lylejk wrote:
That's pretty much the highpass filter idea Kevin. Rob's idea does appear to hold promise. Still have to deal with the occasion hot bits though. Guess there isn't really a good way to test smooth voids.
![Smile :)](./images/smilies/icon_smile.gif)
That is because from an image perspective, there is no difference between a smooth void with CCD noise and the actual image.
One suggestion is to run the built in despeckle filter on the image (good for removing CCD noise), separate that to LAB then edge detect the L layer to get a good mask:
Attachment:
x.png [ 389.31 KiB | Viewed 841 times ]
You can use that mask to resharpen the despeckeled image, or blend in the original, or whatever.
(I messed up and let the image wrap instead of smear, hence the stuff on the left edge)
-Rob A>