ILAV   21219
INSTITUTO DE INVESTIGACION EN LUZ, AMBIENTE Y VISION
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
Does melanopsin help to explain color constancy in natural environments?
Autor/es:
BARRIONUEVO, PABLO A.; CAO, DINGCAI
Reunión:
Encuentro; AIC 2019 Midterm Meeting; 2019
Resumen:
Visual information is conveyed from the retina to the brain by the Magnocellular (MC), Parvocellular (PC) and Koniocellular (KC) pathways. The MC pathway codifies luminance information, while the PC pathway codifies ?red-green opponency? and the KC pathway the ?blue-yellow opponency?. These pathways have been shaped by the natural environment during evolution. Natural image statistics are the most efficient representations of environments and they can predict how neural responses should vary to encode them. A visual attribute that allows us to keep a stable color perception of the visual environment disregarding changes in chromaticity of natural illuminations throughout the day is color constancy. On the other hand, the recently discovered melanopsin photopigment has shown a unique photon-counting capability that may modulate visual processing. If melanopsin is related with color constancy, the perceptual invariance with respect to the illuminant should be function of melanopsin excitation. The aim of this work was to assess whether melanopsin could help to achieve color constancy. Data from hyperspectral natural images containing primordial foliage information and spectral reflectance information at each pixel (https://personalpages.manchester.ac.uk/staff/david.foster/), together with irradiance values of 21 daylight illuminants (Correlated Color Temperature: 3635K ? 24770K) covering different phases of the day from moon light to sun light, were used to compute L-, M-, S-cone and melanopsin excitations for each pixel. Using D65 illuminant as a reference, the geometrical distance d in the cone chromaticity [L/(L+M) and S/(L+M)] was computed between daylight illuminants and the reference illuminant for each pixel in each image. Distance d data were correlated to melanopsin chromaticities [I/(L+M)] using different multiplicative coefficient for each cardinal axis.For all of the scenes, higher correlated color temperature (CCT) values were negatively related to the coefficient of the melanopsin excitation in the L/(L+M) axis but positively related the coefficient of melanopsin excitation in the S/(L+M) axis.Based on these analyses, to achieve color constancy, melanopsin contribution should affect in opposite ways the KC [corresponding to the S/(L+M) axis] and the PC [corresponding to the L/(L+M) axis] pathways, but the sign depends on the CCT of the illuminant. For example, for CCT values related to clear sky conditions, KC signals need a positive melanopsin contribution, and the PC pathway a negative contribution. Instead scenes under low CCT illuminants, such as those related with dawn and dusk, have worse constancy when melanopsin is considered.