山区星载SAR图像的预处理及软件平台搭建

发布时间：2018-05-17 03:22

本文选题：星载SAR + 预处理　；参考：《电子科技大学》2014年硕士论文

【摘要】：星载合成孔径雷达具有全天时、全天候的对地观测能力,在山区自然灾害的监测和预警上有着重要的应用,但是相干斑噪声的干扰以及由山区地形起伏导致的几何畸变现象将使星载SAR图像在这方面的应用大打折扣,本论文针对该这两个问题开展了对山区星载SAR图像预处理的相关研究工作,主要包括了以下几部分的内容:首先,阐述了相干斑的产生机理及其统计特性,基于完全发育的相干斑模型,具体论述了三种相干斑抑制算法,即多视处理,LEE滤波和GAMMA-MAP滤波,分别将这几种滤波算法应用于TerraSAR-X数据的体育场场景图和山区场景图中,基于目视评估准则和定量评估准则对所得结果进行分析比较,结果表明:这几种滤波算法都对SAR图像相干斑噪声有较好的抑制作用,其中GAMMA-MAP算法对SAR图像的结构保持较好。其次,对山区星载SAR图像的几何畸变原理进行了分析,简单介绍了两种几何校正方法:多项式校正法和共线方程校正法。重点阐述了基于SAR成像几何的距离多普勒定位模型,在该模型的基础上,分析研究了基于DEM的SAR图像模拟,以及利用模拟SAR图像的地形校正方法,本文以四川省阿坝州汶川县映秀镇区域作为实验区,试验的数据为TerraSAR-X SSC数据、SRTM的90米DEM以及机载系统的25米DEM,试验结果表明:DEM的分辨率是对SAR图像模拟的关键因素,而后向散射模型对其影响很小;对DEM数据的分形插值需要注意是否会引入错误信息;基于DEM的模拟SAR图像与真实图像的纹理特征非常一致;基于模拟图像的地形校正结果显示该方法良好地消除了山区星载SAR图像上的几何畸变,恢复了山体特征。最后,通过编程实现了上述SAR图像预处理方法,使用IDL语言搭建了山区星载SAR图像预处理软件平台,并展示了该软件平台的功能模块以及部分处理界面。
[Abstract]:Spaceborne synthetic Aperture Radar (SAR) has an all-day, all-weather Earth observation capability and has important applications in monitoring and early warning of natural disasters in mountain areas. However, the interference of speckle noise and geometric distortion caused by terrain fluctuation in mountainous areas will greatly reduce the application of spaceborne SAR images in this field. In order to solve these two problems, this thesis has carried out the research on the preprocessing of spaceborne SAR images in mountainous areas, which includes the following parts: firstly, the mechanism of speckle generation and its statistical characteristics are described. Based on the fully developed speckle model, three speckle suppression algorithms, I. e., multi-view processing le filter and GAMMA-MAP filter, are discussed in detail. These filtering algorithms are applied to the stadium scene map of TerraSAR-X data and the mountain scene map, respectively. The results are analyzed and compared based on visual evaluation criteria and quantitative evaluation criteria. The results show that these filtering algorithms can suppress speckle noise of SAR images well, and GAMMA-MAP algorithm can keep the structure of SAR images better. Secondly, the principle of geometric distortion of spaceborne SAR images in mountainous areas is analyzed, and two geometric correction methods, polynomial correction and collinear equation correction, are briefly introduced. The range Doppler location model based on SAR imaging geometry is expounded. Based on this model, the simulation of SAR image based on DEM and the terrain correction method of simulated SAR image are analyzed and studied. In this paper, the Yingxiu Town area of Wenchuan County, Aba Prefecture, Sichuan Province is taken as the experimental area. The experimental data are 90m DEM of TerraSAR-X SSC data and 25m DEM of airborne system. The experimental results show that the resolution of the SAR image is the key factor in the simulation of SAR images. The backward scattering model has little effect on it; the fractal interpolation of DEM data should pay attention to the introduction of error information; the simulated SAR image based on DEM is very consistent with the texture feature of real image; The results of terrain correction based on simulated images show that the proposed method can eliminate the geometric distortion of spaceborne SAR images and recover the mountain features. Finally, the SAR image preprocessing method mentioned above is realized by programming, and the software platform of spaceborne SAR image preprocessing in mountain area is built by using IDL language, and the function module and some processing interface of the software platform are shown.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.52

【参考文献】