The modified Bradford chromatic adaptation transform (CMCCAT2000) is a von Kries type model of adaptation that best accounts for a variety of corresponding colour data sets. The transform works in three stages. First, XYZs arc linearly mapped to a new 'RGB' space. The RGB sensitivities are somewhat like the cones but have their sensitivity concentrated in narrower regions of the visible spectrum. In the second stage of the transform, the Red, Green and Blue responses are multiplied by three scalars to model illuminant change. Finally, RGBs are transformed back to XYZs (in order that well established formulae for colour appearance might be used). The Sharp adaptation transform (SAT), derived from theoretical arguments, is exactly the same as CMCCAT2000 except that the sharp RGB sensor sensitivities have even narrower support. Research has shown that the SAT delivers, statistically, the same performance as CMCCAT2000. In this paper we consider whether there is any reason why CMCCAT2000 or SAT might be preferable if adaptation is considered from an observers viewpoint. Our argument builds on the premise that an observer in making a corresponding colour match is matching surface reflectance. Starting with this premise the adaptation problem is clearly ill-posed: a pair of different surface reflectances might look the same under one light but different under another (this is the metamerism problem). However, we argue that an observer understands this metamerism and so seeks only to make a plausible reflectance match. Let us suppose a reflectance B viewed under a target light is chosen to match reflectance A viewed under a reference light. We say B is plausible if and only if it is possible that if B is also viewed under the reference light it is identical to A. Adopting this definition of plausibility, we found that the Sharp Adaptation Transform supports plausible adaptation but that CMCCAT2000 supports implausible adaptation.
展开▼
