基于Rician分布的SAR图像MRG目标提取算法  

Rician Distribution-Based MRG Target Extract Algorithm for SAR Image

作　　者：孔莹莹[1] 周建江[1] 张焱[1] 

机构地区：[1]南京航空航天大学信息科学与技术学院,南京210016

出　　处：《宇航学报》2011年第10期2200-2209,共10页Journal of Astronautics

基　　金：教育部研究生创新计划(CX08B_078Z)

摘　　要：针对SAR图像特有的乘性噪声和非恒虚警统计特性很难正确提取目标的问题,本文提出了MRG算法对SAR目标进行检测。首先用MRF-ICM算法对SAR图像进行后向散射系数的复原,再在复原后的幅度上叠加相位信息后引入Rician分布寻找自适应的最佳局域Gabor滤波器参数对SAR图像进行目标提取。和以往的算法相比,MRG算法充分利用了SAR图像幅度和相位中所携带的信息,能更准确地提取目标。实验表明:该方法能取得很好的目标检测效果,后期提取出的目标更符合实际目标形态,具有较强的通用性。Due to SAR image inherent characteristics of multiplicative noise and statistical properties of non-constant false-alarm rate（CFAR）,target extraction in synthetic aperture radar（SAR） images is extremely difficult.In this paper,a new target extractor for SAR images named MRG is proposed.First,the MRF-ICM model is used for recovering SAR backscattering coefficients.Second,by selecting the optimal filter parameters based on the outputs of Rician distribution,a Gabor filter is introduced to extract targets from the complex SAR images composed of recovered amplitudes superposed by phases.Compared with the previous methods,the MRG target extractor makes full use of the SAR amplitude and phase features simultaneously.Experimental results indicate that our method achieves satisfied effect of target detection,and the extracted target forms are more true and accurate.Meanwhile,the MRG extractor also has an advantage of strong versatility.

关 键 词：SAR图像 MRG检测器 图像复原 Rician分布 目标提取 

分 类 号：TN95[电子电信—信号与信息处理]