Spectral measurement of daylights and surface properties of natural objects in Japan

Abstract

We present a spectral dataset of daylights and surface reflectances and transmittances of natural objects measured in Japan. Daylights were measured under the sun and under shadow from dawn to dusk on four different days to capture their temporal spectral transition. We separately measured daylight spectra at five different locations (including an open space and a forest) with minimum time difference to reveal whether a local environment alters daylight spectra reaching the ground. We found that colors of natural objects were spread in a limited area of color space, and data points were absent around saturated green regions. Daylight spectra were found to have a larger variation across time, weather, and local environments than previously thought. Datasets are made freely available, expanding past public datasets mainly collected in Northern America and Europe.

Invariant categorical color regions across illuminant change coincide with focal colors

“…Another candidate would be the color of test illuminants we sampled. It is worth noting that in a previous study where temporal daylight variation within a day was measured ( Morimoto et al, 2022 ), we found that most daylights fell within the range between 4,500K and 20,000K. Thus, it is possible that the natural illuminants we used (3,000K or 25,000K) may have induced too great of illuminant changes.…”

“…To show the magnitude of each illuminant change, Figure 2 d shows the chromaticities of 5 test illuminants and the associated chromatic distribution of the 97 color samples at a mid lightness level ( L = 0 and 1) under each illuminant. The color temperatures 3,000K and 25,000K were selected as they were around the edges of the natural daylight variation ( Morimoto, Zhang, Fukuda, & Uchikawa, 2022 ). The red and blue illuminants were achieved by activating only red or blue phosphors of the projector, respectively.…”

Invariant categorical color regions across illuminant change coincide with focal colors

“…Another candidate would be the color of test illuminants we sampled. It is worth noting that in a previous study where temporal daylight variation within a day was measured (Morimoto et al, 2022), we found that most daylights fell within the range between 4,500K and 20,000K. Thus, it is possible that the natural illuminants we used (3,000K or 25,000K) may have induced too large of illuminant changes.…”

“…To show the magnitude of each illuminant change, Figure 2 (d) shows chromaticities of five test illuminants and the associated chromatic distribution of the 97 color samples at a mid lightness level (L = 0 & 1) under each illuminant. The color temperatures 3,000K and 25,000K were selected as they were around the edges of the natural daylight variation (Morimoto et al 2022). The ‘Red’ and ‘Blue’ illuminants were achieved by activating only red or blue phosphors of the projector, respectively.…”

Misidentifying illuminant changes in natural scenes due to failures in relational colour constancy

“…Similar inferences can be made with other changes in lighting, not just those due to the gradual elevation of the sun [1][2][3][4][5], but those that are more abrupt, as when direct light is interrupted by a passing cloud or by moving foliage [1,6]. The effects may be amplified by the complex structures of natural environments [7], which can produce large temporal and spatial variations of illumination at ground level, quantified in spatial [8], directional [9,10], and time-lapse scene measurements [11][12][13]. The ensuing changes in appearance constitute a failure of what is usually referred to as colour constancy, that is, the constant perceived or apparent colour of a surface despite changes in the intensity and spectral composition of the illumination [14 -18].…”