Abstract
Abstract: :
Purpose: In natural complex environments the spectral composition of surface illumination varies with location and direction owing to occlusions, mutual reflections and solar elevation. But in spite of their importance to human color vision, these physical variations have received little attention. The aim of the present work was to characterize the spatial and directional distributions of surface illumination in natural scenes with a hyperspectral imaging technique. Methods: Images of rural and of urban environments were obtained with a hyperspectral imaging system in the Minho region of Portugal (Foster et al., 2004, Visual Neurosci., 21, 331–336). The system comprised a low–noise Peltier–cooled digital camera with a spatial resolution of 1344 x 1024 pixels coupled to a fast–tunable liquid–crystal filter mounted in front of the lens. Spectral–radiance functions at each pixel were estimated from gray reference surfaces located in the scene and from calibration data obtained with a telespectroradiometer. To estimate the surface illumination for different locations and directions, grey spheres with a uniform covering of neutral matt paint were placed in each scene. The CIE 1931 (x, y) chromaticities of the illumination and the corresponding correlated color temperatures (CCT) were computed as a function of location and direction. Results: Within the same scene, standard deviations of illumination chromaticities were smaller than 0.04 in CIE (x, y) units and about the same in both axis directions. They were a little larger for urban than for rural scenes. The CCT of the illumination within the same scene varied by about 4000 K and over all scenes from 4000 K to 8000 K. Conclusions: The variation of the spectral composition of surface illumination within natural scenes for the range of scenes and illuminants tested here was generally less than half the range of recorded daylight spectral changes, allowing the visual system to develop efficient local color–constancy mechanisms.
Keywords: color vision • color appearance/constancy • imaging/image analysis: non-clinical