Abstract
Since colors in a natural scene are usually inhomogeneous our color vision must have some mechanism that analyzes inhomogeneous color information to discriminate two scenes. We measured effects of density and color distribution of inhomogeneous stimulus patches on color discrimination. The stimulus was an array of isoluminant random-patches displayed on a CRT monitor. We changed size of each patch and color distribution of patches. The MacLeod-Boynton chromaticity diagram was used to express chromaticity coordinates of the stimulus. The direction of color distribution was on the r axis while discrimination was carried out along either on the r axis or in the b axis. It was shown that the discrimination threshold increased linearly as the patch size increased up to about 10 minutes, and then reached an asymptote with larger sizes. When the color distribution width of stimulus patches was set 20 to 40 times of the uniform discrimination threshold (uniform JND) the discrimination threshold was found to be larger than the uniform JND although the mean color of all patches was constant regardless the color distribution. This means that our color vision cannot use exactly the mean color of the distribution when comparing two stimuli. This result suggests that we employs a different criterion of discrimination when the stimulus is inhomogeneous. When color distribution was on the r axis and color discrimination was performed along b axis, the threshold did not increase, which indicates that at least two independent chromatic mechanisms exist in inhomogeneous color discrimination.