基于机载激光雷达点云数据的DEM构建研究

发布时间：2017-12-28 10:35

本文关键词：基于机载激光雷达点云数据的DEM构建研究　出处：《山东师范大学》2017年硕士论文　论文类型：学位论文

【摘要】：机载激光雷达(LiDAR)测量技术已经成为测取高精度空间数据的有效方法。机载LiDAR的主动式遥感测量,对天气条件要求较低,获取的高时间分辨率与空间分辨率的点云数据能为高精度数字高程模型(DEM)的制作提供可靠的数据基础。现阶段,基于LiDAR点云数据的DEM构建研究中,存在的主要难点在于如何精确的获取LiDAR点云数据,并通过预处理与滤波处理剔除非地面点云数据,同时保留关键地形特征信息。本文针对LiDAR点云数据构建DEM存在的主要问题来开展研究。选取烟台市海岸地区为研究区,通过规划飞行线路、检校线路与飞行质量检验,首次获取了烟台市海岸地区高精度的点云数据。通过原始点云数据质量控制与大地定向完成了点云数据的预处理。在此基础上,使用移动窗口滤波算法与人机交互方法实现了点云数据的精确分类,剔除了非地面点云数据并得到了准确的点云数据。通过内插获取到DEM数据,结合网络实时动态定位(RTK)技术实地测得的地面高程数据对基于LiDAR点云数据获取的DEM数据进行精度评价,得到了一些有益的结论。本文研究的主要内容和成果有:介绍了机载LiDAR系统及其测量原理,详细介绍了系统各部分及其功能,分析了LiDAR点云数据的结构与特点,分析了数据误差来源以及降低误差的方法。并阐明了摄影测量与In SAR的原理与特点,与LiDAR系统进行了对比分析,研究了LiDAR点云数据的优势与不足。首次获取得到烟台市海岸地区的高精度机载LiDAR点云数据。对原始点云数据进行大地定向,通过多个坐标系的转换使原始点云数据具有了精确的WGS-84坐标,并通过基于高程统计的方法对点云数据进行初步降噪,去除了较为明显噪声数据,完成对点云数据的预处理。在此基础上,使用基于移动窗口的点云滤波算法对点云数据进行了滤波处理,多次变换窗口阈值,去除掉了大部分的噪声数据。之后使用人机交互的方式对点云数据进行后处理,将地面点云与非地面点云数据进行精确分类,完成了点云数据的进一步降噪处理,基本消除了噪声点云数据。首次使用高精度LiDAR点云数据构建烟台市海岸地区DEM。在处理海量的LiDAR点云数据时,为了避免数据冗余量多大和提高插值精度,采用不规则三角网(TIN)模型来模拟地表形态。使用基于Delaunay三角剖分算法对全部点云数据进行拟合,获取研究区内点云数据的不规则三角网。在此基础上,通过线性内插获取DEM。采用线性内插方法是因点云数据的空间密度极高,极大的降低了线性插值的误差;并且对于含有数百万三角形的三角网有较好的运算速度。详细介绍了网络实时动态定位(RTK)技术,并使用此技术在研究区获取了151个精确的高程数据。使用实测数据基于逻辑分析法与高程误差分析法,判断等高线逻辑关系与坡度异常值,分析高程中误差、平均绝对误差与平均误差,对LiDAR点云数据生成的DEM数据进行精度评价。实测高程与DEM高程的平均误差与平均绝对误差分别为0.105m与0.231m,整个地区的中误差为0.285m。
[Abstract]:Airborne laser radar (LiDAR) measurement technology has become an effective method for measuring high precision spatial data. The active remote sensing measurement of airborne LiDAR requires less weather conditions, and the acquired point cloud data with high temporal resolution and spatial resolution can provide reliable data basis for the production of high-precision digital elevation model (DEM). At present, the main difficulty of DEM construction based on LiDAR point cloud data lies in how to get LiDAR point cloud data accurately, and exclude non ground point cloud data through preprocessing and filtering processing, while preserving key terrain feature information. This paper studies the main problems existing in the construction of LiDAR point cloud data for the existence of DEM. From the coastal areas of Yantai city as the study area, by planning the flight line, line calibration and flight quality inspection, for the first time to get a high precision of point cloud data in coastal area of Yantai city. The pre processing of point cloud data is completed through the quality control and geodetic orientation of the original point cloud data. On this basis, we use mobile window filter algorithm and human-machine interaction method to achieve accurate classification of point cloud data, eliminate non ground point cloud data and get accurate point cloud data. Through interpolation to get the DEM data, combined with the ground elevation data measured by network real-time dynamic positioning (RTK) technology, we evaluate the accuracy of DEM data based on LiDAR point cloud data, and get some useful conclusions. The main contents and achievements of this paper are as follows: the airborne LiDAR system and its measurement principle are introduced, and the parts and functions of the system are introduced. The structure and characteristics of LiDAR point cloud data are analyzed, and the sources of data errors and the methods of reducing errors are analyzed. The principle and characteristics of photogrammetry and In SAR are clarified, and the LiDAR system is compared and analyzed, and the advantages and disadvantages of LiDAR point cloud data are studied. The high precision airborne LiDAR point cloud data are obtained for the first time in the coastal area of Yantai. For the Earth Orientation on the original point cloud data conversion through multiple coordinate system to the original point cloud data with the WGS-84 coordinates, and through the method of elevation based on the statistics of the point cloud data of noise reduction, removal of obvious noise data preprocessing on point cloud data. On this basis, we use point cloud filtering algorithm based on mobile window to filter the point cloud data, and transform the window threshold many times to remove most of the noise data. After that, post processing of point cloud data is done by means of human-machine interaction. The ground point cloud and non ground point cloud data are accurately classified, and the point cloud data is further de noised and processed. The noise point cloud data is basically eliminated. The first use of high precision LiDAR point cloud data to build the coastal area of Yantai DEM. When dealing with massive LiDAR point cloud data, the Irregular Triangulation Network (TIN) model is used to simulate the surface morphology in order to avoid data redundancy and improve interpolation accuracy. Using the Delaunay triangulation algorithm, all the point cloud data are fitted to obtain the Irregular Triangulation of the point cloud data in the study area. On this basis, DEM is obtained by linear interpolation. The linear interpolation method is because the density of point cloud data is very high, which greatly reduces the error of linear interpolation, and has a good speed for the triangulation network containing millions of triangles. The network real-time dynamic positioning (RTK) technology is introduced in detail, and 151 precise elevation data are obtained in the research area using this technology. Based on the measured data, based on the logical analysis and the height error analysis, we judge the logical relation between the contour line and the slope abnormal value, and analyze the height error, the mean absolute error and the average error. We evaluate the accuracy of the DEM data generated by LiDAR point cloud data. The average error and the mean absolute error of the measured elevation and the DEM elevation are 0.105m and 0.231m respectively, and the middle error of the whole area is 0.285m.
【学位授予单位】：山东师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN958.98

【相似文献】