ROD CONTRIBUTION TO LARGE-FIELD COLOR MATCHING

The linear algebraic features of foveal colorimetry do not apply to 10 -degrees degree colorimetric fields. A light presented as a foveal 2-d egrees field stimulates the three classes of cone photoreceptors, and a light presented as a 10-degrees field stimulates the cone and rod ph otoreceptors. While no more than three primary lights are required to match either of these lights, large-field color matches with three pri maries do not obey Grassmann's laws of linearity because the relative contribution of the rod signal to a color match changes with retinal i lluminance. A three-primary set of color-matching functions (CMFs) aff ected by rod-intrusion, therefore, cannot serve as the basis for color imetry. Stiles and Burch removed the effects of rod activity from thei r 10-degrees color-matching data to create a set of CMFs that have the same linear algebraic features as foveal, 2-degrees colorimetry. They did this by selecting the primaries appropriately, making the matches at high retinal illuminance levels, and removing the small residual e ffects with numerical techniques. The resulting CMFs formed the major portion of the CIE 1964 10-degrees Standard Observer. These CMFs descr ibe color matches at the level of the cone-photoreceptor and, therefor e, can be generalized to predict color matches with any set of primari es, but they cannot account for the effects of rod activity in the mat ching sample. If two colorimetric fields differ in the amount of rod a ctivity they produce, calculations with the 1964 Standard Observer may predict their metamerism inaccurately. This article summarizes Stiles and Burch's methods for removing the effect of rods, gives examples o f the implications for colorimetry, and suggests procedures for preven ting the adverse effects of rod activity in applied situations. It als o shows that three recently published sets of 10-degrees CMFs may be i nfluenced by the effects of rod activity. (C) 1994 John Wiley & Sons, Inc.