PolInSAR Coherence Region Modeling and Inversion: The Best Normal Matrix Approximation Solution
PolInSAR Coherence Region Modeling and Inversion: The Best Normal Matrix Approximation Solution
复制标题
DOI：
10.1109/tgrs.2014.2332553
复制
发表时间：
2015-02
影响因子：
8.2
通讯作者：
Yi Cui;Y. Yamaguchi;Hiroyoshi Yamada;Sang-Eun Park
中科院分区：
文献类型：
--
作者：
Yi Cui;Y. Yamaguchi;Hiroyoshi Yamada;Sang-Eun Park
In this paper, we address the theoretical aspects of coherence region modeling and its inversion methods for polarimetric interferometric synthetic aperture radar (PolInSAR). Instead of only considering the geometrical shape of the coherence region, we focus on directly modeling the whitened interferometric cross-correlation matrix of the PolInSAR data. In particular, we consider three classes of generic scattering models that respectively contain single, double, and triple phase centers within one resolution cell. We then demonstrate, for each class, that the modeled whitened interferometric cross-correlation matrix is normal. Based on this property, we propose, for each case, efficient algorithms to obtain the best normal matrix approximation solution for model inversion from the observed whitened interferometric cross-correlation matrix. In addition, we show how a simple and effective criterion can be designed from the derived solutions for model validation. Finally, we verify the solutions with both simulated and real data. The results prove the proposed method to be a very promising tool for PolInSAR applications.
