一种结合极化分解和多尺度MRF的极化SAR数据分类方法  

A classification method of PolSAR data by combining polarimetric decomposition and multi-scale markov random field

作　　者：段艳[1] 张剑清[1] 李畅[2] 孙明伟[1] 汪宏晨[3,4] 

机构地区：[1]武汉大学遥感信息工程学院,武汉430079 [2]华中师范大学城市与环境科学学院,武汉430079 [3]武汉大学 [4]武汉大学学报信息科学版编辑部,武汉430079

出　　处：《华中师范大学学报（自然科学版）》2014年第3期436-441,共6页Journal of Central China Normal University：Natural Sciences

基　　金：国家自然科学基金项目(41301519)

摘　　要：提出一种结合Pauli极化分解和多尺度马尔可夫随机场的极化SAR数据分类方法.该方法先将极化SAR数据进行Pauli极化分解来构造极化SAR特征数据,然后进行3层小波分解,最后进行尺度内和尺度间的迭代分类,获取极化SAR数据的分类结果.NASA/JPL数据的实验结果证明,无论是在地物轮廓保持性上还是地物细节描述性上,该方法效果都比较好.In this paper,we proposed a classification method of polarimetric SAR data by combining the Pauli polarization decomposition and multi-scale markov random field.We first created the features of polarimetric SAR data by doing Pauli polarization decomposition.Then,these features were decomposed by three-layer Haar wavelet decomposition.Finally,iterative classifications were carried out in a layer and between layers to produce classification result of the polarimetric data.The experimental results by using NASA / JPL data show that the proposed method have better performance both on keeping the contour of the features and describing the details of the features.

关 键 词：极化分解 多尺度 马尔可夫随机场 极化SAR 分类 

分 类 号：P407.8[天文地球—大气科学及气象学]