Multibaseline InSAR Data Processing Technology And Application Research

Multi-baseline interferometry synthetic aperture radar(InSAR) technique is an important trend in the development of InSAR.Its main content is the fusion of multiple channels’ data imaging and interferometric processing in order to reduce the difficulty of data processing, and to improve the digital elevation model(DEM) for accuracy.This paper aims at the study of multibaseline InSAR data processing technology.Tthe concrete research content and innovation are as follows:(1) First the multi-baseline maximum likelihood phase uwrapping algorithm is studied. Based on a single point of interferometric phase likelihood function to the multi-point joint likelihood function analysis, this paper introduces the principle of the algorithm,and error sources are analyzed.The characteristics of the algorithm and the suitable conditions are summarized as well.Simulation results show that the algorithm has the ability of dealing with complex terrain but sensitive to noise.Secondly the multi-baseline graph cut phase uwrapping algorithm is studied. From the establishment of the model,the determination of energy function to the introduction of graph cut theory,its principle of phase unwrapping is analyzed.Through the simulation experiment compared with the single baseline graph cut method and the maximum likelihood method,points out that the algorithm has higher uwrapping precision and antinoise performance under high abrupt change and big noise.(2) A phase unwrapping error compensation method based on impulse noise detection is proposed.Due to that the unwrapping erro of the above two algorithms has the similar characteristics with the impulse noise,by detecting the erro point of the initial unwrapping phase, replace the error point with several adjacent pixels of the median point for compensation.Simulation results show that the method has simple principle,fast alculating,and remarkable effect.(3) An improved method of multi-baseline least squares phase unwrapping algorithm is proposed.First the principle of multi-baseline least-square algorithm is studied, and points out that it still has the inherent defects of the least square methods.Secondly solving formula of the error of unwrapping phase is derived, which points out the technology of through solution of principal value of the error phase to get the real value and then adding to the initial unwrapping phase for compensation.The iterative compensating method with FFT least squares is clarified,which has the convergence and can improve the unwrapping precision.Finally the complementary technology is applied to multi-baseline least squares algorithm.The analysis of simulation experiments shows that the method makes full use of calculation quickness of FFT least-square,and significantly improves the precision by iterative compensation to the unwrapping erro,overcoming the shortcomings of the original algorithm.(4) The distributed parallel processing technology based on MATLAB cluster environment is studied,whice realizes the fast and efficient parallel processing of multi-baseline maximum likelihood algorithm.The feasibility of the multi-baseline phase unwrapping algorithms studied and improved in this papare is tested using airborne three-baseline measured data.The results show that the maximum likelihood method performs badly due to the poor noise resistance and the susceptibility to the change of ratio of the baseline length and the interference phase, so the practical is questioned;the unwrapping results of multi-baseline graph cut method is clearer in local details, and has better resistance to noise and better effects;the result of multi-baseline least-square method is too smooth,and has larger overall deviation,and its local information is not prominent;the improved multi-baseline least-square method reduces the unwrapping erro and smoothing effect,and improves unwrapping accuracy to a certain extent.