Comment on 💞 FairScan > Syncthing > Paperlees-ngx

• llii@discuss.tchncs.de ⁨3⁩ ⁨weeks⁩ ago

  Its a hacked together shell script. I wanted to learn shell scripting, but I should’ve probably programmed it in python.

  But it basically does this:

  1. scan pages as png via scanimage in gray scale
  2. convert the image into a high contrast b/w one with image magick
  3. sort pages if I have to combine two scans (front and back side)
  4. create a single pdf of all the document pages via image magick
  5. move pdf to the paperless folder

  I won’t share it because it’s really ugly and some of the functionality can be done with paperless native, like combining the pages of two scans. Which would’ve been easier.

  source

  • RazzleDazzle@discuss.tchncs.de ⁨2⁩ ⁨weeks⁩ ago

    Thanks! And that’s ok, I’m more interested in the steps than the code itself. Do you apply any optimizations with image magik in step 2?

    source

    • llii@discuss.tchncs.de ⁨2⁩ ⁨weeks⁩ ago

      Yeah, I do this:

      magick “$file” -limit memory 1GiB -limit map 2GiB -threshold 60% -define connected-components:area-threshold=5 -define connected-components:mean-color=true -connected-components 8 output.png

      It makes the scan 1-bit black and white. It mostly removes bright background images, if there are any on the pages. I think it works quite well for basic black and white documents, but color graphics and graphs aren’t preserved.

      source