Automatic feature modeling techniques for volume segmentation applications

Abstract

In many volume segmentation and visualization tasks, the ability to correctly identify the boundary surface of each volumetric feature of interest in the data is desirable. This surface can be used in subsequent quantitative studies of the segmented features. In this paper, we present an automatic approach to generate accurate representations of a feature of interest from volume segmentation. Our method first locates a set of points, which tightly define the boundary of the volumetric feature. This set of points can then be used to construct a boundary surface mesh. We also describe how to construct an anti-aliased volume representation of the segmented feature from this point set to enable high-quality volume rendering of the feature. These three representations - point set, boundary surface mesh, and anti-aliased volume segment - have a wide variety of applications.