Method of Constructing Insar Phase Image Model Using Nonlinear Vector Surface

Abstract

The invention relates to radar and optical interference image processing, in particular to InSAR (Interferometric Synthetic Aperture Radar) phase image processing. The method comprises the following steps: firstly, carrying out multi-level wavelet decomposition on a disperse InSAR phase image, and expressing a phase as a sum form of a low-frequency part and a high-frequency part; utilizing a maximum value of the high-frequency component of an estimation window internal phase point to determine the direction of a main vector, and smoothing the estimation window internal phase according to different module extremum changes of a signal and a residual point; and in the estimation window, solving a phase vector sum of each decomposition direction, and constructing a nonlinear vector curved surface model. The phase model obtained by a nonlinear part, which considers the undulation change of a testing area terrain factor, of the nonlinear phase vector curved surface can estimate the phase change tendency of the testing area, and assists in phase filtering and unwrapping, and the accuracy of a processing result is improved.

Description

technical field [0001] The present invention relates to radar and optical interference image processing, and more particularly to synthetic aperture radar interferometry (InSAR) phase image processing. Background technique [0002] InSAR technology uses the phase information carried by the side-looking radar echo signal to calculate the three-dimensional data of the ground. Obtaining the InSAR phase image from the initial radar data format SLC (Single-Look Complex Image), and then obtaining the final elevation data product requires registration of two images, complex conjugate multiplication, phase filtering, phase unwrapping, and baseline estimation. processing steps to obtain the digital elevation model. [0003] InSAR phase image is an intermediate product of SAR data processing, and it is also the basic data of image filtering and phase unwrapping. Since the subsequent steps need to be converted into a mathematical model for processing, constructing an accurate and effe...

AI Technical Summary

Problems solved by technology

[0013] (3) The characteristics of the linear model cause the estimation window to be small and fixed, and the calculation time is long, and the residual points cannot be effectively suppressed, resulting in a large model error in the low coherence area

However, the FFT method still cannot take into account the details of terrain and phase changes, and cannot be distinguished according to the line-of-sight, distance, and azimuth directions of terrain changes, and the use of the mean value of the estimation window to estimate the phase is not accurate, and it is easy to cause error transmission

Images