机载LiDAR点云的多层建筑物三维轮廓模型自动重建

发布时间：2018-06-23 21:34

本文选题：激光雷达 + 多层建筑物　；参考：《测绘科学技术学报》2016年05期

【摘要】：以机载LiDAR点云数据为研究对象,提出一种新的基于点云数据的多层建筑物三维轮廓模型高精度自动重建方法。在已完成建筑物结构提取及轮廓规则化处理的基础上,利用多层屋顶轮廓在水平投影面内的相邻关系,将各层屋顶中同等级屋顶的相邻关系概括为平行边、不平行且不相交、相交3种相邻形式,结合多层屋顶的层级结构信息对相邻轮廓边界进行一致性处理。实验证明本文方法可以进一步消除多层建筑物各屋顶轮廓的规则化处理误差,使相邻轮廓边界在水平投影面内严格重合,同时重建后建筑物三维轮廓模型的正确性与完整性较高,拐点的定位精度优于激光点平均间距。
[Abstract]:Taking airborne LiDAR point cloud data as the research object, a new high precision automatic reconstruction method for 3D contour model of multi-story buildings based on point cloud data is proposed. On the basis of the extraction of building structure and the regularization of the outline of the building, the contiguous relation of the multi-story roof contour in the horizontal projection plane is used to generalize the contiguous relation of the roof of the same grade in each floor as the parallel edge, which is not parallel and does not intersect. Three adjacent forms are intersected, and the contiguous contour boundary is processed in accordance with the hierarchical structure information of the multi-story roof. Experimental results show that the proposed method can further eliminate the regularization error of each roof profile of multi-storey buildings, and make the adjacent contour boundaries coincide strictly in the horizontal projection plane. At the same time, the 3D contour model of the reconstructed building has higher accuracy and integrity, and the accuracy of the inflexion is better than the average distance between laser points.
【作者单位】：信息工程大学;
【基金】：国家自然科学基金项目(41501506) 河南省基础与前沿技术研究计划项目(152300410005)
【分类号】：P237

【相似文献】