JMM’s image/document scanning notes

ArchWiki’s page on SANE is a good resource.

Scanning on Nix

So I’m scanning from a network-connected scanner (a Brother scanner).

The default GUI I use to scan is GNOME’s Document Scanner. But you may want to use the scanner from command line. For this, use the nixpkg sane-backends which provides scanimage.

To get a list of devices, use:

scanimage -L

And you’ll get something like:

device `v4l:/dev/video0' is a Noname Integrated Camera: Integrated C virtual device
    device `escl:http://192.168.1.XX:80' is a Brother XXXX series adf,platen scanner

Here’s how to scan to a file named test1.png:

scanimage --device "escl:http://192.168.1.XX:80" --format=png --resolution=300dpi --mode=color --output-file=test1.png --progress

Here’s how you’d save it as a WebP in one step:

scanimage --device "escl:http://192.168.1.XX:80" --format=png --resolution=300dpi --mode=color |
    ffmpeg -hide_banner -f image2pipe -c:v png -i - -f webp -quality 90 -preset text -compression 6 -y "output.webp"

Get scanner options

How to get scanner options for a device:

scanimage --device "escl:http://192.168.1.XX:80" -A

Scan an OCR’d PDF

This will output a PDF named output.pdf.

scanimage --device "escl:http://192.168.1.XX:80" --format=png --resolution=300dpi --mode=gray |
    ffmpeg -hide_banner -f image2pipe -c:v png -i - -f image2pipe -c:v mjpeg -q:v 13 - |
    tesseract - stdout --oem 1 -l eng --dpi 300 pdf >output.pdf

Here I’m using FFMpeg to convert to a JPEG to reduce the size of the output PDF. Also note that the DPI is set to 300 here. Setting it explicitly prevents tesseract from having to guess, which seems like it can cause the output PDF to be a weirder size.

Scan multiple images into one PDF

Normally you’d just use GNOME’s Document Scanner for this, but you can also use the following bash script:

imgs=()
while
    read -p "Continue (y/n)? " -n 1 -r
    echo
    [[ $REPLY =~ ^[Yy]$ ]]
do
    scanned=$(mktemp scanned-XXXXXXXX.jpg)
    scanimage --device "escl:http://192.168.1.XX:80" --format=png --resolution=300dpi --mode=gray |
	ffmpeg -hide_banner -f image2pipe -c:v png -i - -f image2pipe -c:v mjpeg -q:v 13 - >"$scanned"
    imgs+=("$scanned")
done
if (( ${#imgs[@]} )); then
    read -p "Save as (default output.pdf): " -r
    img2pdf "${imgs[@]}" -o "${REPLY:-output.pdf}" && rm "${imgs[@]}"
fi

This relies on sane-backends, ffmpeg, and img2pdf (to convert JPEGs to PDF without quality loss).

Multipage PDF with OCR

And here’s the same thing, but with OCR.

#!/usr/bin/env bash
# Interactively scan a bunch of pages into one OCR’d PDF
# 🄯2023 Josh Mōller-Mara
set -euo pipefail
pages=()
while
    read -p "Continue (y/n)? " -n 1 -r
    echo
    [[ $REPLY =~ ^[Yy]$ ]]
do
    scanned=$(mktemp scanned-XXXXXXXX.pdf)
    scanimage --device "escl:http://192.168.1.XX:80" --format=png --resolution=300dpi --mode=gray |
	ffmpeg -hide_banner -f image2pipe -c:v png -i - -f image2pipe -c:v mjpeg -q:v 13 - |
	tesseract - stdout --oem 1 -l eng --dpi 300 pdf >"$scanned"
    pages+=("$scanned")
done
if (( ${#pages[@]} )); then
    read -p "Save as (default output.pdf): " -r
    pdftk "${pages[@]}" cat output "${REPLY:-output.pdf}" && rm "${pages[@]}"
fi

Brother scanner issues

I’m currently using a Brother MFC-L2710DW to scan. Which looks like it can use either the brscan4 or brscan5 drivers. Unfortunately this is non-free software that’s just packaged as a .deb or .rpm and requires agreeing to some EULA.

Normal scanning works mostly fine without any proprietary drivers. However, I think this means that I can’t adjust brightness, since I get the following error if I try to set --brightness=100:

scanimage: attempted to set inactive option brightness

It maybe seems like historically brscan may have released source code, but no longer does? I’m not sure why they changed.

To-dos

See a blog post on compressing hand-written notes linked in my notes on OCR
Figure out how to automatically threshold and maybe crop receipts

Image/document scanning notes