Representing Hierarchies of Visual Regard in Eye-Tracking Analysis

Abstract

Abstract The author developed a method for representing data from eye-tracking recordings: a form of graphical analysis that illustrates hierarchical densities of visual regard without obscuring the original pictorial stimulus. Across three different case studies, subjects’ fixation patterns were used to propagate Voronoi generating points. Integrating both fixation locations and their respective dwell times, randomized Gaussian distribution provided a technique to augment Voronoi generating seeds and enhance graphical resolution. Color pixel values were then used to fill in resultant Voronoi cells in relation to color values provided by the original stimulus. This analysis allowed for effective differentiation of viewing behaviors between different subjects, with emphasis on a subject’s attentional distribution rather than on graphic icons.