Non-linear redundant mapping of contrast in multivariate glyph displays

Abstract

Although capable of independent action for standard tasks, automated systems still need monitoring and support from human managers. In order for a single observer to effectively monitor large groups of these systems a summary display is necessary. This provides a top-level glance across a group of systems, allowing prioritization of actions. An effective tool for implementing this type of display is data visualization. Glyphs are a graphical method to summarize different information for a specific entity in a compact fashion. They use different aspects of their components (e.g. size or orientation) to represent real world information about a system (e.g. amount of fuel or direction of movement). By applying knowledge from visual attention and perception research, guidelines can be proposed for glyph usage. This dissertation focused on the use of contrast as a non-linear redundant cue to enhance perception of values in a specific range within a particular glyph feature. This method was tested to see if it would provide faster visual search across a set of glyphs when the observer was trying to find the glyph that had the most or least of a variable. Results suggest that using contrast in this manner is an effective means to improve performance. This effect however is moderated by several factors. For example, the number of the observed glyphs that are showing data in that range at one time as well as dynamic changes in the values being observed can degrade the effect. Overall, this application of contrast shows potential as a highlighting method for data variables where the monitoring of a small range of values could provide observers an advantage. Future research should investigate application of this method in more complex real world simulations to test if the results of this basic visual attention task generalize.