Visualization of Time-Varying Multiresolution Data Using Error-Based Temporal-Spatial Reuse

摘要

In this paper, we report results on exploration of two-dimensional (2D) time varying datasets. We extend the notion of multiresolution spatial data approximation of static datasets to spatio-temporal approximation of time-varying datasets. Time-varying datasets typically do not change 'uniformly,' i.e., some spatial sub-domains can experience only little or no change for extended periods of time. In these sub-domains, we show that approximation error bounds can be met when using sub-domains from other time-steps effectively. We generate a more general approximation scheme where sub-domains may approximate congruent sub-domains from any other time steps. While this incurs an O(T2) overhead, where T is the total number of time-steps, we show significant reduction in data transmission. We also discuss ideas for improvements to reduce overhead.