Local color pixel distributions provide information that is useful for object recognition but are dependent on the scene illumination. We develop a method that assigns color descriptors to an object that depend on the distribution of spectral reflectance across the object and not on the illumination. For a trichromatic system, the method assumes a three-dimensional linear model for surface spectral reflectance. We present examples demonstrating the system's ability to recognize model objects in cluttered scenes independent of scene illumination.
David Slater, Glenn Healey, "Illumination-Invariant Recognition of Local Color Distributions using Linear Models for Spectral Reflectance" in Proc. IS&T 3rd Color and Imaging Conf., 1995, pp 34 - 36, https://doi.org/10.2352/CIC.1995.3.1.art00009