Research On Geometry Modeling Method Of Multi-Dimensional SAR Imaging Data

Three-dimensional Synthetic Aperture Radar(SAR) not only has the capability in all-weather and all day-night observation as the tranditional SAR, but also can get the elevation information of target, so it is widely used in the detection of earth targets and acquisition of information. This paper focuses on the geometric modeling methods of the Interferometric SAR and downward-looking SAR data. The main work is as follows:(1) For the DEM data of InSAR, this paper proposes a new DEM simplification algorithm based on the curvature entropy and Gaussian Mixture Model(GMM). The DEM is generally very huge and should be simplified before modeling. The curvature and entropy can represent the complexity of terrain and can be used in the simplification of terrain. This paper employ the curvature entropy to simplify the data clustered by GMM, and analyze the results in quantitative and qualitative aspect. The experimental results show that the proposed algorithm outperforms traditional simplification algorithms in accuracy and terrain retention, making it more suitable for the complex terrain.(2) For the regular three-dimensional volume data obtained by the downward-looking SAR, this paper proposes a reconstruction method which can repair the holes automatically.Usually, there are holes in the data due to the imaging method and geometry angle of downward-looking SAR, and bring a chanllenge for the reconstruction. Firstly, the Otsu algorithm is employed in the denoising of SAR data. Then the Chan-Vese model of image processing method is applied to the geometry modeling of the downward-looking SAR data, and the distance and inner product functions are employed as the initial surface and contour indicator function. Compare with the marching cube algorithm, the algorithm proposed is more effective on the repair of holes.