星载SAR影像直接与间接正射校正技术研究

发布时间：2018-03-01 07:00

本文关键词： 合成孔径雷达数字高程模型距离-多普勒定位模型直接正射校正间接正射校正　出处：《中国科学院大学(中国科学院遥感与数字地球研究所)》2017年硕士论文　论文类型：学位论文

【摘要】：由于合成孔径雷达(Synthetic Aperture Radar,SAR)侧视成像的成像原理,影像中存在叠掩、阴影和透视收缩等几何畸变,因此对SAR影像进行正射校正有重要的意义。就校正方法而言,对SAR影像进行正射校正的方法有很多,基于数字高程模型(Digital Elevation Model,DEM)对其进行间接正射校正是目前应用较多的一类方法,该类考虑到了地形的细节,校正精度较高。相比之下,SAR影像的直接地理定位方法应用的却比较少,由于其定位精度的不准确,及校正影像中出现大面积空洞像元的问题,严重影响了直接地理定位方法的应用。随着SAR卫星定姿、定轨越来越准确,SAR影像直接地理定位精度得到大大的提升,且其直观、简单、便于理解的特性日益受到人们的关注,但是,其定位精度究竟如何,较间接校正方法的差异有多大,校正结果能否直接用于其他应用是人们比较关注的问题;且由于直接校正结果影像中存在大量的空洞像元,这也从一定程度上制约了其应用,传统的邻域内插的方法只对平坦区域的零星空洞像元起作用,而对于山区等地形起伏较大的区域的大面积空洞像元就显得有些不足了,这就需要一种新的对大面积空洞像元的处理方法,使其能够生成与间接校正类似的校正结果。基于此,论文提出一种空洞像元处理方法,处理的原则是借附近像元的值进行填充,只是内插邻域会根据空洞区域的面积自适应变大,解决了直接正射校正影像中地形起伏较大的区域存在大面积空洞像元的问题。同时,给出了基于距离-多普勒(Range Doppler,RD)定位模型及3种不同的后向散射模型的影像模拟结果,对不同后向散射模型的模拟结果影像做比较分析,最终,根据模拟影像完成对SAR影像的校正工作。采用高空间分辨率的光学卫星影像做基准影像,分别对直接校正与间接校正方法的校正影像做精度评价,评估直接校正与间接校正结果的几何精度,并且对两种方法的校正影像做直观分析,评估直接校正空洞像元处理效果。结果表明,直接校正影像精度略低于间接校正精度,但两者差别不大,两种方法校正影像误差均在一个像元之内;直接校正空洞像元处理后校正影像得到很好的改善,生成与间接校正类似的校正结果。
[Abstract]:Because of the imaging principle of synthetic Aperture Radar synthetic Aperture Radar (SAR) side-view imaging, there are geometric distortions such as overlay, shadow and perspective contraction in the image, so it is very important to correct the SAR image by orthography. There are many methods for orthophoto correction of SAR image. Indirect orthophoto correction based on digital elevation model digital Elevation model (DEM) is one of the most widely used methods, which takes into account the details of terrain. The accuracy of correction is relatively high. In contrast, the direct geographical location method of SAR images is less used, because of the inaccuracy of its location accuracy and the problem of large area voids in the corrected images. Along with the attitude determination of SAR satellite, the accuracy of direct geographic location of orbit determination is greatly improved, and the characteristics of direct geographic location, such as intuitive, simple and easy to understand, have been paid more and more attention to. However, the location accuracy and the difference of indirect correction methods are different, and whether the correction results can be directly used in other applications is a problem that people pay more attention to, and because there are a large number of voids in the image of direct correction results, This also restricts its application to a certain extent. The traditional neighborhood interpolation method only works on scattered voids in flat areas, but it is not enough for large area voids in mountainous areas with large topography. This requires a new method to deal with large area voids, which can generate similar correction results as indirect correction. The principle of processing is to fill by the value of nearby pixel, but the interpolation neighborhood will adaptively enlarge according to the area of cavity area, which solves the problem that there is a large area of voids in the region with large terrain fluctuation in the direct orthographic correction image. At the same time, The image simulation results based on range Doppler range Doppler (RD) localization model and three different backscatter models are given. The simulation results of different backscatter models are compared and analyzed. According to the analog image, the correction of SAR image is completed. The optical satellite image with high spatial resolution is used as the reference image, and the accuracy of the correction image of direct correction method and indirect correction method is evaluated respectively. The geometric accuracy of direct correction and indirect correction is evaluated, and the direct correction image is analyzed intuitively to evaluate the effect of direct correction cavity pixel processing. The results show that the accuracy of direct correction image is slightly lower than that of indirect correction. However, the difference between the two methods is not obvious. The error of the two methods is within one pixel, and the correction image after direct correction of cavity pixel is well improved, and the correction result is similar to that of indirect correction.
【学位授予单位】：中国科学院大学(中国科学院遥感与数字地球研究所)
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.52

【相似文献】