Spectrally Based Material Color Equivalency: Modeling and Manipulation.

摘要

A spectrally based normalization methodology (Wpt normalization) for linearly transforming cone excitations or sensor values (sensor excitations) to a representation that preserves the perceptive concepts of lightness, chroma and hue is proposed resulting in a color space with the axes labeled W, p, t. Wpt (pronounced "Waypoint") has been demonstrated to be an effective material color equivalency space that provides the basis for defining Material Adjustment Transforms that predict the changes in sensor excitations of material spectral reflectance colors due to variations in observer or illuminant. This is contrasted with Chromatic Adaptation Transforms that predict color appearance as defined by corresponding color experiments. Material color equivalency as provided by Wpt and Wpt normalization forms the underlying foundation of this doctoral research. A perceptually uniform material color equivalency space ("Waypoint Lab" or WLab) was developed that represents a non-linear transformation of Wpt coordinates, and Euclidean WLab distances were found to not be statistically different from DeltaE* 94 and DeltaE00 color differences. Sets of Wpt coordinates for variations in reflectance, illumination, or observers were used to form the basis of defining Wpt shift manifolds. WLab distances of corresponding points within or between these manifolds were utilized to define metrics for color inconstancy, metamerism, observer rendering, illuminant rendering, and differences in observing conditions. Spectral estimation and manipulation strategies are presented that preserve various aspects of "Wpt shift potential" as represented by changes in Wpt shift manifolds. Two methods were explored for estimating Wpt normalization matrices based upon direct utilization of sensor excitations, and the use of a Wpt based Material Adjustment Transform to convert Cone Fundamentals to "XYZ-like" Color Matching Functions was investigated and contrasted with other methods such as direct regression and prediction of a common color matching primaries. Finally, linear relationships between Wpt and spectral reflectances were utilized to develop approaches for spectral estimation and spectral manipulation within a general spectral reflectance manipulation framework -- thus providing the ability to define and achieve "spectrally preferred" color rendering objectives. The presented methods of spectral estimation, spectral manipulation, and material adjustment where utilized to: define spectral reflectances for Munsell colors that minimize Wpt shift potential; manipulate spectral reflectances of actual printed characterization data sets to achieve colorimetry of reference printing conditions; and lastly to demonstrate the spectral estimation and manipulation of spectral reflectances using images and spectrally based profiles within an iccMAX color management workflow.