3D-FEATURE RECOGNITION FROM MEASURED DATA

Authors

  • M. Janssens Materialise N.V., Leuven, Belgium
  • W. Van Wijck Department of Industrial Engineering, Stellenbosch University
  • N.D. Du Preez Department of Industrial Engineering, Stellenbosch University

DOI:

https://doi.org/10.7166/10-1-369

Abstract

ENGLISH ABSTRACT: This paper presents a method to automatically extract analytical entities like planes, spheres and cylinders from a file containing a cloud of points. The method facilitates the manipulation and reduction of large data sets and the evaluation of it. It can be used as a design tool, a quality control tool, data-processing tool or a data reduction tool. From a database of points, the user can automatically extract a subset of points belonging to an analytical entity of interest, within a predefined but adjustable level of confidence. If necessary, the dimensional parameters of the entity can also be calculated.
The method is based on the subtle statistical properties of the least-squares technique that makes it compliant with the strict regulations in the co-ordinate measuring arena. Its robustness guarantees . the applicability to less accurate environments than precision engineering.

AFRIKAANSE OPSOMMING: Hierdie artikel bespreek 'n metode met behulp waarvan analitiese voorwerpe soos vlakke, sfere en silinders outomaties vanuit 'n wolk van datapunte, onttrek kan word. Die metode is geskik vir die hantering, manipulasie, reduksie en evaluasie van groot data-stelle. Dit kan gebruik word as 'n gereedskapstuk vir ontwerp, gehaltebeheer, dataverwerking en data-reduksie. Gegewe 'n databasis van punte, kan die gebruiker die subset van punte wat tot enige analitiese voorwerp van belang behoort, outomaties binne 'n voorafgespesifiseerde, maar verstelbare, vlak van vertroue onttrek. Hierbenewens en indien nodig, kan die dimensionele parameters en afmetings van die betrokke entiteit ook bereken word.
Die algoritme maak van die kleinste-kwadrate metode gebruik, sodat elke passings-parameter statisties kwantifiseerbaar en verantwoordbaar is. In hierdie opsig voldoen dit aan die streng regulasies wat die koordinaat-meet arena kenmerk. Die robuustheid van die metode, maak dit ook geskik vir toepassing in minder akkurate omgewings as presisie-ingenieurswese,

Downloads

Published

2011-11-05

How to Cite

Janssens, M., Van Wijck, W., & Du Preez, N. (2011). 3D-FEATURE RECOGNITION FROM MEASURED DATA. The South African Journal of Industrial Engineering, 10(1). https://doi.org/10.7166/10-1-369

Issue

Section

General Articles