Diagnosis of the Ill-condition of the RFM Based on Condition Index and Variance Decomposition Proportion (CIVDP)

35th International Symposium on Remote Sensing of Environment

地球科学;生态环境科学

Qing, Zhou^1,2 ; Weili, Jiao^1 ; Tengfei, Long^1

Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences, Beijing, China^1

Graduate University of Chinese Academy of Sciences, Beijing, China^2

关键词: Classical least squares;    Ill-conditioning;    Least squares estimation;    Multicollinearity;    Physical parameters;    Rational function model;    Reliable methods;    Variance decomposition;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/17/1/012220/pdf
DOI  :  10.1088/1755-1315/17/1/012220

学科分类：环境科学（综合）

来源: IOP

【 摘 要 】

The Rational Function Model (RFM) is a new generalized sensor model. It does not need the physical parameters of sensors to achieve a high accuracy that is compatible to the rigorous sensor models. At present, the main method to solve RPCs is the Least Squares Estimation. But when coefficients has a large number or the distribution of the control points is not even, the classical least square method loses its superiority due to the ill-conditioning problem of design matrix. Condition Index and Variance Decomposition Proportion (CIVDP) is a reliable method for diagnosing the multicollinearity among the design matrix. It can not only detect the multicollinearity, but also can locate the parameters and show the corresponding columns in the design matrix. In this paper, the CIVDP method is used to diagnose the ill-condition problem of the RFM and to find the multicollinearity in the normal matrix.