Analysis of score functions for the automatic registration of terrestrial laser scans

authored by: Christoph Dold, Claus Brenner
Abstract: Terrestrial laser scanners are more and more popular, e.g. for surveying purposes, cultural heritage, city modelling or architectural applications. Usually different points of view are necessary to acquire an object completely. The task of finding the relative orientation between several scan positions is termed as registration. Research work in data driven methods for the registration of point clouds was frequently published the last decade. A distinction between coarse registration and fine registration has been drawn. We use a featurebased registration to search for a coarse orientation. The transformation parameters are calculated by establishing correspondences between extracted planes. An efficient search strategy delivers possible candidates. The focus of this paper lies in the verification phase that is responsible for picking the correct solution. A score function is used to evaluate the candidates of probabilistic solutions for the searched transformation. The goal of a score function is to get a reliable indication, which solution is best. Ideally, the score function attains its maximum for the correct solution and is significantly larger than the score of any wrong solution.
Organisation(s): Institute of Cartography and Geoinformatics
Type: Conference article
Journal: International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
Volume: 37
Pages: 417-422
No. of pages: 6
ISSN: 1682-1750
Publication date: 2008
Publication status: Published
Peer reviewed: Yes
ASJC Scopus subject areas: Information Systems, Geography, Planning and Development
Sustainable Development Goals: SDG 11 - Sustainable Cities and Communities
Electronic version(s): https://www.isprs.org/proceedings/XXXVII/congress/5_pdf/74.pdf (Access: Closed)