基于目标相干散射特性的极化SAR图像分解分类方法

基于对目标极化相干散射特性的分析,我们改进了Cloude和Lee等人提出的极化特征分解及非监督分类算法,以适应高分辨率极化SAR图像中复杂的地物细节特征。实验结果表明,相对传统方法,该方法更能够保留目标的细节特征、准确地估计目标极化相干矩阵,因此能够获得更好的分解分类结果。另外,该方法还具有较好的收敛性和鲁棒性。

Decomposition and Classification Method of Pol-SAR Based on the Analysis of Target Coherent Scattering Property

Based on the analysis of targets' polarimetric coherent scattering properties,we improved the polarimetric decomposition and unsupervised classification algorithm proposed by Cloude and Lee,making it adaptable to the complex terrain and target details in high-resolution Pol-SAR images.Our experiment results demonstrate that,compared with the conventional ones,the proposed method is able to further preserve the detailed target features,obtain much more accurate estimation of targets' polarimetric coherency matrix,and thus leading to a better decomposition and classification result.In addition,the proposed method is with good convergence and robustness.