Multiple Datasets Visualization with Isosurface Extraction

G. Khanduja and B.B. Karki (USA)


Multiple datasets visualization, Isosurface extraction, Scalar volume visualization, Visualization application


Multiple datasets visualization (MDV) processes concurrently more than one datasets of a given type in the same visualization. In this paper, we propose two isosurface-based MDV techniques, namely, all-in memory (AIM) and only-polygons-in-memory (OPIM) methods, which exploit the well-known Marching Cubes algorithm. These techniques differ in the way they use the memory space and provide interaction option. Performance analysis shows that the calculated times for the generation and rendering of polygons representing isosurface(s) increase non-linearly with increasing the number (N) of individual scalar volume datasets and the frame-rates drop below one even for a relatively few datasets. To improve the performance, we adopt a dynamic-resolution concept at the volume data level. Using different threshold criteria related to the frame-rate and time-N variation, the volume data are sub-sampled and reloaded when N crosses the threshold limit. This switching to a low-resolution mode frees up the memory space for additional data to be processed. In this way, we are able to maintain a linear time-N relation, enhance the frame-rate and increase the maximum N that can be handled at a time. The proposed MDV techniques are expected to be useful for an increasing number of applications where multiple datasets representing multiple samples or multiple conditions or multiple times are involved.

Important Links:

Go Back