Since astronomical CCD imaging is performed at extremely low light levels, noise (or uncertainty) in the data is usually a dominant factor in the resulting image. This is because there will always be areas of the image that have little signal and, therefore, small values of SNR (signal-to-noise ratio). Therefore a thorough understanding of this noise is needed in order to properly minimize its effects. This post deals with one aspect of this uncertainty termed readout noise.
Readout or read noise or, sometimes, electronic noise of a CCD is usually quoted in terms of electrons per pixel that is added to the final signal upon readout of the device. The average (one standard deviation) level of this additive uncertainty is the read noise. (For my CCD camera the specified readout noise is 15 electrons per pixel although measurements suggest the readout noise for my setup is closer to about 20 electrons per pixel.) On-chip binning (2×2) or (3×3) that creates a super-pixel out of 4 or 9 unbinned pixels has improved noise characteristics due to having only one read operation per superpixel and therefore, only one readout noise hit. For a more thorough treatment of readout noise, click on the link below:
http://users.libero.it/mnico/glossary/readnoise.htm
Readout noise represents the absolute minimum noise that will always be injected into a resulting image. This source of noise is independent of both the length of exposure and the brightness of the light source. Most often, however, it is not the dominant source of noise. A second source of uncertainty called shot noise is a result of the quantum uncertainty of the light (photons) that strike the CCD. This level of noise can usually be made to overwhelm or swamp the readout noise by taking exposures of a sufficient length. Shot noise will be dealt with in a later post.