Reconstruction dynamique de surface par convection géométrique

In this paper, we present a selective and evolutive method for reconstructing a surface from a point cloud that makes it possible to produce a simplified triangulated surface and to update it dynamically in an efficient manner. This method is based on a geometric convection algorithm and extends recent work on dynamic surface reconstruction introduced in (Allègre et al., 2005). We present a new data structure that significantly speeds up the original selective reconstruction algorithm and makes it possible to handle point set models with millions of sample points efficiently. We also show how to control the quality of the resulting meshes. The proposed method can also take into account some attributes like color in the selective reconstruction process. MOTS-CLÉS : reconstruction de surface, simplification de nuages de points, convection géométrique, correction dynamique.