[Author Prev][Author Next][Thread Prev][Thread Next][Author Index][Thread Index]

Re: [APML] median vs. average vs. sigma



Brian,
 
    I'll make this quick since this is still a film site (anyone not wanting to read this please hit "delete" now) ...
 
   Median combining has a dirty little secret about it ... if all your frames are not EXACTLY similar, median combining will actually reject some of the frames ... after all, it's looking for anything that is not exactly similar ... if you have some frames that aren't, they get the knife. The bad thing about this ... you loose the "averaging" effect when you ditch frames and your signal to noise is worse because of it. Signal to noise is directly related to the square root of the number of frames that have been averaged, so if you start loosing frames, the noise in your image will increase ... sometimes dramatically.
 
     Averaging, on the other hand, is locked in ... EVERY frame is accounted for and all of them are combined so you end up with the smoothest image possible and the highest S/N ... this would be all you needed except for one thing ... those pesky cosmic ray hits that appear in a RANDOM pattern all over your frames ... averaging does nothing to get rid of those. You have the smoothest background, but it is littered with the cosmic ray debris.
 
     Sigma processing is the best of both worlds ... although the exact math is over my head, it essentially does a true average of the frames AND at the same time rejects any "outliers" ... it is vital to set up the parameters correctly when doing a sigma combine ... Ray Graylak's site has a good explanation of it (I use Ray's Sigma 10 for my combining). I have run test after test and the sigma combine beat the median combine every time ... if you want the maximum S/N, it's the only way to fly.
 
         Tony
_______________________________________________
Astro-Photo mailing list
Astro-Photo@seds.org
http://seds.org/mailman/listinfo/astro-photo