Visual inspection of multivariate volume data based on multi-class noise sampling

Zhiyu Ding,Ziang Ding,Weifeng Chen,Haidong Chen,Yubo Tao,Xin Li,Wei Chen

DOI: https://doi.org/10.1007/s00371-015-1070-6

2015-01-01

Abstract:Visualizing multivariate volume data is useful when the user wants to inspect the correlational distributions of multiple variables in a spatial field. Existing solutions commonly rely on color blending or weaving techniques to show multiple variables on a sampling point, probably causing heavy visual confusion. This paper presents an alternative solution that employs a multi-class sampling technique to generate spatially separated sampling points for multiple variables and illustrates the sampling points of each variable individually. We combine this new sampling scheme with the conventional direct volume rendering mode, iso-surface mode, and the cutting plane mode to support interactive inspection of volumetric distributions of multiple variables. The effectiveness of our approach is demonstrated with the IEEE VIS Contest 2004 Hurricane dataset and a 3D nuclear fusion simulation dataset.