Abstract
Abstract: :
Purpose: The aim of this work was to test the ability of observers to detect spatial illuminant gradients in real three–dimensional scenes. Methods: Observers viewed a three–dimensional scene consisting of several coloured man–made and natural objects. The scene subtended 10 deg visual angle and was illuminated by a computer–driven LCD data projector (Epson EMP–5600 with 800×600 resolution and 2220 ANSI lumen) driven by a color–graphics card (VSG2/5; Cambridge Research Systems, UK) with 12–bit resolution in each of the R, G, and B channels. Light of correlated colour temperature of 25000 K always illuminated the left–hand–side of the scene and the right–hand–side could be illuminated by any of 21 different illuminants with correlated colour temperatures in the range 25000 K – 4000 K. The 21 illuminants were equally spaced in CIE 1976 (u',v';) colour space with constant luminance 158 cd/m2 measured at a barium sulphate plug placed in right and left–hand–side of the scene. Spatially, the scene was therefore illuminated by a variable gradient. In each trial of the experiment, observers having viewed the scene for 90 s reported whether the colour of the illuminant was uniform across the scene. There were 90 observers, who each performed a maximum of 3 trials, 1 trial per week to avoid colour–memory effects. Results: Observers could reliably detect that the illuminant was not spatially uniform only when the correlated colour temperature of the right side of the scene was smaller than 4348 K, thus corresponding to a change from 25 000K to 4348 K across the all scene. Conclusions: Observers are remarkably insensitive to spatial changes in the colour temperature of an illuminant, showing a level of illuminant constancy that may explain simultaneous colour constancy.
Keywords: color appearance/constancy • color vision • constancy