Linking Multidimensional Feature Space Cluster Visualization to Multifield Surface Extraction

Abstract

Data sets resulting from physical simulations typically contain a multitude of physical variables. So, visualization methods should take into account the entire multifield volume data rather than concentrate on one variable. We have developed a visualization approach based on surface extraction from multifield volume data. The extracted surfaces segment the data with respect to an underlying multivariate function. Decisions on segmentation properties are based on the analysis of a multidimensional feature space. We perform feature space exploration using automated multidimensional hierarchical clustering. The hierarchical clusters appear as a cluster tree in a 2D radial layout. In this layout, the user can select clusters of interest. A selected cluster in feature space corresponds to a segmenting surface in object space. On the basis of the segmentation property induced by the cluster membership, we extract surfaces from the volume data.