Abstract
Visual clutter is intuitive: Given a visual scene, we can easily get a sense of how cluttered it is. However, quantifying the amount of clutter in a scene is not trivial. To this date, several mathematical models of clutter have been proposed. Some of these metrics linked visual clutter in scenes to the number of objects that are present in a scene, their organizations, and texture elements. However, our understanding of the role of these factors in measuring quantitative vs. perceptual clutter is limited. To examine how these factors contribute to quantifying clutter, we generated a set of semi-realistic desk images with multiple objects arrangements. We varied the number of the objects (set size), their organization (clean or messy), and the complexity of the desk surface textures (high or low). We then measured the amount of clutter in these images using several existing clutter metrics. Our results demonstrate significant correlations between clutter metrics. However, further analysis revealed some variations. The effect of set size or the complexity of the background texture on clutter measurements depends on the choice of the clutter metric. Overall, our findings suggest that existing clutter metrics make use of different features while measuring visual clutter in naturalistic images. Further experiments will expand these results to human performance measures to understand perceptual implications.