Une approche floue du recalage 3D genericite et robustesse 1995

Abstract:  An important problem in computer vision is to recover how features extracted from images are connected to existing models. In this paper, we focus on solving the registration problem, i.e obtaining rigid displacement parameters between several 3D data sets, whether partial or exhaustive. The difficulty of this problem is to obtain a method which is robust to outliers and at the same time accurate. We present a general method performing robust 3D location and fitting based on fuzzy clustering method. The fuzzy set approach is known to its practical efficiency in uncertain environment. To illustrate the advantages of this approach on the registration problem, we show results on synthetic and real 3D data. Moreover, we propose a tool to compare two 3D data sets in terms of similarity and dissemblance.