Graph-Based Analysis of Visual Scanning Patterns: A Developmental Study on Green and Normal Images

Abstract

AbstractThe present study investigated the visual scanning pattern of children with typical development in three different age groups(4–6,6–8,8–10 years old). We used a data set from one related research, which included images with different low-level features: Green and Normal. This study analyzed age-associated inter-individual differences and was intended to show that graph profiling combined with a fixation time approach could help us to better understand the developmental visual pattern. Thus, degree centrality as one of the graph theory measures was implied to analyze gaze distribution. We explored the influence of bottom-up features, comparing the first 2 s (early phase) with the interval from 4 to 6 s (late phase) of scene exploration during age development. Our results indicated that degree centrality and fixation time increased with age. Furthermore, it was found that the effects of saliency are short-lived but significant. Moreover, we found that Green images during the early phase play an important role in visual anchoring, and the children’s performance was significantly different between 4–6 y and 6–8y-group. This comparative study underscores the ability of degree centrality as a developing innovative measure to perform eye-tracking data analyses.