Efficient surface mesh reconstruction from unorganized points using neural network

In this paper, a new approach for the automatic reconstruction from unorganized points is presented, where first an artificial neural network is used to order the data and form a grid of control vertices with triangle topology. Then, we present a general scheme for mesh simplification and optimization that allows to control the geometric approximation as well as the element shape and size quality (required for numerical simulations). The new approach makes possible the construction of adapted geometric meshes for surfaces by specifying the element sizes (and directions) so as to bound the error below a user-given threshold value. The experimental results show that our methods are accurate and simple to implement.