Feature preserving smoothing of 3D surface scans

Abstract

With the increasing use of geometry scanners to create 3D models, there is a rising need for effective denoising of data captured with these devices. This thesis presents new methods for smoothing scanned data, based on extensions of the bilateral filter to 3D. The bilateral filter is a non-linear, edge-preserving image filter; its extension to 3D leads to an efficient, feature preserving filter for a wide class of surface representations, including points and "polygon soups."