The raw field image is that taken directly by the CCD of the object of interest without any additional processing

A dark field image is taken with the light collector/CCD covered; therefore any signal recorded by the CCD is due to "aark current."  The "dark current" produced on the CCD is directly proportional to the exposure time (nearly) but decreases as the CCD temperature decreases.  Dark current results because electrons in the high speed tail of the Maxwell Boltzmann distribution have some finite probability of being captured by a pixel element.  Although dark current is consistently reproducible (for a fixed temperature and exposure time), there is still noise in the dark current that can overwhelm a weak true signal.

A flat field image is is taken with the telescope/CCD pointed at a uniformly bright field of view (e.g., a uniformly illuminated screen, a cloudy sky, a blue sky).  The flat field image not only contains dark current appropriate to its exposure time, but also includes effects due to pixel-dependent quantum efficiency (or sensitivity to photons).  However, because flat field images are usually taken for times so short (less than 5 seconds) that dark current contributions can be ignored.  Since a pixel's quantum efficiency may be wavelength dependent, separate flat field images should be taken for all filters used in taking raw images.

A bias field image is taken for the shortest possible exposure time in total darkness.  It measures any "pre-charge" that resides on a CCD's pixels.  Any bias precharge is independent of temperature and exposure time

Therefore, to obtain a calibrated image (CI) from the raw image (RI), the dark field image (DI), the flat field image (FI) and the bias field image (BI), the quantity

must be calculated for each pixel element in the CCD array.