利用基元分解的机载点云复杂建筑物自动重建

Building Reconstruction from Airborne LiDAR Point Clouds Using Primitive-based Method

针对复杂建筑物自动重建中重建类型有限和边界线确定困难的问题,提出了一种基于基元分解的机载点云建筑物重建方法。该方法通过对屋顶面构建拓扑图并依据最小环原则进行拓扑分解,将复杂建筑物分解成多个简单基元;根据拓扑图的基本要素建立一个涵盖常见基元的建筑基元库,通过将复杂建筑物分解得到的基元与建筑基元库中的基元进行匹配,将复杂建筑物的重建转化为简单基元的重建,解决复杂建筑物重建类型有限的问题;综合利用建筑物整体外轮廓和屋脊线来确定基元边界线,克服数据缺失等因素对边界精度的影响。实验结果表明,该方法可有效地对多种复杂建筑物进行自动重建,具有可靠的精度。

Automatic reconstruction of 3 D building models is still difficult due to the building boundary error and the limitation of building types.In this paper,we propose a primitive-based method for the automatic reconstruction of complex buildings from airborne LiDAR point clouds.We construct topology graphs of object surfaces,and divide complex buildings into primitives by topological decomposition;we create a building primitive library according to the elements of topology graphs,and the problem is converted into the reconstruction of building primitives by matching topology graphs,so that we can overcome the limitation of building types;we determine building boundaries by combining contours and ridges of the complex building to avoid the negative influence of the missing data.Experiment demonstrates the feasibility of this method,which automatically reconstructs complex buildings with reliable geometric accuracy.