A spectral classifier for pixels on the basis of measurements made with two or more broad, spectrally-overlapping sensitivity curves has been called Artificial Color, because animals appear to do the same thing. While Artificial Color appears very attractive in tests to date, it has the drawback of being pixel-by-pixel. We know there are neighborhood effects in the perceived color in animals. The purpose of this article is to introduce and study the effects of neighborhood operations in Artificial Color.
Jian Fu, H. John Caulfield, Terrence Mizell, "Applying Median Filtering with Artificial Color" in Journal of Imaging Science and Technology, 2005, pp 498 - 504, https://doi.org/10.2352/J.ImagingSci.Technol.2005.49.5.art00007