地形辐射校正在极化SAR影像分类中的应用  

Application of terrain radiometric correction in polarimetric SAR imagery classification

作　　者：刘文祥[1,2] 张继贤[2] 黄国满[2] 赵争[2] 卢丽君[2] 魏钜杰[2] 

机构地区：[1]中国矿业大学环境与测绘学院,江苏徐州221116 [2]中国测绘科学研究院,北京100830

出　　处：《测绘科学》2014年第3期98-102,共5页Science of Surveying and Mapping

基　　金：国家863计划主题项目(2011AA120401)

摘　　要：由于SAR斜距成像几何方式及地形起伏的影响,原始SAR影像存在透视收缩、叠掩、阴影等严重的几何畸变和辐射畸变。其中,叠掩区具有强烈的后向散射回波,在极化SAR影像的分类研究中容易造成林地与居民地等地物混分,降低分类的精度。针对该问题,本文研究一种地形辐射校正方法,引入投影角计算后向散射系数γ0,有效地解决了地形起伏造成的辐射畸变问题。选取一景全极化Radarsat-2影像进行实验验证,分别对地形辐射校正前后的极化SAR影像进行了复Wishart监督分类。通过对分类结果的比较,表明经本文地形辐射校正方法处理后,极化SAR影像的分类精度得到了改善。Primitive SAR imagery exists serious geometric and radiometric distortions, e. g. fore shortening, layover and shadow, due to the effect of SAR imaging geometry and the topographic re lief. Among them, layover regions usually reflect strong baekscattering return. This phenomenon easily leads to the misclassification between forestland and habitation so that it decreases the classification accura- cy for polarimetric SAR imagery classification. This paper presented a terrain radiometric correction meth od to solve this problem, which uses the projection angle to compute the 70 backscattering coeffi- cient. Then one scene of Quad-polarization Radarsat-2 imagery was used to verify the validity of the meth- od, and a complex Wishart supervised classification method was applied to classify the polarimetric SAR images prior to and after terrain radiometrie correction, respectively. The classification results showed that the classification accuracy of polarimetric SAR images would be improved and increased from 87 % to 92% after processing with the proposed method.

关 键 词：极化SAR 地形辐射校正 投影角 复Wishart监督分类 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]