基于幅相一致性校正的稳健植被参数反演方法  

Forest Parameters Inversion Based on Nonstationarity Compensation and Mapping Space Regularization

作　　者：卢红喜[1] 宋文青[1] 李飞[1] 王英华[1] 刘宏伟[1] 保铮[1] 黄海风[2] 

机构地区：[1]西安电子科技大学雷达信号处理国家重点实验室,西安710071 [2]国防科学技术大学电子科学与工程学院,长沙410073

出　　处：《电子与信息学报》2015年第2期283-290,共8页Journal of Electronics & Information Technology

基　　金：国家自然科学基金(61271024;61201292;61201283);新世纪优秀人才支持计划(NCET-09-0630);全国优秀博士学位论文作者专项资金(FANEDD-201156);国家部级基金;中国航天科技集团公司航天科技创新基金;中央高校基本科研业务费专项资助课题

摘　　要：植被参数反演是极化干涉合成孔径雷达(Pol In SAR)的重要应用。传统反演方法未考虑观测样本数据幅度和相位的非平稳特性,以及观测信号非均匀分布对其散布区域线性变化主导方向估计的影响。针对这些问题,该文首先采用经过幅度和相位一致性校正的数据样本估计极化相干矩阵,提高了极化干涉复相干系数的估计性能,并提出了映射空间均衡化(MSR)处理技术以消除观测信号非均匀分布对主导方向提取的影响,通过引入主成分分析(PCA)方法进一步提高了参数反演算法的性能。利用欧空局(ESA)发布的软件PolSARPro仿真验证了该文方法在植被参数反演方面具有更好的稳健性和估计精度。Forest parameters inversion is an important application of Polarimetric Interference Synthetic Aperture Radar（Pol In SAR）. The traditional inversion method does not take into account the amplitude and phase non-stationary of observation, and its non-uniform distribution effect on the estimation of the principal linear change direction. Aiming at these problems, an amplitude and phase calibration approach is proposed to compensate the polarization coherence matrix nonstationarity, to enhance the performance of complex coherences estimation. Moreover, this paper develops a Mapping Space Regularization（MSR） technology which promises to be able to eliminate the non-uniform distribution effect of sample coherences on the linear variation of complex coherences. Based on MSR, the Principal Component Analysis（PCA） is further introduced to the linear variation model extraction. Processing results of ESA PolSARpro simulated data verify the better robustness and estimation accuracy of the proposal in forest parameters inversion.

关 键 词：极化干涉合成孔径雷达 植被参数反演 非平稳校正 映射空间均衡化 主成分分析 

分 类 号：TN957.51[电子电信—信号与信息处理]