Date of Award
Program or Major
Master of Science
Flow visualization is the graphical representation of vector fields or fluids that enables an observer to visually perceive the forces or motions involved. The fields being displayed are typically dynamic and complex, with a vector direction and magnitude at every point in the field, and often with additional underlying data that is also of interest to the observer. Distilling this mass of data into a static, two-dimensional image that captures the essential patterns and features in a way that is intuitively understandable can be a daunting task.
Historically, there have been many different techniques and algorithms to generate visualizations of a flow field. These methods differ widely in implementation, but conceptually they involve the association of significant aspects of the data field (e.g., direction, velocity, temperature, vorticity) to certain visual parameters used in the graphic representation (e.g., size and orientation of lines or arrows, foreground and background color, density/sparsity of graphical elements). For example, the velocity of a field could be mapped to color, line width, line length, arrow head or glyph size, etc. There are many such potential parameter mappings within each technique, and many value ranges that can be used to constrain each parameter within a given mapping, resulting in a virtually limitless number of possible permutations for visually representing a flow field. So, how does one optimize the output? How can one determine which mappings and what values within each mapping produce the best results? Such optimization requires the ability to rapidly generate high-quality visualizations across a wide variety of parameter mappings and settings.
We address this need by providing a highly-configurable interactive software system that allows rapid, human-in-the-loop optimization of two-dimensional flow visualization. This software is then used in a study to generate quality visual solutions to a two-dimensional ocean current flow plus surface temperature over a variety of parameter mappings. The results of this study are used to identify relevant rules and patterns governing the efficacy of each combination of parameters, and to draw some general conclusions concerning 2D flow visualization parameter mapping and values.
Mitchell, Peter W., "The perceptual optimization of two-dimensional flow visualizations using human-in-the-loop local hill climbing" (2007). Master's Theses and Capstones. 339.