Abstract
It has been shown that human perception of surface gloss depends on the correspondence between specular highlights and the underlying surface shading. However, it is not known how gloss perception might depend on surface roughness. Increased roughness could increase the perceptual evidence in support of a gloss interpretation. Alternatively, roughness may increase the computational demand on estimating detailed shape from the underlying surface shading, resulting in a loss of correspondence information. We measured how human judgments of glossiness depend on surface roughness as a function of the spatial congruence between matte shading and specular highlights. We simulated surfaces of varying roughness by modulating depth according to a random Brownian process. We then manipulated the perceived gloss by rendering separate matte and gloss image layers, and combining these layers with various spatial displacements (Anderson & Kim, 2009). Subjects performed a 2AFC comparison task in which they indicated which of two surfaces looks glossier. Consistent with previous results, we found that the discriminability of gloss dropped as a function of the congruence between the gloss and matte layers. In addition, we found that the ability to distinguish gloss from matte increased with surface roughness, supporting the idea that increased evidence for correspondence improves gloss perception
Meeting abstract presented at VSS 2014