Shape classification and normal estimation for non-uniformly sampled, noisy point data

► Robust normals even in the presence of noise in the data and non-uniform sampling. ► Construction of a well-behaved, minimal one-ring neighborhood from the k-nearest neighbors. ► Local shape estimation (flat, bowl, saddle, ridge, corner, sharp edge, or boundary). ► An optimization algorithm to find the above that cross-validates the results. ► Identification of outliers and poorly reconstructed areas. ► Two novel surface reconstruction algorithms based on the constructed one-rings.