ScanSnap iX500 on Omarchy / Arch Linux

Setup guide and batch scanning script for the discontinued Fujitsu ScanSnap iX500 on Omarchy (Arch Linux-based) systems, using SANE.

The repo contains one script, scansnap-scan: load the ADF, run it, get a single deskewed PDF, optionally OCR'd and optionally uploaded to Paperless-ngx.

---

Overview

The Fujitsu ScanSnap iX500 is discontinued but an excellent duplex document scanner. It works on Linux through SANE's built-in fujitsu backend, with no proprietary drivers and no firmware files.

What works:

• USB scanning (WiFi is not supported on Linux)
• Duplex (double-sided) ADF scanning, whole stack in one pass
• 50-600 DPI
• Lineart, Grayscale and Color modes

USB IDs:

• iX500: 04c5:132b
• iX500EE (Enterprise Edition): 04c5:13f3

USB IDs are hardware identifiers burned into the device firmware. They never change and uniquely identify the scanner model.

---

Install

# Core: scanning + PDF assembly + image processing
yay -S sane img2pdf imagemagick

# Optional: GUI frontend
yay -S simple-scan

# Optional: OCR (searchable PDFs). ocrmypdf is AUR; pick your language packs.
yay -S ocrmypdf tesseract-data-eng tesseract-data-nld

# Optional: only needed for --paperless
yay -S jq 1password-cli

# Optional: full-text search across scanned PDFs
yay -S pdfgrep

What each package is for:

Package	Why
sane	Core libraries and the fujitsu backend. Arch has no separate sane-backends package; the backends are in here.
img2pdf	Lossless image-to-PDF assembly
imagemagick	Automatic deskew and --trim (provides magick, ImageMagick 7)
ocrmypdf	--ocr, adds a searchable text layer
jq, curl	--paperless metadata resolution and upload
simple-scan	GUI alternative, has its own batch support

Run yay -Ss tesseract-data for the full list of 128+ OCR language packs.

Then install the script:

git clone https://github.com/peterberkenbosch/omarchy-scansnap.git
install -Dm755 omarchy-scansnap/scansnap-scan ~/.local/bin/scansnap-scan

Make sure ~/.local/bin is on your PATH.

---

Find your scanner

scanimage -L

device `fujitsu:ScanSnap iX500:1566072' is a FUJITSU ScanSnap iX500 scanner

That trailing number is your unit's serial. The script's built-in default carries the author's serial, so export your own:

# ~/.bashrc, ~/.zshrc, or wherever you keep shell config
export SCANSNAP_DEVICE="fujitsu:ScanSnap iX500:YOURSERIAL"

Other detection routes:

sane-find-scanner
lsusb | grep -i fujitsu
# Bus 001 Device 006: ID 04c5:132b Fujitsu, Ltd ScanSnap iX500

Permissions

Usually nothing to do. systemd's 70-uaccess.rules grants the logged-in seat an ACL on tagged USB devices, and the sane package tags the iX500 for exactly that. Verify after installing sane:

grep -A 2 "ScanSnap iX500" /usr/lib/udev/hwdb.d/20-sane.hwdb

# Fujitsu ScanSnap iX500
usb:v04C5p132B*
 libsane_matched=yes

libsane_matched=yes is what earns the device its uaccess tag. If you still get permission errors (non-systemd setup, or scanning from a non-seat session such as a cron job), fall back to group membership:

sudo usermod -a -G scanner $USER
# log out and back in

---

Usage

# Duplex color 300dpi, all pages in the ADF into one PDF
scansnap-scan

# Single-sided grayscale at 200dpi
scansnap-scan --simplex --gray --dpi 200

# Searchable PDF
scansnap-scan --ocr

# Receipt or small document: crop the surrounding whitespace
scansnap-scan --simplex --gray --trim

# Somewhere other than the default
scansnap-scan --output ~/Desktop

Options

Option	Effect
--bw, --lineart	Black & white
--gray, --grayscale	Grayscale
--color	Color (default)
--dpi N	Resolution, 50-600 (default 300)
--simplex	Single-sided (ADF Front)
--duplex	Double-sided (default)
--ocr	OCR with ocrmypdf, Dutch + English, PDF/A output
--trim	Crop whitespace around each page
--output, -o DIR	Output directory
--paperless	Upload to Paperless-ngx after scanning
--title, --correspondent, --type, --tags, --created	Paperless metadata hints
--keep-local	Keep the local PDF after a successful upload
--help, -h	Full help

Environment

Variable	Meaning
SCANSNAP_DEVICE	Device string from scanimage -L. Set this.
SCANSNAP_OUTPUT_DIR	Default output directory (default ~/Documents/Scans)
PAPERLESS_URL	Paperless-ngx instance URL, required for --paperless
PAPERLESS_TOKEN	Paperless API token
PAPERLESS_OP_ITEM	1Password secret reference to read the token from

---

Automatic deskew

Every page is deskewed automatically. There is no flag to enable it and no flag to skip it.

