多极化层析SAR植被高度反演基线融合方法  

Fusion method for vegetation height inversion with multibaseline PolInSAR data

作　　者：邢成 杨健[1] 殷君君[2] 张瞻婕 Cheng XING;Jian YANG;Junjun YIN;Zhanjie ZHANG(Department of Electronic Engineering,Tsinghua University,Beijing 100084,China;School of Computer&Communication Engineering,University of Science and Technology Beijing,Beijing 100083,China)

机构地区：[1]清华大学电子工程系,北京100084 [2]北京科技大学计算机与通信工程学院,北京100083

出　　处：《中国科学：信息科学》2023年第3期606-621,共16页Scientia Sinica(Informationis)

基　　金：国家自然科学基金(批准号:62171023);高分辨率对地观测重大专项航空观测系统(批准号:30-H30C01-9004-19/21)资助项目。

摘　　要：植被高度反演属于森林定量遥感,在森林测绘、资源评估和生态监测等领域发挥着重要作用.本文基于多极化层析SAR数据进行基线融合植被高度反演,将森林散射相干模型推广到高维空间,通过复数域到幅度–相位域的转化降低模型非线性度,并提出了基于相干分布方差的广义幅相距离,进而发展了多基线联合参数反演方法.本文通过实测P波段极化层析SAR数据对所提方法进行了验证,同时将其与单基线反演法、基线选择法和欧式距离融合法进行了对比分析.实验结果表明,本文所提方法的反演结果与激光雷达获取的高度图相比,相关性更强,均方根误差更低,显著提高了植被高度反演的精度和稳定性.Vegetation height inversion is a mainstream research topic in quantitative forest remote sensing,which can play a pivotal role in forest mapping,resource assessment,and ecological monitoring.This article is focused on multibaseline fusion for vegetation height inversion through polarimetric tomographic synthetic aperture radar(Pol-tomoSAR)data.First,the forest scattering model is extended to high dimensions to adapt to multibaseline conditions.Second,the nonlinearity of the model is reduced by a transformation from the complex domain to the amplitude-phase domain.Finally,a fusion algorithm,which takes the distribution variance into consideration,is proposed based on the generalized distance.The proposed method is verified by P-band Pol-tomoSAR data and compared with the single baseline method,baseline selection method,and Euclidean fusion method.Experiments show that the proposed method produces more reliable results in terms of the root mean square error and correlation coefficient,thereby significantly improving the accuracy and stability of vegetation height inversion.

关 键 词：森林遥感 极化层析SAR 植被高度反演 多基线融合 

分 类 号：Q948[生物学—植物学] TN957.52[电子电信—信号与信息处理]