Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!apple!agate!eos!jbm
From: jbm@eos.UUCP (Jeffrey Mulligan)
Newsgroups: comp.graphics
Subject: Re: Gray levels and color
Keywords: color gray
Message-ID: <4347@eos.UUCP>
Date: 16 Jul 89 20:10:56 GMT
References: <20722@hodge.UUCP>
Organization: NASA Ames Research Center, California
Lines: 58

jdm@hodge.UUCP (jdm) writes:


>In a company I onced worked for there was a guy who could
>look at a black and white photograph of a prinited circuit board
>and read the color codes on the resistors, capacitors, wires, etc.
>I could look at the same photograph and make an educated guess
>of what the colors were based on the intensity of the gray levels
>and the most likely color combinations that would occur on a
>PC board, but I could not read them with 100% accuracy like
>this guy could.

>It turned out that this guy was color blind in such a way that
>he could see colors in levels of gray.  Apparently they were the
>correct colors for he never seemed to need to guess or deduce.
>He actually saw those colors on the B&W photograph.

Note that there are several types of "color blindness," including
red-blindness (protanopia) and green-blindness (dueteranopia).
For a protanope, reds will look dark, and greens light, while
for a deuteranope the opposite correlation between hue and
lightness will hold.  

Each type of photographic film has it's own spectral sensitivity;
some films are insensitive to red (and can therefore be handled under
a safelight) while others must be handled in total darkness.
Naturally, it is the goal of the film manufacturer to make
the film's spectral sensitivity correspond as closely as possible
to the normal eye's luminosity function; if your color-blind friend
could correctly read the resistor values from the photograph,
however, it was a lucky combination of circumstances.

> Now I am
>wondering "just what are the correlations between intensity
>(levels of gray) and color (either RGB or HSI)"?

There are no bright browns or dark yellows.  Other than that
there are no correlations imposed by nature, although in any
given scene or image there are bound to be some correlations.

>Reducing the Hue and Saturation in a photograph would leave only
>a gray level picture of the intensities.  How would one use such
>an intensity scale to reconstruct the orginal color of the 
>photograph using either RGB or HSI?

Not possible from a methematical standpoint.  My guess is that
for film colorization, a human operator just picks whatever colors
he thinks would look nice.


Jeff Mulligan


-- 

	Jeff Mulligan (jbm@aurora.arc.nasa.gov)
	NASA/Ames Research Ctr., Mail Stop 239-3, Moffet Field CA, 94035
	(415) 694-6290