Automatic Edge Detection From Point Clouds Collected by Terrestrial Laser Scanners

Abstract

The continuous advancement of precise and high‐speed laser scanning technology has enabled its widespread utilization in construction surveying. Consequently, the surge in demand for reconstructing three‐dimensional (3D) models from point cloud data has promoted the development of automatic or semiautomatic data processing processes. In this study, an automated method for extracting edge points from 3D point clouds collected by terrestrial laser scanners is introduced. The automated method analyzes the geometric attributes of relevant objects. Nine data samples with progressively escalating complexity levels were used to evaluate the method. The results revealed that the method can easily extract straight edges or intersections at 90° angles and the point density does not significantly influence the extraction of edge data, particularly for elements of considerable size. Moreover, structures with features larger than four times the neighborhood searching distance can be effectively extracted with minimal noise points by the automated method. This study underscores the substantial impact of the geometrical features of the target object on setting the appropriate parameters for extracting edge points. The automated method developed in this study is cost‐effective, accurate, and can be applied with terrestrial laser scanners for creating information‐rich 3D models of infrastructure.