Outline: Color



Two provocative statements
    We are all color blind.
    Color is a perceptual construction that is not, per se in the physical stimulus.

Physics of light - reprise
Light as a wave
Frequency/wavelength
Spectral power distribution
Monochromatic versus broadband lights
White light is a mixture
Various mixtures are perceived as white

Color mixture
Additive
Methods
Mixing lights (multiple slide projectors)
Mixing over space (e.g., television)
Mixing over time (color wheel)
Primary Colors (red, green and blue, can get most colors from these)
Subtractive
Methods
Filters/gels
Paints/dyes
Filters/gels are really multiplicative (of spectral transmission characteristics), not subtractive
Paint mixing is more complicated to model.
The perceptual result of additive mixture is predictable given the perceived colors of the components
For subtractive color mixture this is not the case

Aside:  "Dimensionality''
"Space is 3-dimensional."  This means that you can describe a location in space with 3 numbers, such as (x,y,z) for some fixed coordinate system. Lots of different sorts of coordinate systems work.  If they cover the space and are not redundant, then they require 3 numbers.
The space of all possible visible lights is infinite-dimensional (all possible spectral power distributions)
The space of all possible perceived colors is 3-dimensional (e.g., hue, saturation and brightness).
Hue:  the color itself (blue, green, lime green, chartreuse, ...)
Saturation:  the intensity of color, that is, the degree of pure color versus pastel, that is, the amount of white that has been mixed in
Brightness/Value:  the percept of physical intensity (independent of hue and saturation)

Psychophysics of color mixture and color matching
The task: color matching
Additive structure of the task
Primaries and test patch
Matchable test patches
Result:  3-dimensionality of color space (trichromacy)
Relate to the CIE chromaticity diagram
Monochromacy/Dichromacy ("color blindness") = 1- and 2-dimensionality
Relate to the principle of univariance
Thus there are pairs of lights that differ physically (have different spectral power distributions) but are indistinguishable to human observers.  These are called metamers.

Color Deficiencies
Dichromats
Protanope - missing the L cone
Deuteranope - missing the M cone
Tritanope - missing the S cone
Monochromats

Color opponency
Color afterimages
Unique hues (Hurvich & Jameson)
Hering opponent process theory: red-green, yellow-blue, black-white
Physiological evidence:  color opponent and double opponent receptive fields
Simultaneous color contrast
What is "brown"?
Color afterimages

Color image formation
Illuminants:  spectral power distribution
Surfaces:  spectral reflectance function
Lights/Signals:  spectral power distribution (the product of the above two)
Receptors:  spectral sensitivity function
The resulting color signal:  3 numbers (L-, M- and S-cone responses)
Color constancy as illuminant estimation and why it's difficult