A 3D Automatic Optical Inspection (3D-AOI) System for Architecture Inspection

Abstract

Most architecture inspections, that were traditionally relied on human operations by professionals or experts, could be dangerous and time-costly. With the recent developments of unmanned aerial and ground vehicles, some visual inspections were done based on image acquisitions from the camera on an unmanned vehicle. This paper presented a 3D Automatic Optical Inspection (3D-AOI) system for bridge inspection. A stereovision camera that was installed on top of an unmanned aerial vehicle (UAV) acquired a 3D point cloud of the inspected area of the bridge’s cap beam. Image processing steps were used to recognize the bridge crack pattern. Finally, the voxels of the crack pattern were transformed to the plane that was perpendicular to the camera direction for the measurements of the geometrical features of the crack. In our implementation, the measurement error was less than 5% as the normal of the inspected plane had a 40°-angular difference between the camera direction.