Monday, October 22, 2007

Color space dimensionality

Today RocketRoo posted a comment to my short August post on a paper on Multiscale contrast enhancement. Since that is a few months ago, I will reply with a new post. Here is the comment:

Re: Multiscale contrast: achromatic dims

In this recent study, http://compbiol.plosjournals.org/perlserv/?request=get-document&doi=10.1371/journal.pcbi.0030179, Vladusich, Lucassen and Cornelissen provide evidence that brightness and darkness form the dimensions of a two-dimensional achromatic color space. This color space may play a role in the representation of object surfaces viewed against natural backgrounds, which simultaneously induce both brightness and darkness signals. The 2-D model generalizes to the chromatic dimensions of color perception, indicating that redness and greenness (blueness and yellowness) also form perceptual dimensions. Collectively, these findings suggest that human color space is composed of six dimensions, rather than the conventional three.

Posted by RocketRoo on 10/22/2007 11:45 AM

Let me first admit that I just read the first few paragraphs and not the complete article cited.

I am in violent disagreement with the authors. The opponent color model was first proposed by Leonardo da Vinci, (see chapters CLX and CLXII of his Trattato della Pittura, Langlois, Paris, 2nd edition, 1701) then discussed by Wolfgang Goethe (see also here) in his virtual diatribe with Isaac Newton. The first modern theory of color opponency was proposed in 1872 by Ewald Hering and was very hotly debated until G.E. Müller and Erwin Schrödinger reconciled Helmholtz's and Hering's theories in the zone theory of color vision, which they based on the 1904 law of coefficients proposed by Johannes A. von Kries.

The matter was finally settled in 1956 when Gunnar Svaetichin was able to record from horizontal fish retinas and show opponent response in red-green and yellow-blue potentials. At the time, he showed that each horizontal cell is presumed to inhibit either its bipolar cells or the receptors, with further processing occurring in the amacrine cells and in the retinal ganglion cells.

You an always postulate a mathematical model approximating some phenomenon, but at the end what counts is physics and you cannot contradict the physiology on which color vision is based.

What may be confusing the authors is that much of color science is for aperture color. If the appearance mode is different, then part of ordinary colorimetry falls apart. This is why there are color appearance models and the authors should have based their work on these.