摘要
在分析了典型的极化目标分解和地物分类算法基础上,提出了融合Yamaguchi分解和H/α(H为散射熵,α为地物散射角)平面分解结果的迭代处理目标分类方法.首先,通过获取4种散射分量及地物的散射熵和散射角,结合6个参量,将极化合成孔径雷达图像中的地物初始分类;然后,利用相干散射矩阵服从Wishart分布的特性进行迭代,获得最终分类结果.实验结果证明,该算法提高了分类性能,运算量小,分类效果较好.
Based on the analysis of typical polarized target decomposition and classification,the paper proposes a new scheme for iterative classification of polarimetric SAR image,which blends the outcomes of Yamaguchi decomposition and H/α decomposition.This technique extracts four decomposition coefficients of four scattering mechanism components through Yamaguchi decomposition,the scattering entropy and angle through H/α decomposition first;then the initial classification of the POLSAR images is done by the combination of the 6 parameters mentioned above.The final result is obtained by iterative classification due to coherence scattering matrix following wishart distribution.The classification performance improved,better effectiveness and less amount of computation is demonstrated by the experimental results of polarimetric SAR data.
出处
《同济大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
2011年第9期1345-1349,共5页
Journal of Tongji University:Natural Science
基金
"十一五"国家科技支撑计划(2006BAJ09B01)
关键词
极化合成孔径雷达
图像分类
Yamaguchi分解
散射熵
散射角
polarized synthetic aperture radar
imaging classification
Yamaguchi decomposition
scattering entropy
scattering angle