A normal observer has trichromatic color vision that is determined by the three
types of cones (L, M, S that stand for long, middle and short wavelength).
Every color can be represented by three variables. Adding luminance as the
fourth variable will completely describe the perception of the steady light from
an object with certain size.
The values of these variable for certain object can be calculated from its
spectral radiance. A set of these values can correspond to unlimited number
of spectral radiances. In other words, these spectral radiances match each other
and are perceived as the same color.
The neural visual pathways start from cones and end at visual cortex. It is commonly
accepted that the color matching takes place at cone level. In other words,
if the outputs of cones are the same, then the colors are the same. Logically,
the outputs of the three types of cones are the ideal coordinates of color space
- LMS space. The spectral sensitivities of cones are used to calculate the
outputs of cones.
Unfortunately the widely used color space 1931 CIE coordinates are not LMS space.
The author strongly recommends the review by
Stockman and Sharpe(1999) to understand the principles, history, legacy
of color space. They also proposed a set of cone fundamentals (spectral sensitivities).
CIE 1931 color space is used here simply because it has been used as an international
standard for color monitor manufacturing even its accuracy has been seriously challenged.