基于TerraSAR-X影像的城市建筑物的三维重建

发布时间：2018-05-04 01:35

本文选题：TerraSAR-X + 三维重建　；参考：《长安大学》2015年硕士论文

【摘要】：合成孔径雷达(Synthetic Aperture Radar,SAR)是20世纪50年代末研制成功的一种主动式微波传感器。由于外界环境对SAR观测没有影响,因此,它可以在任何环境、任何时间进行对地观测,又由于SAR技术可以穿透植物和某些地面,所以,合成孔径雷达技术在测绘、资源勘探、地质以及环境监测等许多的领域中得到了广泛的应用。在当前世界各国城镇化速度进展加快、城市快速发展和不断扩张的形势下,高分辨率的SAR对地观测可以为城区结构的分析和城区实时变化的信息的提取供强有力的数据支撑。从城区环境中提取出建筑物在军事监视、追踪、任务规划、训练和损失估计方面具有重要的意义;在地图绘制、土地使用调查和城市规划等方面也得到广泛应用。特别地,随着近年来城市的经济发展,自然资源配置、环境保护等问题越来越受到关注,这使得对城市地理环境变化信息的需求急剧增加,其中一个重要的需求就是对建筑物的三维重建。但是,早以前的SAR影像的分辨率较低,使得利用SAR影像进行城市建筑物的三维重建的研究受到了限制。2007年德国TerraSAR-X高分辨率雷达卫星的发射成功标志着星载SAR进入了一个全新时代,它具有重访周期短、空间分辨率高、波长短、多极化、多模式、高精度监测地表微小形变等优势。所以研究利用高分辨率的TerraSAR-X影像进行城市建筑物的三维重建具有很强的理论意义和现实意义。本文主要研究了利用TerraSAR-X高分辨率影像对西安城区建筑物进行三维重建,并对生成的城市DEM的精度进行了评定。其主要内容包括:1、利用三幅2009年12月3日、2009年12月14日以及2009年12月25日获取的TerraSAR-X影像进行了西安市区建筑物三维重建的实验,利用两两干涉的方法,通过影像的配准、去平地效应、滤波、相位解缠、基线精化等一系列数据处理流程,获取了西安城区四个研究区域的DEM,每一个研究区域生成三幅DEM。论文对GAMMA软件处理TerraSAR-X影像时的一些参数的选择也做了简单的说明,最后对实验结果进行了分析。2、城区的建设使得地表发生了较大的形变,形变相位会造成生成的DEM产生误差;在城市建筑物密集、相邻建筑物边界高度跳跃性较大的区域也会使解缠结果出现误差,导致生成的DEM产生误差。实验采用中值滤波的方法对获得的DEM进行了滤波,取得了较好的滤波效果。3、为了减弱偶然误差的影响,将每个研究区域的三幅DEM进行了均值处理。4、将每个研究区域的路面看做一个零基准面,然后在每个研究区域的路面上选择30个随机点,求出这些随机点高程值的误差值,利用误差的均值和标准差这两个指标对生成的DEM做了精度评定。相对于平面,四个研究区域DEM精度最高是东郊,可以达到6米左右,最低的是城墙内,精度是10米左右。
[Abstract]:Synthetic Aperture radar (SAR) is an active microwave sensor developed in the late 1950s. Because the outside environment has no effect on SAR observation, so it can do earth observation in any environment, at any time, and because SAR technology can penetrate plants and some ground, so synthetic aperture radar technology is mapping, resource exploration, Geology and environmental monitoring are widely used in many fields. With the rapid development of urbanization and the rapid development and expansion of cities in the world, the high-resolution SAR Earth observation can provide strong data support for the analysis of urban structure and the extraction of urban real-time change information. The extraction of buildings from urban environment is of great significance in military surveillance, tracking, task planning, training and loss estimation, and is also widely used in mapping, land use survey and urban planning. In particular, with the economic development of cities, the allocation of natural resources, environmental protection and other issues have been paid more and more attention in recent years, which makes the demand for information on changes of urban geographical environment increase rapidly. One of the important requirements is the three-dimensional reconstruction of buildings. But the resolution of earlier SAR images was low, The success of German TerraSAR-X high-resolution radar satellite launch in 2007 marks a new era for spaceborne SAR, which has a short re-visit period and high spatial resolution. The advantages of short wavelength, multi-polarization, multi-mode, high-precision monitoring of surface micro deformation, etc. Therefore, it is of great theoretical and practical significance to study the 3D reconstruction of urban buildings using high resolution TerraSAR-X images. In this paper, the 3D reconstruction of Xi'an urban buildings with TerraSAR-X high resolution image is studied, and the accuracy of the generated city DEM is evaluated. The main contents include: 1, using three TerraSAR-X images obtained on December 3, 2009, December 14, 2009 and December 25, 2009, to carry out the experiment of 3D reconstruction of urban buildings in Xi'an, using the method of pairwise interference and registration of images. A series of data processing processes, such as levelling effect, filtering, phase unwrapping and baseline refinement, are used to obtain DEM of four study areas in Xi'an urban area, each of which generates three DEMs. In this paper, the selection of some parameters in processing TerraSAR-X image by GAMMA software is also explained simply. Finally, the experimental results are analyzed. 2. The construction of urban area makes the earth surface deformed, and the deformation phase will cause the generated DEM error. In the area where the buildings are dense and the height of adjacent buildings is high jump, the result of unwrapping will also appear error, which leads to the error of the generated DEM. In order to reduce the influence of accidental error, the method of median filter is used to filter the DEM in the experiment, and a better filtering effect is obtained. The average value of three DEM in each research area is treated by means. 4, the pavement of each research area is treated as a zero datum, then 30 random points are selected on the road surface of each research area, and the errors of the height values of these random points are obtained. The accuracy of the generated DEM is evaluated by the mean error and standard deviation. Compared with the plane, the DEM accuracy of the four study areas is the highest in the eastern suburb, which can reach about 6 meters, and the lowest is within the city wall, and the precision is about 10 meters.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：P237

【参考文献】