Visibility Histograms in Direct Volume Rendering

Abstract

This thesis introduces visibility histograms as a method for analyzing volumetric datasets. These histograms show how much the data points within a 3D dataset that have the same scalar value influence the image which is created by rendering the dataset with a particular transfer function and from a particular viewing direction. These histograms can be used to gain insights into the internal structure of volumetric datasets, in particular information about occlusions. Furthermore, the possibility of automatically calculating transfer functions which generate a particular visibility histogram when applied to a dataset from a particular viewing direction is explored. Two methods which can be used to calculate a matching transfer function for a visibility histogram are explained, one of which is based on a genetic algorithm approach, while the other is an heuristic.