Publications / 2013 Proceedings of the 30th ISARC, Montréal, Canada

Autonomous Modeling of Pipes within Point Clouds

Mahmoud Ahmed, Carl T. Haas, Ralph Haas
Pages 1093-1100 (2013 Proceedings of the 30th ISARC, Montréal, Canada, ISBN 978-1-62993-294-1, ISSN 2413-5844)

Three-dimensional models are in increasing demand for design, maintenance, operations and construction project management. Point-clouds are the main output of automatic data collection using laser-scanners and photogrammetric technologies. Pipe-works may comprise 50% of the value of important construction projects such as industrial and research facilities. We have developed a practical and cost-effective approach, based on the Hough-transform and judicious use of domain constraints, to find, recognize and reconstruct/model 3D pipes within point-clouds fully automatically. The developed approach allows pipe growing within a 3D point-cloud. It samples the point-cloud to sequential orthogonal slices, detects pipe’s cross-sections within sampled slices and grows the pipe along the cross sections centerlines. It is validated using laser-scanner data from construction of the Engineering-VI building on the University of Waterloo (UW) campus. The system works on a typical laptop. Recognition and localization results are within a few millimeters. The presented approach opens the gate to broad applications of pipe-network modeling.

Keywords: No keywords