Date of Award

Fall 2010

Project Type

Thesis

Program or Major

Computer Science

Degree Name

Master of Science

First Advisor

R Daniel Bergeron

Abstract

The process of making observations and drawing conclusions from large data sets is an essential part of modern scientific research. However, the size of these data sets can easily exceed the available resources of a typical workstation, making visualization and analysis a formidable challenge. Many solutions, including multiresolution and adaptive resolution representations, have been proposed and implemented to address these problems.

This thesis describes an error model for calculating and representing localized error from data reduction and a process for constructing error-driven adaptive resolutions from this data, allowing fully-renderable error driven adaptive resolutions to be constructed from a single, high-resolution data set. We evaluated the performance of these adaptive resolutions generated with various parameters compared to the original data set. We found that adaptive resolutions generated with reasonable subdomain sizes and error tolerances show improved performance daring visualization.

Share

COinS