机载激光扫描中复杂建筑物轮廓线平差提取模型

Adjustment Model of Boundary Extraction for Urban Complicate Building Based on LiDAR Data

摘要:

建筑物轮廓线提取与规则化是房屋3维重建等处理中的重要步骤，目前大多方法面向多边形规则建筑物轮廓线的提取，而无法适用于包含圆弧轮廓线的不规则建筑物轮廓线提取.针对城市中这一类复杂不规则建筑物，提出一种结构化提取特征点的方法，并判断特征点的属性，对直角处的特征点进行条件平差，优化特征点的位置，而对圆弧处的特征点之间的边界点分段拟合圆弧，以得到平滑的符合实际情况的轮廓线.该方法可有效提取包含圆弧的建筑物轮廓线，最后通过上海陆家嘴地区的建筑物LiDAR（机械激光扫描）数据验证该方法的可行性和提取精度.

Abstract:

The extraction and regulation of building boundary is a crucial step towards 3D building reconstruction, most methods that are appropriate for buildings of regular polygon cannot extract irregular building boundary which contains arc lines. A structure method for extracting building feature points is put forward by considering the irregular building boundary. By judging the property of feature points, feature points located at rectangular corner are adjusted with the parallel requirements to obtain accurate feature points, boundary points between adjacent arc feature points are fitted to an arc so as to get the real smooth boundary. Building boundary that contains arc lines can be efficiently extracted by proposed method. As a case study, LiDAR point clouds of Lujiazui in Shanghai are used to prove the feasibility and extraction accuracy of the method.