Time-Preserving Visual Attention Maps
Michael Burch · Smart innovation, systems and technologies · 2016
Exploring the visual attention paid to a static scene can be done by visual analysis in form of attention maps also referred to as heat maps. Such maps can be computed by aggregating eye movement data to a density field which is later color coded to support the rapid identification of hot spots. Although many attempts have been made to enhance such visual attention maps, they typically do not integrate the time-varying visual attention in the static map. In this paper we investigate the problem of incorporating the dynamics of the visual attention paid to a static scene in a corresponding attention map. To reach this goal we first compute time-weighted Voronoi-based density fields for each eye-tracked person which are aggregated or averaged for a group of those people. These density values are then smoothed by a box reconstruction filter to generate aesthetically pleasing diagrams. To achieve better readability of the similar color values in the maps we enrich them by interactively adaptable isolines indicating the borders of hot spot regions of different density values. We illustrate the usefulness of our time-preserving visual attention maps in an application example investigating the analysis of visual attention in a formerly conducted eye tracking study for solving route finding tasks in public transport maps. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.