基于机载LIDAR点云数据的建筑物边缘提取

发布时间：2018-03-05 19:20

本文选题：LIDAR数据　切入点：DSM深度影像　出处：《长安大学》2013年硕士论文　论文类型：学位论文

【摘要】：上世纪末，机载LIDAR技术在三维地球空间信息的实时获取方面有了突飞猛进的发展，成为一种获取高时空分辨率地球空间信息的全新技术手段。由于能够自动的、快速精确的、不受天气影响的获取三维坐标，使人们从传统的人工单点地形表面数据获取变成连续自动的数据获取，为地形地物识别提供了新的数据源和信息源。因此，如何从LIDAR数据中自动化、智能化识别地物成为需要解决的关键问题和新的研究领域。本文以机载LIDAR点云数据为基础，通过一定的算法进行数据空洞填补，将离散LIDAR点云数据进行规则格网重采样，得到数字表面模型（DSM），接着按照高程进行灰度量化生成DSM深度影像，通过直方图阈值分割、小面积去除等操作对DSM深度影像进行预处理，得到独立的建筑物区域。最后运用Canny边缘检测算子检测建筑物的粗略边缘，利用hough变换获得建筑物边缘的直线段组，通过线段合并将本应该属于一条直线段的破碎小线段合并，通过直角约束对线段的倾角进行调整，获得了符合实际情况的建筑物边缘信息。本文以机载LIDAR点云为基础，在无其他辅助数据的情况下，运用ArcGIS、ERDAS等软件进行数字图像处理，运用MATLAB软件进行算法优化，实现了基于航空LIDAR点云数据的建筑物边缘检测。
[Abstract]:At the end of 0th century, airborne LIDAR technology has made rapid progress in the real-time acquisition of three-dimensional geospatial information, and has become a new technology for obtaining geospatial information with high spatial and temporal resolution. The acquisition of 3D coordinates, which is not affected by weather, makes people change from traditional artificial single-point topographic surface data acquisition to continuous automatic data acquisition, which provides a new data source and information source for terrain and ground object recognition. How to automate and intelligently identify ground objects from LIDAR data has become a key problem and a new research field. In this paper, based on airborne LIDAR point cloud data, a certain algorithm is used to fill the data holes, and discrete LIDAR point cloud data are resampled by regular grid resampling. The digital surface model (DSM) is obtained, and then the DSM depth image is generated according to the grayscale of height. The DSM depth image is preprocessed by histogram threshold segmentation, small area removal and so on. Finally, Canny edge detection operator is used to detect the rough edge of the building, hough transform is used to obtain the straight line segment group of the building edge, and the broken small line segment which is supposed to belong to a straight line segment is merged through line segment merging. The obliquity of the line segment is adjusted by the right angle constraint, and the building edge information is obtained according with the actual situation. Based on airborne LIDAR point cloud and without other auxiliary data, this paper uses ArcGIS ERDAS and other software to process digital image, and uses MATLAB software to optimize the algorithm to realize building edge detection based on airborne LIDAR point cloud data.
【学位授予单位】：长安大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：P23

【引证文献】