A Clustering-based Visualization Technique to Emphasize Meaningful Regions of Vector Fields

Abstract

This paper proposes a vector field visualization approach that extracts and visualizes grouped regions of static 3D vector fields of similar curvature behavior. These regions are argued to ease the recognition of regions of potential interest and accelerate the general exploration process of vector fields. Our approach detects regions of similar geometric stream properties such as integral curvature and visualizes them by means of compact cluster boundaries. To supplement existing approaches our method combines information on relevant scales to extract meaningful semantical aspects of the overall field structure. For proof of concept we illustrate our results based on real and synthetic data sets.