SC|05
Workshop on State-of-the-Art Visualization Techniques for
Gleaning Insights in Large Time-Varying Volume Data



Temporal Features Encoding and Extraction
Han-Wei Shen
Ohio State University
11:00am-11:30pm, November 18, 2005
Room 302, Washington State Convention and Trade Center, Seattle, WA

In this talk, we will present novel space-time data encoding and feature tracking schemes for large scale time-varying data. To allow for visualization of data at arbitrary spatial and temporal scales, we devise a multiresolution data management framework utilizing wavelet transform. In the algorithm, the underlying time-varying data set is converted into a space-time multiresolution data hierarchy, called wavelet-based time-space partitioning (WTSP) tree. We partition the WTSP tree and distribute the wavelet-compressed data across a cluster of computation nodes to achieve parallelism and load balancing. To allow the user to freely explore the data set, features must be tracked in an efficient manner. To achieve the goal, we devised an efficient time-varying isosurface tracking algorithm. Our algorithm can rapidly identify corresponding isosurfaces by utilizing higher dimensional geometry information. We also design a high-dimensional volume rendering algorithm that allows the scientist to visually track the features at interactive rates.