Toggle Main Menu Toggle Search

Open Access padlockePrints

A Correspondence-Based Strategy for Automatic Registration of Terrestrial Laser Scanning Data

Lookup NU author(s): Dr Darion Grant

Downloads

Full text for this publication is not currently held within this repository. Alternative links are provided below where available.


Abstract

The absence of explicit point correspondences among overlapping terrestrial laser scanning data limits the performance of automatic registration schemes. The popular Iterative Closest Point (ICP) and its variants, solve the correspondence problem implicitly while minimizing some distance metric. Other approaches perform low-level processing to obtain surface properties from which correspondences are established, and then they conduct the registration, which, in general is not as accurate as the ICP methods. This paper presents an approach that addresses the registration issue by dealing directly with the correspondence problem, without the use of derivative surface properties, except for local surface tangents. An Iterative Network Matching (INM) scheme is developed, in which a subset of points from one dataset comprise a 3D network. The corresponding network on the other dataset is obtained through a modified trilateration network adjustment, but without control, and a novel strategy is employed in which the local surface tangents are utilized as constraints. The main purpose of this paper is to extend the recent introduction of the INM methodology, by proving the validity of the proposed use of surface constraints for free network adjustment. From preliminary experiments with synthetic and real data, INM yielded global correspondence and registration RMSE that were as accurate as the ICP method of Chen and Medioni, 1991, and in some cases, improvements up to an order of magnitude in correspondence RMSE, were realized by INM.


Publication metadata

Author(s): Grant D, Bethel J, Crawford M

Publication type: Conference Proceedings (inc. Abstract)

Publication status: Published

Conference Name: Opportunities for emerging geospatial technologies : American Society for Photogrammetry and Remote Sensing annual conference

Year of Conference: 2010

Pages: 1-11

Publisher: American Society for Photogrammetry and Remote Sensing

URL: http://www.asprs.org/a/publications/proceedings/sandiego2010/sandiego10/Grant.pdf


Share