Abstract
Abstract. The paper deals with the 3D reconstruction of bridges from Airborne Laser Scanning point clouds and cadastral footprints. The generated realistic 3D objects can be used to enhance city models. While other studies have focused on bridge decks to fill gaps in digital elevation models, this paper focuses on the decomposition of superstructures into construction elements such as pylons, cables and arches. For this purpose, the bridge type is classified, and a combination of model-based and data-based methods is used that are built on the detection of arcs, catenaries, and line segments in the point clouds. The described techniques were successfully applied to create 3D models of the Rhine bridges in the German state of North Rhine-Westphalia.
Highlights
In the last decade, work on automatic computation of 3D city models from Airborne Laser Scanning (ALS) point clouds and cadastral footprints has focused on building roofs
This was achieved with a general data-based approach not restricted to suspension bridges: first, planes were matched to laser scanning points, and points were projected onto these planes, thereby generating 2D images with sparsely set pixels
Even if manual corrections are necessary in some cases, the superstructures generated by the presented methods are much more realistic than the fully automatically computed structures in (Goebbels, 2021) on which we have built
Summary
Work on automatic computation of 3D city models from Airborne Laser Scanning (ALS) point clouds and cadastral footprints has focused on building roofs. Whilst the previous study (Goebbels, 2021) of our research group focused on assembling smooth surfaces of bridge decks from planar polygons, it dealt with simple but fast reconstruction of building elements that are above the deck level of the bridge This was achieved with a general data-based approach not restricted to suspension bridges: first, planes were matched to laser scanning points, and points were projected onto these planes, thereby generating 2D images with sparsely set pixels. These pixels were connected to nearest neighbors by line segments, contours were detected, and extruded in 3D. This algorithm makes bridges recognizable, it does not allow to distinguish between different construc-
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
