Fisher纹理分布下基于匹配滤波的极化SAR图像CFAR检测方法  

A New CFAR Detection Method of Polarimetric SAR Imagery Based on Matched Filter Under Fisher Texture

作　　者：张嘉峰 杨子渊 张鹏[1] 刘涛[1] ZHANG Jia-feng;YANG Zi-yuan;ZHANG Peng;LIU Tao(School of Electronic Engineering,Naval University of Engineering,Wuhan,Hubei 430030,China;No.92118 Unit of PLA,Zhoushan,Zhejiang 316000,China)

机构地区：[1]海军工程大学电子工程学院,湖北武汉430030 [2]中国人民解放军92118部队,浙江舟山316000

出　　处：《电子学报》2019年第12期2533-2543,共11页Acta Electronica Sinica

基　　金：国家自然科学基金(No.61372165,No.61771483)

摘　　要：在已有的极化合成孔径雷达(PolSAR)图像恒虚警(CFAR)检测方法中,存在着高分辨下杂波模型适用性差的难题。为此提出了一种Fisher分布下的CFAR检测方法,并定义虚警损失率(CFAR Loss,CL)以量化评估算法的恒虚警保持性能.首先,在乘积模型框架下引入Fisher纹理变量,推导出了多视极化匹配滤波(Multi-look Polarization Matched Filter,MPMF)检测量的概率密度函数(PDF).然后,对PDF积分得到了虚警概率的闭合解析式,并设计了CFAR检测流程.仿真数据和机载合成孔径雷达(Airborne SAR,AIRSAR)数据实验结果表明,与基于K分布、G0分布、Wishart分布的CFAR检测算法以及双参数恒虚警(two-Parameter CFAR,2P-CFAR)算法相比,新方法具有良好的恒虚警保持性能和检测性能,具有较强的鲁棒性,且运算时间未明显增加,相比于其他检测方法,品质因数(Figure of Merit,FoM)平均高出12.80%.There is a problem of the poor applicability of clutter models under high resolution with the existing constant false alarm rate(CFAR) detection methods in polarimetric synthetic aperture radar(PolSAR) imageries.To solve the problem,a CFAR detection method is proposed under Fisher texture,and the CFAR loss(CL) is defined to quantitatively evaluate the CFAR maintenance performance of detection methods.Firstly,the probability density function(PDF) of the multi-look polarization matched filter(MPMF) metric is derived based on product model combining the hypothesis of the Fisher texture.Secondly,the PDF of the MPMF metric is integrated,and the analytical expression of the false alarm rate is obtained.The process of the proposed CFAR detection method is also designed.Compared with other methods based on K,G0 and Wishart distribution,as well as the two-parameter CFAR(2 P-CFAR) detector via simulation data and airborne SAR(AIRSAR) data,the proposed method has good constant false alarm maintenance performance and detection performance,strong robustness and no significant increase in operation time.Compared with other detection methods,the figure of merit(FoM) is 12.80% higher on average.

关 键 词：极化合成孔径雷达(PolSAR) 恒虚警(CFAR) Fisher分布 虚警损失率(CFAR Loss CL) 多视极化匹配滤波(MPMF) 虚警概率 品质因数(FoM) 

分 类 号：TN95[电子电信—信号与信息处理]