Section: New Results

Extraction of cylinders and cones from minimal point sets

Participants : Laurent Busé, André Galligo, Jiajun Zhang.

The extraction of geometric primitives from 3D point clouds is an important problem in reverse engineering. These 3D point clouds are typically obtained by means of accurate 3D scanners and there exists several methods for performing the 3D geometric primitives extraction. An important category among these method are based on a RANSAC method. For such methods, the primitives are directly extracted for the input point cloud. The basic idea is to extract a particular elementary type of shape, such as planes, spheres, cylinders, cones or tori, from the smallest possible set of points and then to judge if this extracted primitive is relevant to the full point cloud. Therefore, for this category of methods it is very important to compute a particular type of shape through the smallest possible number of points, including normals or not. If the extraction of planes and spheres is easy to treat, the cases of cylinders, cones and tori are more involved. In this work, we aim at developing methods for extracting these geometric primitives from the smaller possible number of points (counting multiplicities if normals are taken into account). Another objective is also to provide methods for extraction without using estimated normals in order to improve the accuracy of the extracted geometric primitive, or to use mixed data depending of the applied context (some points with normals and some other points without normals). A paper is in preparation.