How to process scans of printer documents to improve quality?

Question

Document flow

I decided to clean up and throw out papers for which the originals are unnecessary, having previously digitized (scanned) them.

Documents are mostly printed a la Word / Excel - all sorts of textures (respectively, text and plates of black lines on a white sheet plus a stamp with a signature (usually blue, but this is not very important)), old contracts, etc.

The problem is that after scanning, the background is not white, but partly or completely grayish. If you reduce the brightness, the quality of the lines / text deteriorates. If you keep it honest B / W, then ugly notches appear, etc.

Accordingly, the task is to make the gray background pure white (#FFFFFF in RGB), and the text / lines - clear and black (maybe even a little smooth (anti-aliasing)).

If desired, of course, I can do this manually by selecting dirt with tools like “magic wand”, but this is broken and something tells me that this task should be relatively easy to automate.

From the OS available Windows and Linux, I'm not afraid of the command line :-) but I had no experience with automatic graphics processing before. Scanning is manual (in the sense of not streaming), preferably JPEG or TIFF formats.

Let me know if anyone knows the tool.

UPDATE: Here's an example - imgur.com/emxvgKc
- this is the best we could get from a scanner (HP ScanJet 2400). Pay attention to the gray ruler on a white background, in the original, of course, it was not there, this is the scanner carriage driving. In principle, if you just remove it, it's already good.

Answer 1

There is an article: www.djvu-soft.narod.ru/scan/twdragon_djvu.htm
In particular, a section about the scankromsator program

Answer 2

If it's serious and normal, then OCR
And if not seriously, then you just need to edit the gamma, cutting off the light part to the level of the gray background and stretching the rest to the stop.
imagemagick can do this for non-boiling consoles and scripts.
www.imagemagick.org/script/command-line-options.php
If the console is still very scary or the amount of work is not large, then there is a levels color filter in gimp (levels in the Russian locale) - very clearly and simply do exactly what is needed.

Answer 3

Yes, at least a 3x3 median filter will help (unless, of course, the scan resolution is not very small).

Answer 4

Gimp + bimp . Then either a Gaussian smear or a ticketer filter.

Answer 5

ScanKromsator is what you need.