Color constancy in natural scenes independent of an explicit illuminant cue

The ability of observers to make reliable judgments of surface color despite changes in an illuminant (“color constancy”) has sometimes been attributed to their estimating the spectral properties of the illuminant in order to discount its effects. To test how information about the illuminant contributes to surface-color judgments, observers’ performance in two surface-color-matching experiments with natural scenes, acquired with a high-resolution hyperspectral imaging system, was subjected to a multivariate analysis of variance. In the first experiment, the sky was directly visible to the observer, and its color was varied. In the second experiment, a large gray sphere was introduced into the scene so that its illumination by the sun and sky was directly visible to the observer and the color of that illumination was varied. Although observers’ surface-color matches varied across conditions, there was no reliable effect of the illuminant cue. Even when the sky was invisible, performance did not worsen. Judging surface color in natural scenes seems not to require knowledge of the illuminant. Introduction Human observers can make reliable judgments of surface color despite changes in the spectral properties of an illuminant that is not in the field of view. Some theories of this color constancy assume that the observer estimates the spectral properties of the illuminant, in order to compensate for its effects, an estimation that is theoretically possible in some circumstances [1-4]. For the task of matching colored papers in Mondrian-like patterns, it appears to be unnecessary [5], but it is possible that in scenes where explicit knowledge about an illuminant is available rather than indirect inference, there might be an effect on observers’ judgments. To test this hypothesis, two surface-color-matching experiments were performed with images of natural scenes obtained from high-resolution hyperspectral images [6]. The test surface in each scene formed part of the scene and an explicit cue to the illuminant was manipulated in different ways to bias the information about its spectrum. By contrast with analyses of observers’ performance summarized by a univariate constancy index [7], which may mask the effects of some stimulus manipulations, here CIE u′, v′ values of observers’ matches were subjected to a multivariate analysis of variance (MANOVA). In the first experiment, the sky (but not the sun) in the scene was clearly visible to the observers, and its color was varied. In the second experiment, a large gray sphere was introduced into a scene so that the illumination upon it (avoid preempting details in Stimuli) was also clearly visible, and the color of that illumination was varied. As detailed later, the MANOVA revealed no significant effect of illuminant color in either scene, nor an effect of initial test surface color. Judging surface color in natural scenes seems to be independent of knowledge of illuminant color.