方向性粗糙度特征(DRF) 扩展分形特征 

分 类 号：TN959.1[电子电信—信号与信息处理]