Author: Hansen Thorsten Giesel Martin Gegenfurtner Karl R.
Publisher: Society for Imaging Science and Technology
ISSN: 2158-6330
Source: Conference on Colour in Graphics, Imaging, and Vision, Vol.2006, Iss.1, 2006-01, pp. : 119-124
Disclaimer: Any content in publications that violate the sovereignty, the constitution or regulations of the PRC is not accepted or approved by CNPIEC.
Abstract
Discrimination of different chromatic hues is a fundamental visual capability. Traditional measurements of color discrimination have used patches of a single homogeneous color. Everyday color vision, however, is based on natural objects which contain a distribution of different chromatic hues. Here we study chromatic discrimination using photographs of various natural fruit objects. In a fouralternative forced-choice (4AFC) experiment four stimuli were briefly presented in a 2×2 arrangement. Three of the stimuli were identical (test stimuli) and the fourth one (comparison stimulus) differed. The stimuli were either homogeneous patches of light or digital photographs of fruit objects. In addition to the natural stimuli we also used synthetic chromatic textures which resembled the natural objects both in spatial frequency and chromatic distribution. The stimuli were displayed on top of a homogeneous gray background. The mean color of the comparison stimulus was varied along eight different directions in color space relative to the test stimulus. Discrimination thresholds were measured along these eight directions and ellipses were fitted to the resulting threshold contours. We measured discrimination at the adaptation point and at 8 test locations away from the adaptation point. In agreement with earlier studies, we found that discriminability was best when the test stimuli had the same average color as the adapting background. However, when natural objects were used as stimuli, thresholds were elevated and threshold contours were elongated in a way that reflected the distribution of hues in the image. For test stimuli that had an average color different from the background, threshold contours for fruit objects and homogeneous patches were identical. The results for the synthetic color textures resembled those for the natural objects. We conclude that the distribution of hues within natural objects leads to a specific increase in discrimination thresholds along the direction of maximal chromatic variation.