Abstract
Humans can quickly perceive various surface qualities such as glossiness and heaviness. Also, it seems intuitively plausible that complexity of image information contributing to perception are largely different across surface qualities; for example, perception of glossiness may depend on simpler image information such as low-level image statistics than heaviness. In this study, we psychophysically measured effects of stimulus durations on perception of different surface qualities as an index of complexity of image information contributing to perception. In the experiment, two of the 35 material samples were simultaneously presented on an LCD monitor for a duration in the range from 34 ms to 150 ms. The observer judged which of the two samples gave stronger perception of a certain surface quality with a 2AFC procedure (e.g., which one gave "stronger glossiness"). Seven surface qualities such as glossiness, warmness, and heaviness were employed as target qualities. Preference scale values were calculated on the basis of Thurston's scaling method for pairwise comparison as measures of perceived strength of surface qualities. In the results, magnitudes of effects of stimulus duration were significantly different between surface qualities; some surface qualities such as glossiness and warmness were well judged even under short durations, while other qualities such as heaviness were not. In an additional analysis, we calculated 23 low-level image statistics from the material samples used in the experiment and investigated relationship with surface quality perception. The analysis showed that perception of surface qualities, which were not strongly affected by stimulus durations, tended to be strongly related with low-level image statistics. For example, warmness was strongly related to mean chromaticity regardless of the stimulus duration. These results suggest that complexity of image information contributing to surface quality perception may considerably differ across various surface qualities, and specific surface qualities may strongly depend on simple image information.
Meeting abstract presented at VSS 2016