The first part of this process involves two steps:
- Calibration - each frame contains not only the "signal" from the object but also "noise" originated in the way the camera sensor converts the signal into data. Furthermore, for various reasons, the signal is not "measured by the same yardstick" in all areas of the sensor. This process is intended to correct these two effects in order to reconstruct the signal.
- Combination - as the calibration is not perfect it does not completely eliminate noise, but the signal to noise ratio can be improved substantially combining many frames, resulting in a combined image much less noisy than the individual frames.
I have made video tutorials of K3CCDTools and DeepSkyStacker (in spanish) I hope them to be useful for beginners.