Maalek Reza, Lichti Derek D, Ruwanpura Janaka Y
Department of Civil Engineering, University of Calgary, Calgary, AB T2N 1N4, Canada.
Department of Geomatics Engineering, University of Calgary, Calgary, AB T2N 1N4, Canada.
Sensors (Basel). 2018 Mar 8;18(3):819. doi: 10.3390/s18030819.
Automated segmentation of planar and linear features of point clouds acquired from construction sites is essential for the automatic extraction of building construction elements such as columns, beams and slabs. However, many planar and linear segmentation methods use scene-dependent similarity thresholds that may not provide generalizable solutions for all environments. In addition, outliers exist in construction site point clouds due to data artefacts caused by moving objects, occlusions and dust. To address these concerns, a novel method for robust classification and segmentation of planar and linear features is proposed. First, coplanar and collinear points are classified through a robust principal components analysis procedure. The classified points are then grouped using a new robust clustering method, the robust complete linkage method. A robust method is also proposed to extract the points of flat-slab floors and/or ceilings independent of the aforementioned stages to improve computational efficiency. The applicability of the proposed method is evaluated in eight datasets acquired from a complex laboratory environment and two construction sites at the University of Calgary. The precision, recall, and accuracy of the segmentation at both construction sites were 96.8%, 97.7% and 95%, respectively. These results demonstrate the suitability of the proposed method for robust segmentation of planar and linear features of contaminated datasets, such as those collected from construction sites.
从建筑工地采集的点云的平面和线性特征的自动分割对于自动提取建筑施工元素（如柱、梁和平板）至关重要。然而，许多平面和线性分割方法使用依赖场景的相似性阈值，这些阈值可能无法为所有环境提供通用的解决方案。此外，由于移动物体、遮挡和灰尘导致的数据伪像，建筑工地的点云中存在异常值。为了解决这些问题，提出了一种用于平面和线性特征的鲁棒分类和分割的新方法。首先，通过鲁棒主成分分析程序对共面和共线点进行分类。然后使用一种新的鲁棒聚类方法——鲁棒完全连锁法对分类后的点进行分组。还提出了一种鲁棒方法，独立于上述阶段提取平板地板和/或天花板的点，以提高计算效率。在从卡尔加里大学的复杂实验室环境和两个建筑工地采集的八个数据集中评估了所提出方法的适用性。两个建筑工地分割的精度、召回率和准确率分别为96.8%、97.7%和95%。这些结果证明了所提出的方法适用于对受污染数据集（如从建筑工地收集的数据集）的平面和线性特征进行鲁棒分割。
