Abstract
It has been shown that color constancy is improved in real environments compared to simple two-dimensional patterns. This fact suggests that natural environment is important for color constancy. Natural environment consists of a variety of materials and they could contribute to recognizing illumination and obtaining stable appearance of objects. In this study, we examine whether the materials of a test stimulus influence to the degree of color constancy. We built a booth arranged like a normal room illuminated by fluorescent lamps with correlated color temperature 5000K or 2700K. Test stimuli made of papers, plastic, wool were prepared. Four colors (gray, yellow-green, orange, and purple) were tested for each material. Observers evaluated the color appearance of test stimulus using an elementary color naming method. Two viewing conditions were tested: a natural viewing condition and a limited-view condition in which observers only viewed a test stimulus through a black viewing box. The results showed that in the natural viewing condition, differences on the color appearance of test stimuli under two illuminations were small, meaning good constancy. However, they did not show any clear differences between materials. In the limited-view condition, differences in color appearance between two illuminations were generally large, suggesting the low degree of color constancy. Some results, however, showed good color constancy even though the surrounding of test stimuli was dark. We further tested an additional condition using test stimuli with uniform color on a CRT monitor. The results suggested that the high degree of color constancy in the limited-view condition was due not only to an adaption in lower level but also material perception. Our results imply that material perception would not have large influence to color constancy in normal environment, whereas it would contribute to color constancy in a condition lacking information on natural environment.
Meeting abstract presented at VSS 2012