Fitting of Constrained Models to Poor 3D Data

Abstract

In this work we have addressed the question of whether it is possible to extract parametric models of features from poor quality 3D data. In doing this we have examined the applicability of an evolutionary strategy to the problem of fitting constrained parametric models. In the first phase, a background surface is fitted and removed leaving points of discontinuity associated with the artifacts. Then the discontinuities are classified, using the RANSAC algorithm, into drilled hole artifacts or blade edges suggesting drilled slots. This information, as well as the set of discontinuity points is passed to the Genocop III algorithm proposed by Michalewicz for optimization using a priori geometric constraints. Results, example times for convergence and comparisons with known ground truths are given.