Image processing is the act of applying various corrections to the "raw" images coming from the CCD camera, so that they can be used for photometry and other quantitative measurements.
Why two cookbooks? Well, we have noticed that the median level of a bias frame (typically near 975 counts) can vary by 0-5 counts over the course of several hours. If you are working on frames for the RR Lyrae star program (V and I filters, exposure times of 10 sec or more), this won't be a problem. They sky will have several hundred counts in it, and since we are measuring photometry differences between pairs of stars, any error incurred by subtracting off the wrong bias level will cancel out.
However, if you are processing images where the sky level will be close to zero, and precise sky subtraction over larger areas is critical, you may want to get the best estimate possible for the bias level by using a bias image taken soon after your star (object) image. This may occur if you are using the U filter (very low throughput), taking very short exposures, and/or using very narrow filters (e.g., for cometary comae). Otherwise, you could face the problematic situation where you get a negative sky value!
Updated 23 Feb 2000