Device color-space characteristics and transformations are dependent on the illuminants used. The mismatch of white points is a frequently encountered problem. It happens in situations where the measuring and viewing of an object are under different illuminants, the original and reproduction use different illuminants, and the different substrates are under the same illuminant. To correct these problems, the transform of white points is needed. White-point conversion techniques are developed for converting between different illuminants strictly on the physical quantities, without any appearance transform. In this context, white-point conversion is different from chromatic adaptation or color constancy in which the illumination difference is treated with an appearance transform (see Chapter 3). For white-point conversion, we are asking the question: What are the tristimulus values (or other colorimetric specifications) of an object under illuminant A if the only information available is under illuminant B? Thus, it is strictly a mathematical transform.
In this chapter, we present four methods of white-point conversion. The first method is based on the transform between CIEXYZ and colorimetric RGB. By using a colorimetric RGB space as the intermediate connection between two white points, the white-point conversion is scaled by the ratios of the proportional constants of red, green, and blue primaries. The second method performs the white point-conversion via tristimulus ratios. The third method uses the spectral difference of the white points. Source tristimulus values are corrected from the white-point difference to obtain destination tristimulus values. The fourth method uses polynomial regression. Theory and derivation are given in detail for each method. Conversion accuracies of these methods are compared using a set of 135 data points. Advantages and disadvantages of these methods are discussed.
13.1 White-Point Conversion via RGB Space
Xerox Color Encoding Standards (XCES) provide a method for white-point conversion, utilizing an intermediate colorimetric RGB space because tristimulus values are a linear transform of the colorimetric RGB, as shown in Section 6.2.
Online access to SPIE eBooks is limited to subscribing institutions.