Share Email Print

Proceedings Paper

Estimation of 3D reconstruction errors in a stereo-vision system
Author(s): A. Belhaoua; S. Kohler; E. Hirsch
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

The paper presents an approach for error estimation for the various steps of an automated 3D vision-based reconstruction procedure of manufactured workpieces. The process is based on a priori planning of the task and built around a cognitive intelligent sensory system using so-called Situation Graph Trees (SGT) as a planning tool. Such an automated quality control system requires the coordination of a set of complex processes performing sequentially data acquisition, its quantitative evaluation and the comparison with a reference model (e.g., CAD object model) in order to evaluate quantitatively the object. To ensure efficient quality control, the aim is to be able to state if reconstruction results fulfill tolerance rules or not. Thus, the goal is to evaluate independently the error for each step of the stereo-vision based 3D reconstruction (e.g., for calibration, contour segmentation, matching and reconstruction) and then to estimate the error for the whole system. In this contribution, we analyze particularly the segmentation error due to localization errors for extracted edge points supposed to belong to lines and curves composing the outline of the workpiece under evaluation. The fitting parameters describing these geometric features are used as quality measure to determine confidence intervals and finally to estimate the segmentation errors. These errors are then propagated through the whole reconstruction procedure, enabling to evaluate their effect on the final 3D reconstruction result, specifically on position uncertainties. Lastly, analysis of these error estimates enables to evaluate the quality of the 3D reconstruction, as illustrated by the shown experimental results.

Paper Details

Date Published: 17 June 2009
PDF: 10 pages
Proc. SPIE 7390, Modeling Aspects in Optical Metrology II, 73900X (17 June 2009); doi: 10.1117/12.827346
Show Author Affiliations
A. Belhaoua, Lab. des Sciences de l'Image, de l'Informatique et de la Télédétection (France)
S. Kohler, Lab. Modélisation Intelligence Processus Systèmes (France)
E. Hirsch, Lab. des Sciences de l'Image, de l'Informatique et de la Télédétection (France)

Published in SPIE Proceedings Vol. 7390:
Modeling Aspects in Optical Metrology II
Harald Bosse; Bernd Bodermann; Richard M. Silver, Editor(s)

© SPIE. Terms of Use
Back to Top