Reducing Aliasing Artifacts in Iso-Surfaces of Binary Volumes

Abstract

This paper presents a method for filtering binary volumes in order to reduce the aliasing artifacts that typically result when trying to visualize the surfaces described by such a binary partitioning. The proposed method is a reformulation of the technique of constrained elastic surface nets from the literature. The strategy is to estimate a 3D surface using an iterative relaxation process, which finds the surface of minimal area while treating the binary input data as a set of constraints. Unlike surface nets, the proposed method does not require the generation of any explicit surface representation; it acts directly on the volume. This is done by treating the level sets of the volume as deformable surfaces moving under a mean curvature flow constrained by a discretely sampled inside/outside function. Because the proposed method does not require generating a surface mesh, it avoids the various singularities and topological problems associated with iso-surface generation. It also avoids the artifacts that an irregular mesh topology can produce in deformation processes. If the goal is volume rendering, the proposed method avoids the time consuming process of converting surface meshes back into volumes.