中文摘要：

提出了一种基于原始离散LiDAR点云的多层结构建筑物激光点自动提取方法。首先对原始离散Li-DAR点云进行不规则三角网构建,然后利用分阶段区域生长的原理逐阶段遍历TIN中的各三角面。依据三角面的法向量方位角判定建筑物的各层结构,依次提取出多层建筑物的墙面及各级屋顶面,最后将提取出的各层屋顶面数据进行分类存储,确保提取出的各栋建筑物激光点之间相互独立且包含各层屋顶结构等细部特征信息,便于后续各建筑物的三维重建。

英文摘要：

A novel method of multistorey building automatic extraction from raw and discrete Light Detection and Ranging（LiDAR） data was presented.In the proposed method,the TIN model of the raw and discrete LiDAR data is firstly built,then each triangle of the TIN model is traversed according to the staged region growing algorithm.In the algorithm,each triangulation can be identified as a Wall Triangle（WT） or a None-Wall Triangle（NWT） by the orientation of its normal vector,and the walls or multistorey building roofs are identified in a sequential order until all of the triangles are traversed.After that,the multistorey structure of each building can be finally extracted,and each one of the buildings is recognized as an unique entity which is differentiated form the others.