InSAR DEM Quality Improving Method Based On FBM

InSAR (Interferometric Synthetic Aperture Radar) is one of the key techniques to obtain high accuracy DEM (Digital Elevation Model) of the large area. But in some regions the DEM contains the spikes and the precision of the DEM cannot be further improved because of the limitations of the wrapped phase solution algorithms. The existing solution algorithms are mostly based on the following assumptions: continuous and gentle terrain, the terrain which is fully sampled by the interference phase signal, the absolute value of phase difference between adjacent pixels being less than π. The assumption is violated when the topography is discontinuous or when the wrapped phase is noisy or contains residues of phase images with poor coherence. The three conditions are the main reasons and lead to the wrapped phase with a large amount of spikes, and they are unable to overcome at present. Therefore this article focuses on the reconstruction of the wrapped phase. We know that InSAR image’s magnitude is proportional to the backscattered intensity and its phase is relate to the surface elevation, so this article using FBM (Fractional Brownian Motion) and InSAR image’s feature to provide a model that relates the phase and magnitude. At last this article uses the model and PolSAR data together to reconstruct the wrapped results. Then the Center-Point Displacement is proposed to interpolate the voids of the DEM produced by the reconstruct phase for improving the precision of the DEM.The main work and achievements of this dissertation are as follows:(1) The method of extracting FBM local parameter dynamically is proposed based on conventional methods. Base on the fractal analysis of the original data and the terrain effect on the fractal parameters, the window size is determined by the terrain and the local parameters are extracted according to the selected window size dynamically. It can obviously improve the precision of the fractal parameters and can fit the actual situation of the local terrain.(2) the wrapped phase is reconstructed by using the magnitude and the radiation characteristics of PolSAR (Polarimetric SAR) data. The scattering mechanism from fractal surface can provide the model to describe the relation between magnitude and phase components in InSAR image. Based on the provided model and polarization data, in the region of containing the residues, the amplitude and phase information can be used together to reconstruct the wrapped results. This method aims at solving the residual of the wrapped results. The using of magnitude and the radiation characteristics of PolSAR data can not only avoid to make the phase more smooth, but also increase the details of phase image and edge information.(3) Based on the DEM generated from the reconstructed phase and the Mid-Point Displacement method, the Center-Point Displacement is proposed. This method uses more information points to interpolate the spikes of the DEM generated from the reconstructed phase, and the final DEM fully displays the local details.