EIV Model Adjustment Theory And Its Application In Optical Satellite Remote Sensing

The widely known error-in-variables(EIV)model is often encountered in the mathematical model,widely used in signal processing,communications engineering,computer vision and other fields.The simplest way to solve the parameter estimation of the EIV model is to use the least square(LS)solution ignoring the randomness of the coefficient matrix,but it is no longer the optimal solution.The total least squares(TLS)technology was proposed to deal with both the random observational vector and the random elements of the coefficient matrix.Now the theoretical investigation and application research points at home and abroad.Combined with the common data fitting problem,This paper summarizes the existing TLS technology and analysis how to rational use of TLS technology.Using the Ziyuan-3 satellite image data,the application in optical satellite remote sensing based on TLS has researched.The concrete aspects of this thesis involved are:Based on the correlations between the elements in the coefficient matrix and observation vector,the distribution of the variable elements in the coefficient matrix and the distribution of the variable elements in the EIV model,the TLS technology and its expansion algorithm are summarized,And analyzes its applicability,then gives a way to select the reasonable TLS algorithm for the EIV model in this paper.This paper discusses the there-phase fitting problem of spatial straight line,polynomial curve and circular curve.The problem is discussed from five aspects of mathematical expression form,model construction,algorithm selection,precision evaluation and case analysis.In this process,a general EIV fitting model of spatial straight line is established for the problem of spatial straight fitting,and the parameters are solved by using mixed structure total least square(MSTLS).The problem of solving the circular fitting parameters is based on the form of circular curve parameters,and the mathematic model of the error matrix can be effectively solved by using the TLS.Through the simulation data and examples found that the rational use of TLS can effectively solve the coefficient matrix error of the mathematical model of the parameters of the solution.The application of EIV model adjustment method in the field of surveying and mapping has focused on its application in optical satellite remote sensing,including the modulation and transfer function(MTF)obtaining of optical camera and the block adjustment based on rational function model.In terms of the MTF on-orbit assessment for optical satellite cameras,in this paper,we use the edge method based on TLS to obtain the position of the sub-pixel position,and finally get the value of MTF at each frequency.The experimental results show that the MTF on-orbit assessment based on TLS is feasible,which is more rigorous in theory than the conventional edge method without considering the matrix error of the coefficientmatrix.In the aspect of the remote sensing image block adjustment based on RFM,in this paper,the translation scaling model and the affine transformation model in the compensation model of the rational function model are studied.The compensation coefficient is solved by TLS and TS.The experimental results show that the error of the system can be used to improve the geometric accuracy of the image.TLS can be used to solve the compensation coefficients of the rational function model's compensation model.