Laser-Based Automatic Cross-Sectional Change Detection for Steel Frames

Abstract

The aging infrastructure in the U.S. poses a major challenge for society since thousands of registered dams, bridges and other infrastructure components are now operating for more than five decades and have damage due to age or accumulated damage from hazards. Assessing present structural safety condition of these structures is vital for the process of evaluating their current status and deciding on future steps for rehabilitation. The in situ condition of structures can be captured by laser scanners that create dense 3D point clouds. However, deducting meaningful information from point clouds is generally difficult, and most of the time user-interaction is mandatory. Furthermore, it is even more challenging to automate this process and to detect deteriorated and damaged locations on the structure through point cloud processing. This requires dividing the point cloud into meaningful clusters obtained through segmentation and relationship modeling, followed by object recognition. It is then possible to locate defects on the structure through comparison once the objects from model library are correctly detected and fitted into the captured point cloud. This paper presents a new approach for detecting defects due to material loss through point cloud processing by using object detection and model comparison.