Sharpening Lunar/Planetary Images: Use More Frames

27 Apr 2014.  I recently got a little insight into how the image sharpening in Registax works, and it has implications for how images should be acquired and processed.  I had always assumed the idea was to sort through many planetary/lunar image frames and pick out and use only the sharpest.  That is part of the process, but not all.  Michael Covington, the author of a very nice set of astrophotography web pages and of a book on DSLR astrophotography, described a few days ago on the Deep Sky Stacker Yahoo Group how combining a lot of images that have random noise produces an image that is Gaussian blurred.  Gaussian blurring can be reversed mathematically.  Suddenly I understood why stacked images from Registax always look blurry, usually worse than the individual frames, until sharpening (which reverses that Gaussian blurring) is applied.  So I think this means I should be a bit more liberal about how many frames I stack.  Here’s an example.  In advance, I’ll admit I’m cheating a little bit, because the final sharpened images have been enhanced some in addition to using wavelet sharpening in Registax, but this illustrates the idea. 

This is the stacked image, prior to doing any sharpening.  It definitely looks worse than many of the individual frames.

This is the same data, with 50 frames per alignpoint stacked.

Finally, this is the same data,  but with 500 frames per alignpoint stacked.  I think it is just as sharp as the more selective stack, but is definitely smoother and less noisy.