Hardware-Assisted Feature Analysis and Visualization of Procedurally Encoded Multifield Volumetric Data

Procedural encoding of scattered and unstructured scalar datasets using Radial Basis Functions (RBF) is an active area of research with great potential for compactly representing large datasets. This reduced storage requirement allows the compressed datasets to completely reside in the local memory of the graphics card, thus, enabling accurate and efficient processing and visualization without data transfer problems. We have developed new hierarchical techniques that effectively encode data on arbitrary grids including volumetric scalar, vector, and multifield data. Once the RBF representation is transferred to texture memory, GPU-based visualization using particle advection, cutting planes, isosurfaces, and volume rendering can be performed by functional reconstruction of the encoded data in the fragment pipeline. For the special requirements of flow visualization, we derive the definitions of well known features in RBF space allowing us to integrate pixel accurate hardware-accelerat...

• Type: Journal
• Year: 2005
• Where: CGA
• Authors: Manfred Weiler, Ralf P. Botchen, Simon Stegmaier, Thomas Ertl, Jingshu Huang, Yun Jang, David S. Ebert, Kelly P. Gaither
• Keywords: CGA 2005, Computational Geometry, Feature Detection, Radial Basis Functions, Scalar Datasets