Abstract
The motivation for analyzing a person’s gaze be- havior stems from the eye-mind-hypothesis that correlates the observation of content by viewing it with its cognitive comprehension. A standard approach to quantitative analysis of gaze behavior is to analyze so called heatmaps generated by an eye tracker. This technical process is subject to noise. In order to compensate it, state-of-the-art eye ...
Abstract
The motivation for analyzing a person’s gaze be- havior stems from the eye-mind-hypothesis that correlates the observation of content by viewing it with its cognitive comprehension. A standard approach to quantitative analysis of gaze behavior is to analyze so called heatmaps generated by an eye tracker. This technical process is subject to noise. In order to compensate it, state-of-the-art eye tracking software smooths raw data using linear filters. In this paper, we present an alternative method based on interpolation. We provide empirical data that our method reproduces the actual gaze behavior more precisely. Furthermore, we introduce Empirical Heatmap Decomposition (EHD) to cluster eye movements into classes of similar frequency and amplitude. For evaluation, we present an analysis of gaze data that illustrates how EHD can uncover details in the observed gaze behavior that state-of-the-art heatmaps do not visualize.
More literature (via CORE)