Question? on Processing SubexposuresPosted By Enrico Africa |
Hi Tyler,
The impression I have from the experts here and in other groups is that you almost never add your subexposures. The reason is that when you add, you are summing not only signal but noise as well.
Average is the "best" as far as increasing your final image's signal-to-noise ratio (SNR). If you have a lot of extraneous noise in your image, however (e.g., cosmic ray hits, or a plane in the FOV that does not wreck the image), a median combine does well to eliminate those outliers, but at a cost of a lower SNR.
A lot of the experts recommend the sigma combine method (one handy tool is Ray Gralak's Sigma Reject - a Google search will turn that up), which I am just trying now myself.
For this to be most effective, however, you will need to:
1. Dither your exposures. This means you deliberately move your scope around a little bit - a few pixels - between each subexposure. The reason for this is so that different areas image different parts of the CCD (IOW, so that the bad pixels in your camera are not reinforced when you combine).
2. Take a lot of subexposures. The minimum for a sigma reject to be effective is ten.
I did see a recent thread that stated when an "Add" is used: if you have two sets of images exposed for different times (e.g., one set at 5 minute subexposures, another for 10 minutes). The one set (5 minutes, in this example) will be average, sigma or whatever combined, the other set also combined, then the two combined images summed.
HTH,
Eric
The impression I have from the experts here and in other groups is that you almost never add your subexposures. The reason is that when you add, you are summing not only signal but noise as well.
Average is the "best" as far as increasing your final image's signal-to-noise ratio (SNR). If you have a lot of extraneous noise in your image, however (e.g., cosmic ray hits, or a plane in the FOV that does not wreck the image), a median combine does well to eliminate those outliers, but at a cost of a lower SNR.
A lot of the experts recommend the sigma combine method (one handy tool is Ray Gralak's Sigma Reject - a Google search will turn that up), which I am just trying now myself.
For this to be most effective, however, you will need to:
1. Dither your exposures. This means you deliberately move your scope around a little bit - a few pixels - between each subexposure. The reason for this is so that different areas image different parts of the CCD (IOW, so that the bad pixels in your camera are not reinforced when you combine).
2. Take a lot of subexposures. The minimum for a sigma reject to be effective is ten.
I did see a recent thread that stated when an "Add" is used: if you have two sets of images exposed for different times (e.g., one set at 5 minute subexposures, another for 10 minutes). The one set (5 minutes, in this example) will be average, sigma or whatever combined, the other set also combined, then the two combined images summed.
HTH,
Eric
