基于移动激光扫描点云数据和遥感图像的建筑物三维模型快速建模方法

A Method for Fast Modeling of 3D Buildings from Mobile Laser Scanning Point Clouds and Remote Sensing Data

建筑物的三维建模是城市三维建模和可视化的重要组成部分。本文提出一种基于点云数据与遥感图像的建筑物三维模型快速建模方法。首先,运用改进的RANSAC法从点云数据中提取建筑立面,根据立面区分平顶建筑与人字形屋顶建筑;在此基础上,进一步对建筑物的高度进行提取;之后,利用区域增长法从遥感图像中提取建筑物屋顶轮廓,利用形态学方法对提取出的轮廓进行规则化处理,并基于Freeman链码提取轮廓角点,得到规整的轮廓;最后,根据提取出的建筑高度属性对屋顶轮廓拉伸并进行纹理映射,实现对建筑物的三维重建。通过实例证明,提出的方法能快速、高效地实现建筑物三维模型的重建。

A model is an abstraction of an entity object. Modeling three dimensional （3D） buildings is an important component of urban 3D modelling and visualization. This paper presents a new method for fast modeling of 3D buildings based on mobile laser scanning （MLS） point clouds and remote sensing images. First, buildings facades are extracted from original MLS point clouds data by using adapted RANSAC method, and are further classified into gable roof and plane roof. The heights of different types of buildings are determined. Next, the footprints and roofs of buildings are delineated from remote sensing image by using region growing method and morphological processes. The angular points, which can be used to represent the shape of the building roofs accurately, are identified from building roofs by using Freeman chaining coding algorithm. Finally, the 3 D building models are reconstructed after extruding the building footprint with building heights and mapping texture into corresponding facades. The method is proved to be effective and efficient through a case application.