The results from the ANOVAs showed that crops were perceived to vary significantly in shininess within most of the crop sets. The presence of highlights on a surface is a well-known image feature for the perception of glossiness. According to
Beck and Prazdny (1981), glossiness perception depends on the local presence of highlights, meaning that the direct area surrounding the highlight is perceived to be glossy, but not the whole surface per se. That is, they argue that glossiness perception is the direct response to local visual information, and not the result of some perceptual inference about the reflectance properties of the whole surface. Similar results are discussed by
Berzhanskaya, Swaminathan, Beck, and Mingolla (2005). They found that perceived gloss decreases as a function of the distance from the highlight. Thus when different parts of an object are considered, gloss perception will differ among the different parts depending on their vicinity to the highlights. This local quality of glossiness is in agreement with our results on the perception of shininess differing between the crops of a fabric. We used three image features (mean luminance, coverage of the highlight and contrast of the highlight) to further analyze this relationship between the local image content and the evoked perception. In
Figure 12 (top), we showed that the mean luminance of the crops was highly and positively correlated with almost all the crop sets for shininess. This finding is in line with
Wiebel, Toscani, and Gegenfurtner (2015), who found that the mean luminance of photographs of real materials was a high-performance predictor, followed by the standard deviation of luminance, to differentiate between glossy and matte materials. Highlights are high-luminance regions of the surface, explaining the high correlation we observed between the mean luminance of the crops and the perceived shininess. Coverage of the highlights was also highly correlated with the perceived shininess for most of the crop sets. Coverage of highlights has been shown to be strongly associated with glossiness perception (
Marlow, Kim, & Anderson, 2012;
Marlow & Anderson, 2013), especially when coverage is the most reliable cue for the judgement of glossiness. This happens with objects whose shapes create higher variability in highlights’ coverage rather than contrast or sharpness, under the same illumination. For our stimuli, within the same fabric, the folding configuration caused high variations of coverage that we found to be related to significant variations in shininess perception between the different crops of a fabric. High highlights’ coverage is also related to higher mean luminance, given that the area of the surface covered with highlights, that is, the high-luminance regions, increases. We indeed found the correlation between the mean luminance and the coverage averaged over all the crop sets, to be high and significant (
r = 0.78
p < 0.001). The third image feature that we measured, the highlights’ contrast, overall, was not strongly correlated with perceived shininess. In the three cases in which high and significant positive correlations were found, the contrast was also positively correlated with coverage. The opposite occurred for the only crop set that showed a significant negative correlation between contrast and shininess, that is, the high-contrast highlights covered the smallest regions of the fabrics’ surface.