SAR图像与GIS配准及融合算法研究

发布时间：2018-05-13 20:59

本文选题：SAR图像 + GIS　；参考：《哈尔滨工业大学》2014年硕士论文

【摘要】：合成孔径雷达（SAR）对地观测时不受气候条件影响，并能够穿透植被冠层，其获得的图像轮廓清晰，结构信息保存完整，因此SAR在近代遥感领域中占有极其重要的地位。目前，地理信息系统（GIS）的数据多来自早期的实地勘测和基于光学图像的制作，前者往往更新周期长并且成本较大；后者由于经常受到天气情况的影响或山体阴影的遮挡，获得的数据无法对成像区域进行完备的描述。因此，将SAR图像作为一种数据源对GIS进行更新已成为目前解决GIS数据库更新周期长、完备性差等问题的首要选择。从另一方面讲，由于GIS库中包含了大量经过解译的地理信息，因此，利用GIS中的信息辅助SAR图像解译也已成为SAR图像智能解译领域中一个研究热点。本文主要针对SAR图像对GIS库的更新以及利用GIS数据辅助SAR图像实现目标的快速提取进行研究。其中对于实现第一条的首要条件是对SAR图像与GIS进行配准及融合，，因此，本文围绕SAR图像与GIS的配准及融合问题进行展开研究，为后续SAR图像对GIS库的更新提供先决条件。本文主要研究内容如下： (1)首先研究了SAR图像的特征提取技术，主要对居民区、机场、水域三类目标的提取技术进行了深入研究，旨在为后续SAR图像与GIS的配准做准备。其中，对于居民区的提取主要采取了灰度共生矩阵的方法进行纹理分析；对于机场的提取，提出了一种先区域分割后直线检测的分步提取算法；对于水域的提取，详细研究了一种基于主动轮廓的分割方法。 (2)对于SAR图像与GIS的配准问题，本文在分析基于SURF特征的匹配算法基础上提出了一种基于三角剖分去除误匹配点的算法，经实验表明，该算法可以有效地对匹配过程中出现的误匹配点进行剔除。 (3)在SAR图像与GIS配准的基础上，研究了利用GIS中的信息引导SAR图像中目标的快速提取，通过将GIS中包含的属性信息转换为可视化的图像实现了SAR图像中目标的定位和检测。 (4)在二维SAR图像与GIS配准融合的基础上，详细研究了DEM数据与三维GIS的配准，为后续DEM数据对三维GIS的更新以及变化检测做准备。
[Abstract]:Synthetic Aperture Radar (SAR) is not affected by climatic conditions and can penetrate the canopy of vegetation. The obtained images are clear in outline and complete in structure information. Therefore, SAR plays an extremely important role in the field of remote sensing in modern times. At present, GIS data are derived mostly from early field surveys, which tend to have long updating cycles and high costs, and optical image-based production; the latter are often obscured by weather conditions or mountain shadows. The obtained data can not fully describe the imaging area. Therefore, using SAR image as a data source to update GIS has become the first choice to solve the problems of long updating period and poor completeness of GIS database. On the other hand, because the GIS library contains a lot of interpreted geographic information, using the information in GIS to assist the interpretation of SAR images has become a research hotspot in the field of intelligent interpretation of SAR images. This paper focuses on the updating of GIS library by SAR image and the fast extraction of SAR image using GIS data. The first condition to realize the first item is to register and fuse SAR images with GIS. Therefore, this paper focuses on the registration and fusion of SAR images and GIS images, which provides a prerequisite for the subsequent SAR images to update the GIS library. The main contents of this paper are as follows: Firstly, the feature extraction technology of SAR image is studied, and the extraction techniques of residential area, airport and water area are studied in order to prepare for the registration of SAR image with GIS. Among them, for the extraction of residential areas, the grayscale co-occurrence matrix method is mainly used for texture analysis; for the extraction of the airport, a step extraction algorithm is proposed to segment the region before line detection; for the water area extraction, A segmentation method based on active contour is studied in detail. 2) for the problem of SAR image registration with GIS, this paper proposes a triangulation algorithm to remove mismatch points based on the analysis of SURF feature based matching algorithm. The experimental results show that, The algorithm can effectively eliminate the mismatch points in the matching process. 3) based on the registration of SAR image and GIS, this paper studies how to use the information in GIS to guide the fast extraction of the target in the SAR image. By converting the attribute information contained in GIS into the visual image, the location and detection of the target in the SAR image are realized. 4) based on the fusion of 2D SAR images and GIS, the registration of DEM data and 3D GIS is studied in detail to prepare for the update and change detection of 3D GIS by subsequent DEM data.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P208;TN957.52

【相似文献】