数字高程（DEM）差异性检测及校验方法研究

发布时间：2019-05-07 04:08

【摘要】：数字高程数据差异性的来源可分为主观数据误差和客观数据误差,其中主观数据误差是指人为替换关键区域数据,客观数据误差是指测量时存在的粗大误差。在实际应用中,数据的差异性会给科学研究及应用带来很大的影响。本文的研究目的是得出不同数字高程之间差异性的表达,为DEM错误性的判断提供依据。ASTER GDEM(先进星载热量发射和反射辐射仪全球数高程模型)是目前为止覆盖最为广泛的、免费向用户提供的数字高程数据,具有较好的现势性,目前已经成为科学研究和地质应用领域重要的数据来源。本文以ASTER GDEM为待测用例,以德国宇航局获取的DLR DEM数据为假定真值。具体研究方法:第一是借助遥感图像匹配的思想,使用SURF算法与RANSAC算法结合的方法进行图像的匹配,在继承SURF算法的速度优势的前提下保证了图像匹配的精度。根据匹配所得的特征点的横纵坐标,计算其水平偏差并绘制ASTER GDEM与DLR DEM之间经纬度偏移量的三维图。第二是使用GIS专业软件ArcGis实现小范围的特征点的提取和优化,并根据生成的等高线和特征点的位置关系对提取的特征点进行匹配,计算配对特征点的水平距离,导入matlab中实现ASTER GDEM相对DLR DEM的水平偏移量的表示。实验结果表明:(1)经过RANSAC算法提纯后可以成功剔除误匹配的特征点,SURF算法与RANSAC算法结合的方式进行图像的匹配的结果较为准确。(2)通过对比两种研究方法所得ASTER GDEM相对DLR DEM的水平偏移量结果可以看出,两种方法提取所得到的特征点对的水平偏移量在0-30m和30m-60m之间所占的比例相近,其中基于图像匹配算法的实现方法所提取的特征点数量较少,且最大偏差量较大。使用ArcGis提取山顶点并手动匹配的方法所得配对点较为准确,但由于手动匹配工作量大,不适用于较大的实验区域。
[Abstract]:The difference of digital elevation data can be divided into subjective data error and objective data error, in which subjective data error refers to man-made replacement of key regional data, and objective data error refers to gross error in surveying. In practical application, the difference of data will bring great influence to scientific research and application. The purpose of this paper is to obtain the expression of the difference between different digital elevations, and to provide the basis for judging the error of DEM. ASTER GDEM (Advanced Spaceborne Heat Emission and reflection Radiometer Global Digital elevation Model) is by far the most widely covered. The digital elevation data provided to users free of charge has a good present situation and has become an important data source in the fields of scientific research and geological application. In this paper, ASTER GDEM is taken as the test case, and the DLR DEM data obtained by German space agency is taken as the assumed true value. The main research methods are as follows: firstly, with the help of the idea of remote sensing image matching, the combination of SURF algorithm and RANSAC algorithm is used for image matching, and the precision of image matching is guaranteed on the premise of inheriting the advantages of SURF algorithm. According to the transverse and longitudinal coordinates of the matching characteristic points, the horizontal deviation is calculated and the three-dimensional diagram of the longitude and latitude offset between ASTER GDEM and DLR DEM is drawn. The second is to use GIS professional software ArcGis to realize the extraction and optimization of small range of feature points, and to match the extracted feature points according to the position relationship between the generated contour lines and feature points, and to calculate the horizontal distance of paired feature points. Import matlab to implement the ASTER GDEM relative to DLR DEM horizontal offset representation. The experimental results show that: (1) the mismatched feature points can be removed successfully after purification of RANSAC algorithm. The results of image matching based on SURF algorithm and RANSAC algorithm are more accurate. (2) the horizontal offset of ASTER GDEM relative to DLR DEM can be seen by comparing the two methods. The ratio of horizontal offset between 0 ~ 30m and 30m-60m is similar to that of the two methods, and the number of feature points extracted by the method based on image matching algorithm is smaller, and the maximum deviation is larger. The matching point obtained by the method of extracting mountain vertices by ArcGis and manually matching is more accurate, but because of the large workload of manual matching, it is not suitable for large experimental areas.
【学位授予单位】：天津大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP751

【相似文献】