Differential Visualization of Large-Scale Time-Varying 3D Volumetric Data
The goal of this project is to facilitate the real-time exploration of large (several 100 GB) time-varying volumetric data produced by computational modeling or medical applications. We take advantage of temporal coherence (coherence between consequent time frames) and spatial coherence (coherence between neighboring voxels) of three-dimensional time-varying data to facilitate an efficient visualization process.
As computing power and scanning precision rapidly increases, scientific simulation or measurements generate more and more densely sampled time-varying 3D volume data which are large in space (up to a few billions of voxels per time step) and time (hundreds to thousands of time steps). While the current state-of-the-art graphics hardware allows very fast volume rendering, interactive exploration of large-scale time-varying data is still a challenging research problem. The difficulties mainly come from the fact that only a small proportion of data in the entire time series can fit into main memory and graphics hardware memory at a time and transferring the data between the disk and main memory or between main memory and graphics hardware can become a major bottleneck.
In this project we take advantage of the temporal coherence (coherence between subsequent time frames) and spatial coherence (coherence between neighboring voxels) of three-dimensional time-varying data to facilitate an efficient visualization process. First, we calculate the differential information among a sequence of 3D volumetric data in a preprocessing step. Then, during the visualization, using the differential information, the parts of the output that needs to be updated are identified and only the necessary data are loaded to main memory and processed.