摘要
地面激光扫描技术是获取建筑物三维数据的重要手段之一,但其处理技术在自动化程度、密度适应性以及算法计算量等方面还存在较多问题。为此,本文提出了建筑区域点云的快速自动提取方法。首先,引入SLAM6D（simultaneous localization and mapping with 6 Dof）算法实现点云的自动配准;接着,使用体素重采样解决数据的远近密度差异与重叠区域冗余,并设计了空中管线滤除算子防止建筑分割中的粘连现象;然后,引入车载点云中的特征图法实现地面激光点云的快速分割;最后,使用先验知识从分割单元中识别建筑区域。实验证明,本方法可以从地面点云中提取建筑区域点云。
This paper proposes a fast workflow to extract the points in the building region.First,the SLAM6D algorithm is used to realize automatic registration.Then,we use voxelgrid resampling method to solve the density difference caused by distance and redundancy caused by overlap.In the meantime,we design an air pipeline removing operator to avoid conglutination of building segments.After that,the featureimage method of MLS data is improved to realize the fast segmentation of TLS data.Finally,prior knowledge is used to recognize the building region.Experimental result shows that method proposed in this paper is able to extract the building region accurately and efficiently.
作者
董安国
黄亮
DONG Anguo;HUANG Liang(Jiangsu Province Surveying ＆Mapping Engineering Institute, Nanjing 210013, China)
出处
《测绘地理信息》
2018年第3期112-114,共3页
Journal of Geomatics
基金
江苏省测绘地理信息科研资助项目(JSCHKY20172)
关键词
地面激光扫描
点云配准
点云重采样
特征图法
建筑区域提取
terrestrial laser scanning
registration
re-sampling
feature image
building region extraction
