Improved Density Clustering for Spacing Measurement of Irregular Rebar Mesh from 3D Point Clouds

Proper spacing of rebar mesh is essential for structural integrity and safety, but measuring the spacing performance in three-dimensional (3D) space can be complicated. Existing spacing verification either relies on well-annotated data or encounters challenges when dealing with obscured point clouds. To address these challenges, we propose a novel approach using a projection clustering algorithm. Initially, we project the 3D point cloud onto a frontal view and perform density peak clustering in two directions, enabling accurate recognition of vertical and horizontal boundaries. Then, we iteratively select the centerline of each rebar based on clustering center points, maximizing internal points. Evaluation using the Dimension Quality Assessment System showed that our method achieves 94.5% accuracy in estimating rebar spacing. Compared with classical methods, our approach significantly improves spacing measurement.