Studies On InSAR Imaging Algorithms

Supported by national natural science fundation of china, InSAR imaging techniques are studied in this dissertation. Some new and practical algorithms are proposed. Every procedure of InSAR image formation is expounded from the various aspects. In order to make a comparison between the different processing algorithms, almost all the algorithms are provided with the processing result of real data or simulated data.The chapters and sections of this dissertation are organized according to the routine of InSAR image formation.Chapter one provides an overview of the characteristics of InSAR, presents several examples of its application, reviews its history, summarizes its current state, and predicts its developing tendency. Finally, the main contents of this dissertation are outlined.In the second chapter, the basic principle and the practical problems of SAR are compactly explained at first. Then the basic principle of InSAR is introduced. The unwrapping phase of interferogram is divided into three parts: constant part, linear part and the part related with the height of terrain. And the conversion relation between the interferometric phase and the terrian height is established. Finally, the ability and limitation of an InSAR system originating from a specific SAR system is discussed. The precision of DEM affected by the precision of baseline, height of radar antenna and phase of interferogram is analyzed. The range of InSAR imaging region in determined by the relationship between range resolution on Earth surface and interferometer range-difference resolution, range resolution on Earth surface and the height resolution.In the third chapter, the concept of spatial decorrelation is explained. Because weighting function of range spectrum of InSAR images is not involved in the expression of spatial decorrelation coefficient at present, the relation between spatial decorrelation coefficient and various weighting function is studied. The formula of spatial decorrelation coefficient considering the weigthing process is derived. The effect of different weighting function on spatial decorrelation coefficient is analysed. And the equivalent bandwidth of interferogram is suggested to determine whether the fixed-band filter for spatial decorrelation is needed by the InSAR images. These have been verified and tested by real data processing. After that, various filtering algorithms for spatial decorrelation are analyzed and compared with each other. Finally, coregistration of InSAR images is introduced. A fast coregistration method combined the spectral diversity and scaling theorem for airborne InSAR is proposed.In the forth chapter, various interferogram filtering algorithms are discussed. The algorithm based on PD operator uses PD operator to estimate the higher-order polynomialmodel of interferograms. It can effectively work in the areas with low SNR. The filtered interferogram is weighted to weaken the mosaic effect. The morphological filter on multi-levels method is compared with common morphological filter. It not only has characteristics of common morphological filter, but also can reconstruct the phase of interferogram in a certain extent. The filter algorithm based on anisotropic diffusion formulation is used to filter the interferometric phase image. When selecting the exponential adjusting function and its parameter from the range of [0.2 0.3], the filter based on anisotropic diffusion formulation can get the best result. When wavelet shrink algorithm is used to filter interferogram, it cannot work effectively. But when a certain kind of wavelet shrink algorithm is modified, its can work effectively.In the fifth chapter, the problem of phase unwrapping is studied. Firstly branch cut method and it's implementation based on the Hungarian algorithm are discussed. Then the weighted and non-weighted least mean phase unwrapping algorithm, and the least mean phase unwrapping algorithm based on anisotropic evolution formulation are explained. After that, the minimum cost network flow phase unwrapping algori...