Color atlas theory

Publisher Website
Google Scholar
Abstract
A domain of color science based on color atlases has to be sharply distinguished from conventional colorimetry. If essentially colorimetric results are desired, a còlor atlas is at a severe disadvantage. Even a technically flawless atlas will yield results that depend on ideosyncrasies of the observer (e.g., if the observer is an anomalous trichromat, this shows up in the results) even if the light source is within strict tolerances. If the source is not well specified, the results will degrade even further. However, the use of a color atlas also has some very important advantages. These are, apart from the obvious practical assets such as cost savings, speed, and intuitive trustworthiness, as follows: first, if a sample looks like a fiducial sample (or a small environment of fiducial samples) for several illuminants, then its bidirectional reflection function (BDRF) must be similar to that of the fiducial samples; and second, the equality of materials can be established by any observer (anomalous or not) if the illuminant is varied. Thus the use of an atlas leads to a physical classification (i.e., one based on the BDRF’s) rather than a merely colorimetric classification. In order to serve well, the atlas must be composed of physically well-specified fiducial samples, and different atlases may be needed for different tasks. In this paper the rational construction of such atlases to optimize various desirable properties is discussed.

This publication has 17 references indexed in Scilit: