# Decomposing light into frequency spectrum

Light hits a charge coupled element. The wavelength of the light somehow is translated into a color picture. Where can I learn about methods (algorithms) to decompose light hitting a CCD into frequency spectrum?

-
Note that computational questions per se are off-topic ad should be taken to ScientificComputation. That said, the physics which you are computing about remains on-topic. And, yeah, that leaves some ambiguity. We do the best we can. –  dmckee Mar 14 '13 at 14:07
thanks. I am new here... 😃 –  SpaceDog Mar 14 '13 at 14:08
Accidently, it was exactly a physics question. –  Rafael Reiter Mar 14 '13 at 14:09
not really accidentally... –  SpaceDog Mar 14 '13 at 14:15

A color CCD is made of a monochrom CCD Sensor and an array of filters. The common Bayer pattern is a pattern of 2 green, a blue and a red filter. The filters transmits light on your broad band CCD sensor. A color CCD already does some spectral analysis.

The four physical pixels are read into RGB channels of one color pixel. If you want to see the spectrum of color than see a histogram of RGB channels in your favorite graphics tool.

E.g. the green bandpassfilter allows transmittance of several light frequencies $\nu = \frac{c}{\lambda}$ in the green and possibly overlapping with blue and red filters. Usually the wavelength $\lambda$ is used to define the color of the light.

This relation for visible light is visualized in the spectrum of light. Imagine the intensity on the green pixel is composed of all light transmitted through the green filter. It is not possible to tell "which green" frequency caused the signal on the green pixel. This information about the frequency spectrum is lost. The CCD just delivers electrons and firmware translates it to a digital value. A pure CCD can not decompose the measured intensity. The information is lost.

-
ok, but this histogram in fact is how much of every color RGB we have on the picture and not the frequency in hertz they have. –  SpaceDog Mar 14 '13 at 14:14
@RubberDuck Updated my answer. It is not possible to tell "which green" frequency $\approx 10^{15}\,$GHz caused the signal on the green pixel. See the link for the spectrum of light. It supplies frequency and wavelength at the same picture. This information is lost on a CCD chip. –  Stefan Bischof Mar 14 '13 at 16:56
ah, got it now. Thanks for your help. –  SpaceDog Mar 14 '13 at 20:20

I guess that the CCD sensor already gives you data decomposed into red, green and blue, or more colors if it's built for scientific purposes. In any case, the sensor does the decomposition itself, not you (or a computer).

-
no, this is not what I mean. Now I have the R, G and B of a pixel. Which frequency it is in hertz? –  SpaceDog Mar 14 '13 at 13:58
As soon as the sensor gives you RGB data, you cannot recover the original frequency. It might also be more than one or hundreds of different frequencies. –  Rafael Reiter Mar 14 '13 at 14:01
ah, I see. Thanks. –  SpaceDog Mar 14 '13 at 14:05