机载LIDAR点云的城市平顶建筑物提取方法研究

Extraction Algorithm Study of Urban Flat-topped Buildings Based on Airborne LIDAR Point Cloud

本文以滤波后的机载LIDAR非地面点集作为数据源,提出了基于平面拟合及法向量的区域生长建筑物分类算法,剔除了大量的非建筑物点集,但在这些点云中仍存在少量非建筑物点,为了精确去除非建筑物点,分割出每个建筑物点集,提出了基于点集特征约束的建筑物分割算法,依据点数、点集离地面平均高度及平面面积特征分割每个建筑物,实验结果表明,该算法可完整去除非建筑物点,具有重要的应用价值。

Taking the non-ground point set of the airborne LIDAR as the data source,the paper proposes a building classification algorithm of regional growth method based on plane fitting and normal vector.Although the algorithm preliminarily eliminated non-building point sets,there were a small amount of non-building points.In order to remove non-building points completely and split every building point sets,the building segmentation algorithm based on point set’s features was presented.According to the number of points,the building average height from the ground and plane area of building,the algorithm could split each building.Experimental results show that the algorithm removed non-building points completely.