Three-Dimensional Tunnel Reconstruction Using Photogrammetry and Laser Scanning

Automatic dense high-resolution tunnel surface reconstruction using photogrammetric methods has been realized by the Dibit tunnel measurement system for almost a decade. We report on the current state of the system from the computer vision point of view. The tunnel surface is measured by a laser scanner or stereo photogrammetry. Sensor orientation is accomplished by geodetic means. Laser scan points and stereoscopic reconstructions are projected to a regular grid on the design surface, which contains both radial deviations from the design and image texture represented by digital images. Methods for refinement and increase of accuracy include the usage of an additional distance sensor for each stereo image set, texture based fusion between laser scan and camera image, 3d textured point cloud alignment, and automatic image stitching. The data flow of the main system components from data acquisition to the derivation of quality measures and digital maps of the tunnel surface is shown. With the example of crack detection new exploitation potential for tunnel documentation & inspection is demonstrated.