基于航空LiDAR数据提取建筑物边界的方法研究

Research on Building Outline Extraction Method from Airborne LiDAR Data

激光测距是一项高效、精确的实现地表信息采集的技术。城市区域的高密度激光点云数据的获取使提取兴趣目标对象(建筑物、植被等)成为可能。本文提供一种利用航空激光雷达(LIDAR)数据提取建筑物边界的有效方法,该方法首先通过基于曲面拟合的滤波算法销减地形的影响,然后利用地物的纹理信息区分建筑物与非建筑物要素,最后通过一系列后处理,如小区域去除、矢量化等,最终提取出所需要的建筑物边界。实验表明本文提供的方法用于建筑物边界的提取效果理想。

Laser scanning is a fast and precise technique for resampling the earth into irregular pattern. The large number of laser points hitting planar facades in urban areas makes it possible to extract objects （buildings,vegetation, etc. ） in the areas. This paper presents a new approach to extract building outlines. First of all, filter method based on surface fitting is adopted to eliminate the effects of terrain. Then, texture characters are used to distinguish building from non-building ob- jects. Finally, through a series of post--processing, such as small region removal, vectorization, building outlines are detec ted. Experiments show the method presented in this paper is a reliable method to extract building outlines.