Research On Efficient Multi-baseline InSAR Elevation Reconstruction

The multi-baseline InSAR elevation reconstruction technology studys in this paper is a terrain mapping technology of synthetic aperture radar developed on the basis of phase unwrapping of traditional single-baseline InSAR,which can not only reduce or overcome the influence of complex terrain conditions on the stability of the algorithm,but also improve the accuracy of elevation reconstruction.The main research contents are as follows:(1)The relevant principles of InSAR and multi-baseline InSAR are introduced,and then the traditional multi-baseline algorithm and relevant improved algorithm are briefly reviewed,and experimental simulation and performance analysis are carried out.(2)An extended kalman filter elevation reconstruction algorithm based on optimal gradient estimation is proposed.Firstly,the horizontal and vertical gradients of two groups of interferograms are obtained by two different algorithms: maximum likelihood gradient estimation algorithm and local gradient estimation algorithm.Then,the weighted likelihood function threshold judgment method is used to obtain the best estimated gradient of the current pixel.Then the integral path is guided by the quality map,and the EKF elevation is recursively estimated according to the path from high to low pixel quality until all pixels are reconstructed.The experiment proves that the algorithm can get better unwrapping results in the interferogram with high coherence.(3)A multi-baseline elevation reconstruction algorithm based on closed clustering combined with multi-baseline unscented information filtering phase unwrapping is proposed.Firstly,the closed-form robust clustering algorithm is used to estimate the terrain height roughly,and then the pixels of steep terrain is obtained by the residual handicap recognition algorithm and edge detection.Secondly,a multi-baseline unscented information filtering algorithm with good anti-noise performance is constructed,and a heapsort path-following unwrapping strategy with high efficiency is used to unwrap each pixel;Finally,the probability density function of different weights of each estimated elevation is used as the reliability of detection standard to judge the final estimated elevation of each pixel.The feasibility of this method is fully verified by the results of elevation reconstruction for different terrains.