In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

Marzieh Berenjkoub

will defend her dissertation proposal

Quantify and Visualize Attribute Relations for Flow Analysis

Abstract

Unsteady vector fields (or flows) and their analysis are of paramount importance to a wide variety of scientific and engineering applications. Despite significant advances in the analysis and visualization of unsteady flows, the interpretation of their behavior still remains a challenge. Previous works predominantly focus on the depiction of the geometric characteristics in the flows, which cannot sufficiently reveal the underlying physics of the flow behaviors that are interesting to the experts. To address that, this dissertation work concentrates on the physical properties or attributes of the unsteady flows. In addition, rather than looking at one attribute at a time, this dissertation work proposes to study the behaviors of multiple attributes together at once, which currently receives little attention in the flow visualization community. To achieve this goal, the pairwise relations among attributes are first measured, including their linear correlation and non-linear dependency. Based on these pairwise relations, the spatial relations of the attributes, relations among three or more attributes and the relations of attributes in ensemble data (or data with uncertainty) will be investigated. To help effectively show the relations of attributes for an informative interpretation of the flow behaviors, a number of novel visualization techniques are introduced, including the encoding of relation information along pathlines, a ranking based segmentation, and a novel relation glyphs. The proposed attribute relation measurements and their visualizations have been applied to a number of 2D and 3D unsteady flows to evaluate their effectiveness. Important flow characteristics that cannot be revealed with the previous methods are captured by the proposed methods.

Date: Wednesday, November 7, 2018
Time: 11:00 AM
Place: PGH 550
Advisors: Dr. Guoning Chen

Faculty, students, and the general public are invited.