Shape Recognition in 3D Point Clouds
Universität Bonn, Technical Report number CG-2007-1, May 2007
Abstract
Currently, tasks as basic as selecting all windows in a 3D laser scan of a house require a disproportional amount of user interaction. We present a flexible framework for the rapid detection of such semantically motivated shapes in large point clouds. Semantic entities are represented as constrained graphs that describe configurations of primitive shapes. Experimental results in various scenarios related to the architectural domain demonstrate the feasibility of the approach.
Images
![]() |
Download Paper
Bibtex
@TECHREPORT{cg-2007-1, author = {Schnabel, Ruwen and Wahl, Roland and Wessel, Raoul and Klein, Reinhard}, title = {Shape Recognition in 3D Point Clouds}, number = {CG-2007-1}, year = {2007}, month = may, institution = {Universit{\"a}t Bonn}, abstract = {Currently, tasks as basic as selecting all windows in a 3D laser scan of a house require a disproportional amount of user interaction. We present a flexible framework for the rapid detection of such semantically motivated shapes in large point clouds. Semantic entities are represented as constrained graphs that describe configurations of primitive shapes. Experimental results in various scenarios related to the architectural domain demonstrate the feasibility of the approach.}, issn = {1610-8892} }