Diplomarbeit: Volumetric Base Mesh Construction

Thema & Aufgabe:

High-resolution, regularly sampled trivariate functions (for example, 3D medical data sets, 3D flow field data sets) typically do not consider the inherent complexity of the function. Therefore, the progressive approximation, and thus visualization, of these functions with volumetric subdivision schemes serves two main purposes: 1) a coarse approximation using just a small, "ideally" placed polyhedral cells (possibly quite general cells) may still capture many of the essential qualitative features of a function, and 2) a volumetric subdivision scheme will support progressive data transmission and rendering, possibly even view-dependent exploration techniques.

Therefore the regularly meshed, trivariate data shall be reduced to a coarse base mesh. This base mesh should be optimal in two ways: 1) certain features in the fine mesh are still recognizable in the base mesh and 2) the number of cells in the base mesh is a low as possible. Thus the decimation procedure is guided by an analysis of the data. To reconstruct the initial fine mesh a subdivison approach shall be used. That means that certain subdivision rules have to be constructed that use the information given in the coarse base mesh (and of the data analysis in the decimation procedure) such that subdividing the base mesh using these rules leads to a good approximation of the initial fine mesh. A genetic approach might be useful for the construction of near-optimal volumetric base mesh.

Fragen? Georg Umlauf (Raum 36-236)