<title>Surface reconstruction from scattered measurements</title>

Abstract

Given scattered measurements from a surface, a method for reconstruction of the surface using rational Gaussian surfaces is described. Rational Gaussian surfaces are globally defined parametric surfaces that adapt well to local data. Examples demonstrating reconstruction of 3- D scenes from scattered depth measurements obtained from stereo disparity, reconstruction of 3-D shapes from noisy range data, and segmentation of volumetric images for the extraction of generalized cylinders are described.