Surface modeling by using self organizing maps of Kohonen

The general aim in surface reconstruction from unorganized sampling points is to convert the cloud of points into a standard geometrical and freely formed surfaces obeying to design conditions. In this paper we propose to build a connectivity between the sample points by applying a learning algorithm for specific architectures as toric and spherical maps derived from Kohonen's self organizing maps. We start from a pre-built neural network architecture, and by the use of a learning process, we achieve a topological model of the original surface.