Does Variation in Quality Scores Result from Variation in Visual Attention?

Novel research in image quality assessment attempts to further improve the reliability of objective metrics by taking into account visual attention of the human visual system. Modeling visual attention in an objective metric, however, is not yet a trivial task, partly because earlier work has shown that the added value of including visual attention is content dependent. In this paper, we attempt to better understand the impact of human visual attention on image quality scoring by investigating the variation in visual attention between observers, and to what extent such variation affects the reliability of scoring the image quality. The results, however, show that the relationship between variation in visual attention and variation in quality scores is weak.