Gimp: how to remove background noise/artifacts and enhance handwritten text


I have a lots of scanned handwritten lecture notes (PDFormat) and I want to improve their readability (I have uploaded an example). In particular I'd like to remove the "squares" on the back. I would appreciate any advice you may have. Thanks in advance.


4/19/2014 5:15:00 PM

The scanning we have as an example shows many artifacts which have only little difference in color or brightness. In addition there are heavy JPEG artifacts. Therefore repair tools will have a hard time to separate artifacts from important parts.

For the removal of lines we can try with G'Mic > Repair > Unstrip filter of the Grey's Magic Tools plugin which makes use of Fourier's transformation for artifact removal. But we can see whatever setting we chose some lines will still be there when the handwriting already degrades:

enter image description here

This will become more obvious after we had performed some despeckling (here using the Wavelet-Denoise plugin), and adjustments of contrast:

enter image description here

In case we had many such documents to process all this will need a lot of manual work in addition, and then it will be a matter of days or weeks rather than hours with a sadly doubtful outcome.

Some artifact removal can be achieved, and we may get some better results than the quick examples above with optimal settings or after leaving more of the background noise in the image but from the example here this still may not be quite what we hoped, unfortunately.

We may have more luck with scans where the background is lighter or with colored scans where writing and background may have a different color. In any case, repair is much easier with a source not having so many JPEG artifacts.

4/19/2014 8:04:00 PM