Automatic extraction of bridge dimensional information using 3D point cloud data

Abstract

The advent of advanced 3D LiDAR technology and progress in computer vision have led to the extensive utilization of three-Dimensional Point Cloud Data (3D PCD) across various fields such as robotics and autonomous driving, as well as in the construction industry. Particularly, for infrastructure facilities lacking design drawings or with missing records, automated Scan-to-BIM technology plays a pivotal role in efficiently generating three-dimensional models. However, while recent studies have actively explored the segmentation of 3D PCD essential for automated Scan-to-BIM technologies in infrastructure, subsequent research, such as deriving dimensions of segmented objects, remains relatively underdeveloped. This study proposes a novel method for automatically extracting dimensional information for each component of a bridge using 3D PCD. Unlike existing methods that rely solely on the intrinsic characteristics of 3D LiDAR to derive lengths, the proposed approach is capable of handling challenges such as sloped superstructures and multiple piers with varying lengths. The proposed method combines RANSAC, geometric features, edge detection, and orthographic projection techniques to accurately and automatically extract dimensional information from 3D PCD. This approach not only provides critical data for generating bridge BIM models but also holds potential for quickly predicting damages such as deflections and deformations in bridges.