SignatureSpace: a multidimensional, exploratory approach for the analysis of volume data

Abstract

The analysis of volumetric data is a crucial part in the visualization pipeline, since it determines the features in a volume dataset and henceforth, also its rendering parameters. Unfortunately, volume analysis can also be a very tedious and difficult challenge.

To cope with this challenge, this paper describes a novel information visualization driven, explorative approach that allows users to perform an analysis in a comprehensive fashion. From the original data volume, a variety of auxiliary data volumes, the signature volumes, are computed, which are based on intensity, gradients, and various other statistical metrics. Each of these signatures (or signatures in short) is then unified into a multi-dimensional signature space to create a comprehensive scope for the analysis. A mosaic of visualization techniques ranging from parallel coordinates, to colormaps and opacity modulation, is available to provide insight into the structure and feature distribution of the volume dataset, and thus enables a specification of complex multi-dimensional transfer functions and segmentations.