Perception of Colors: Comparing sRGB Triples & Metrics

[CRGreathouse]

I'm interested in comparing colors (represented as sRGB triples) in terms of human perception. I vaguely recall something about eyes being more sensitive to differences in lightness than to differences in hue or saturation. Does anyone have any information on this?

Better yet, any good metrics for color difference?

 

[mgb_phys]

http://upload.wikimedia.org/wikipedia/commons/9/9a/Optical_grey_squares_orange_brown.svg

See http://en.wikipedia.org/wiki/Color_constancy for some references

 

[CRGreathouse]

Hmm. That's something, but not really what I'm looking for. Those algorithms take pictures as inputs and attempt to determine what color each point represents in the context of the picture. I'm looking for something that takes two colors and gives a distance.

For example, d(A, B) could be the Euclidean distance
sqrt((A.red - B.red)^2 + (A.green - B.green)^2 + (A.blue - B.blue)^2)
but I'm looking for something that more closely reflects human perception.

 

[rcgldr]

For humans there are 3 and sometimes 4 types of color receptors, that have sensitivity curves to certain colors, but it's not quite true red, blue, and green (for the 3 receptor people). There is also a set of shades of grey receptors which work better in low light.

http://en.wikipedia.org/wiki/Color#Color_perception

 

[CRGreathouse]

For humans there are 3 and sometimes 4 types of color receptors, that have sensitivity curves to certain colors, but it's not quite true red, blue, and green (for the 3 receptor people).

True. But I'm working with 3-color devices, so the sensitivities to any wavelengths beyond those three are actually irrelevant to me. :)

http://en.wikipedia.org/wiki/Color#Color_perception

I just found
http://en.wikipedia.org/wiki/Color_difference
which may be what I want. Let me read it and ask the knowledgeables here any questions I may have, :D