What modern methods exist for denoising and deblurring images? Yes, I haven’t forgotten how to Google and spent a lot of time searching for the code of a neural network that would not have been created 5 years ago, but I didn’t come up with much. The idea is to create a modern neural network to remove noise and blur from astronomical images. There are already a couple of options on the market, but they are paid and not available to the average user, or not trained on astronomical data.
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
1
There was a Helsinki Deblur Challenge 2021, where the goal was to de-blur images of ASCII text. Maybe the submitted networks are suitable also for astronomical pictures. At least the winner used U-net architecture, as you might guess. Note that here the problem was gray scale, but I guess you can transform the image into (for example) YCbCr color space, de-blur the Y channel and just apply noise filtering to the color channels.
-
$\begingroup$ Very interesting information, thanks! I assume that it is enough to divide the RGB into three mono layers and get an average PSF, which can already be applied to each layer and end up with a good result. $\endgroup$ Commented Jun 17 at 17:53