基于Freeman 分解和雷达植被指数的极化SAR图像分类  

Polarimetric SAR image classification based on Freeman decomposition and radar vegetation index

作　　者：李成绕 贾诗超 薛东剑[2] LI Cheng-rao;JIA Shi-chao;XUE Dong-jian(School of History Geography and Tourism,Chengdu Normal University,Chengdu 611130,China;School of Earth Sciences,Chengdu University of Technology,Chengdu 610059,China)

机构地区：[1]成都师范学院史地与旅游学院,成都611130 [2]成都理工大学地球科学学院,成都610059

出　　处：《湖北农业科学》2021年第15期132-135,共4页Hubei Agricultural Sciences

基　　金：四川省科技计划项目(2019YJ0505);青海省科技厅重点项目(2019-SF-130)。

摘　　要：研究首先基于影像的相干矩阵提取特征参数,即雷达植被指数(RVI),再对影像的协方差矩阵进行Freeman分解,得到三种散射机制参数,分别为体散射、面散射和二面角散射。然后组合这些特征参数应用于支持向量机(SVM)中,对极化SAR图像进行分类,并与Wishart监督分类比较。结果表明,雷达植被指数有助于提高植被的分类精度,且该方法的分类精度明显高于Wishart监督分类。Based on the coherence matrix of the image,the characteristic parameters,namely the radar vegetation index(RVI),were extracted.Then the Freeman decomposition of the covariance matrix of the image was performed,and three scattering mechanism pa⁃rameters were obtained,which are body scattering,surface scattering and dihedral scattering.These parameters are then combined in⁃to a support vector machine(SVM)to classify the polarimetric SAR images and compare them with the Wishart supervised classifica⁃tion.The results show that the radar vegetation index can improve the classification accuracy of vegetation,and the classification accu⁃racy of this method was significantly higher than the Wishart supervised classification.

关 键 词：Freeman分解 雷达植被指数 支持向量机 Wishart监督分类 极化SAR 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]