Color Bands: Visualizing Dynamic Eye Movement Patterns Michael Burch 1 , Ayush Kumar 2,3 , Klaus Mueller 3 and Daniel Weiskopf 1* 1 VISUS, University of Stuttgart 2 SUNY Korea 3 Stony Brook University Figure 1: A color band showing the time-varying eye movement patterns of an individual participant (time along the horizontal axis): x- and y-coordinates of gaze data as well as their differences can be visually inspected by observing the vertical positions, thickness, and color of the band. ABSTRACT We introduce a visualization technique called color bands for show- ing the time-varying eye movement behavior of eye-tracked people. Our contribution is the clutter-free representation of time-varying x- and y-positions of gaze data. We map these coordinates to ver- tical positions from left to right as in traditional line plots. On top, we display the differences between the x- and y-coordinates by the thickness of the band. Fixation durations are visually encoded as circles of varying diameters. Color coding is used to additionally enhance the distance values and the durations in order to percep- tually benefit from pattern recognition for an individual participant but also to compare the eye movement behavior of several partici- pants. We illustrate the usefulness of our technique in a case study investigating eye movements from a formerly conducted eye track- ing study on the readability of node-link tree diagrams. Index Terms: H.5.2 [Information Interfaces and Presentation]: User Interfaces—Graphical user interfaces (GUI); 1 I NTRODUCTION This paper addresses a problem of visual data analysis of eye track- ing data acquired for individual participants and groups of partici- pants alike. It is difficult to fully analyze and understand all facets of such spatio-temporal data, especially for long eye movement tra- jectories and large numbers of participants. Many existing visu- alization concepts aggregate over participants and over time like visual attention maps [3, 16] or they show the time-varying be- havior of many people as color-coded polylines plotted on top of each other [1, 15], producing visual clutter [14]. Both concepts— referred to as heat maps or gaze plots—are common visual repre- sentations of eye tracking data; however, they either show only half of the truth or do not visually and perceptually scale to large eye movement datasets. In this paper, we argue that a clutter-free static visualization can have many benefits for getting an overview of the time-varying visual attention of several participants taking part in eye tracking studies. To reach this goal we introduce color bands: they encode time along the horizontal axis (in left-to-right reading direction), saccade lengths are proportionally mapped to x-axis sections, sac- cade orientations to vertical gradients, the x- and y-positions of * e-mail: {michael.burch, weiskopf}@visus.uni-stuttgart.de, {aykuma, mueller}@cs.stonybrook.edu gazes to vertical positions, and the differences between x- and y- coordinate values to vertical distances and color. The distances be- tween two consecutive fixations are indicated by the changes of the band shape over time, whereas the fixation durations are encoded as circles of different sizes placed at the end of each polyline segment, see Figure 1. This visual mapping allows us to explore the visual attention of several participants in parallel, avoiding any aggrega- tion or overplotting. Moreover, it results in aesthetically appealing diagrams. We illustrate the usefulness of our novel visualization technique for the example of eye movement data from a formerly conducted eye tracking study [5, 6]. We identify static and dynamic visual patterns, but also patterns of individual participants and groups of them, and finally, have a look at scalability issues concerning algo- rithmic, visual, and perceptual aspects. 2 RELATED WORK There are several visualization techniques dealing with eye move- ment data, as surveyed by Blascheck et al. [2]. Most of them ag- gregate eye movement data over space, time, or participants. For example, heat maps or visual attention maps [3, 4, 16] show a con- densed view of the visual attention of a group of participants ag- gregated over a certain time period. Only the hot spots of visual at- tention are visible by visually inspecting the color-coded hot spots in a corresponding diagram. In contrast, gaze plots [1, 15] do not aggregate over time nor over participants, but overplotting of many polylines leads to visual clutter [14], “a state in which excess items or their disorganization leads to a degradation of performance at some task.” AOI rivers [7] show the time-varying behavior of many eye tracking study participants, but they aggregate over all participants, making it impossible to identify similar strategies of certain peo- ple over longer time periods. This Sankey-based [17] diagram style merges and splits, i.e., single trajectories cannot be traced. The ap- proach is based on the ThemeRiver visualization [9], which uses vertically stacked bands to indicate the number of visits to Areas of Interest (AOIs) over time. We do not specifically focus on AOIs nor do we vertically stack the color bands in our color bands vi- sualization. Our approach is more related to the CloudLines tech- nique [10] using non-stacked, but horizontally time-aligned rivers, which helps better visualize value changes [8] over time and visu- ally compare them with the others. Approaches like parallel scanpaths [13] are other visualization techniques that somewhat resemble our work, but they require the definition of AOIs and lack the visual representation by color bands. We have also added visual features for the fixation duration and ad-