A New Technique

for

Resampling Curvilinear Volumes

David D. Helfrick

Computer Science Department

University of Maryland Baltimore County

Baltimore, Maryland 21228

helfrick@cs.umbc.edu

December 11, 1995

ABSTRACT

Volume rendering large three-dimensional (3D) data sets is an active research area. The rendering of irregular grid structures is particularly difficult. To render curvilinear data, researchers have developed new techniques and extended generic algorithms.

The problem investigated in this research is rendering curvilinear data, and the particular application is rendering computational fluid dynamics (CFD) data. Current rendering systems handling curvilinear data are inadequate. These systems either do not render the 3D data in its entirety, or they are inefficient. Both components, completeness and efficiency, are needed for the visualization system to be of great use to researchers.

A new technique (as well as its predecessor) is presented to resample irregular grids to a regular voxel structure. The regular structure allows the use of traditional volume rendering techniques for image rendering. This technique is applicable to any curvilinear data, and traverses the smaller irregular grid to determine which voxels each cell point influences, rather than traversing the larger resampled rectilinear grid to find the influencing cell points. Compared to the original system, this approach improves the execution time by a factor of thirty. Additionally, the new technique produces smoother 3D images. Possible extensions to the technique are also discussed.

KEYWORDS