0

I used to get rid of noise in my scanned photocopies using a mixture of Brightness and Contrast which worked quite well. Then I found a better method by increasing the Exposure and Black Levels which worked even better. But there are still times where I can't completely remove all of it without needing to manually erase it like in the below example:

enter image description here

Increasing the Exposure and Black Levels will still give a cleaner result but I would still have to manually remove it:

enter image description here

Is this currenty the best method possible or is there a feature in Gimp that can get rid of this type of Noise? It is likely that AI is the only solution, are there any free AI tools that I can use to do this?

0

2 Answers 2

1

That isn't technically "noise". It's a bad photocopy/scan. Noise is something that can be fixed often using noise reduction filters, but that won't work here because that isn't true digital noise like you get from a camera sensor.

Anyway . . .

There is a plugin for GIMP (and other editors) called G'MIC which contains a filter called "Repair Scanned Document", but your 1st example is probably too severe for it to fix.

I tried using it on your 2nd image, the result isn't perfect though. You could of course just paint in white over the parts that didn't quite work.

Here's the plugin dialog and settings. The easiest way to find the filter in G'MIC is to use the search feature, since it contains over 500 filters.

enter image description here

And the resulting image

enter image description hereclick to enlarge

1
1

With plain Gimp:

A bit of thresholding:

enter image description here

Followed by a median blur with a low radius:

enter image description here

1
  • thanks for sharing looks interesting I will try this! Mar 10 at 15:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.