The script does not use the fujitsu driver's --swdeskew. That option mislocks on security-print patterns (the tint blocks printed on bank and government letters) and has been observed rotating an upright page by roughly 30 degrees.

Instead the script runs its own projection-profile deskew:

1. Downscale the page, crop to the paper interior (70%), threshold to bilevel. Cropping first keeps scanner background and paper edges out of the measurement.
2. Score candidate angles from -3° to +3° in 0.25° steps. The score is the standard deviation of the row-brightness profile: text rows line up into sharp bands when the page is straight, so the profile's variance peaks at the correct angle.
3. Refine around the winner in 0.05° steps.
4. If the best angle sits at the search boundary, detection is treated as failed and the page is left alone.
5. Corrections under 0.05° are skipped so already-straight pages are never resampled.

The ±3° clamp is the safety property: a correction can never exceed 3°, so a misdetection costs you a barely-tilted page rather than a ruined one.

---

Paperless-ngx upload

With --paperless, the finished PDF is posted to a Paperless-ngx instance and the local copy is deleted (pass --keep-local to keep it).

export PAPERLESS_URL="https://paperless.example.com"
export PAPERLESS_TOKEN="your-api-token"

scansnap-scan --ocr --paperless \
  --correspondent "Belastingdienst" \
  --type "Beschikking" \
  --tags "business,fy-2026" \
  --title "Voorlopige aanslag 2026"

Instead of putting the token in your shell config, store it in 1Password and let the script read it:

export PAPERLESS_OP_ITEM="op://Private/Paperless API/credential"

If PAPERLESS_TOKEN is unset, the script calls op read "$PAPERLESS_OP_ITEM".

Correspondents, document types and tags are passed by name. The script resolves each name to its Paperless ID and creates it if it doesn't exist yet (with matching_algorithm: 6, "auto"). Paperless then runs its own OCR, classification and workflow rules on the uploaded file, so --ocr is optional here; it is useful if you also want a searchable copy outside Paperless.

The upload prints the task ID and a dashboard URL to follow processing.

---

Getting your token

In Paperless-ngx: profile menu → My Profile → API Auth Token. Copy it into PAPERLESS_TOKEN or your 1Password item.

---

Searching scanned PDFs

Once --ocr has run, the PDFs carry a text layer:

pdfgrep "Belastingdienst" ~/Documents/Scans/*.pdf
pdfgrep -C 3 "factuur" ~/Documents/Scans/*.pdf     # with context
pdfgrep -r -i "belasting" ~/Documents/Scans/       # recursive, case-insensitive
pdfgrep -l "2026" ~/Documents/Scans/*.pdf          # filenames only

---

Manual scanning

The script is a convenience wrapper. The underlying commands, if you want them:

# Inspect every option the backend exposes
scanimage --device "$SCANSNAP_DEVICE" -A

# Single page
scanimage --device "$SCANSNAP_DEVICE" \
  --source "ADF Duplex" --mode Color --resolution 300 \
  --format=png --output-file page.png --progress

# Whole stack in one session, then assemble
scanimage --device "$SCANSNAP_DEVICE" \
  --source "ADF Duplex" --mode Color --resolution 300 \
  --format=png --batch="page_%03d.png" --batch-print --progress
img2pdf page_*.png -o document.pdf && rm page_*.png

Useful backend options: --page-width / --page-height, --brightness -127..127, --contrast -127..127.

---

Desktop entry

~/.local/share/applications/scansnap-ix500.desktop:

[Desktop Entry]
Name=ScanSnap iX500
Comment=Scan documents with Fujitsu ScanSnap iX500
Exec=simple-scan
Icon=scanner
Type=Application
Categories=Office;Scanning;
Keywords=scan;scanner;document;scansnap;

update-desktop-database ~/.local/share/applications/

---

Troubleshooting

Only the front side of each duplex sheet comes out

You are running an old version of this script. Versions before August 2026 scanned with --output-file in a per-page loop, and scanimage performs exactly one acquisition per invocation — for a duplex sheet that is the front side only, and the back was discarded when the process exited. The fix is --batch, which holds a single SANE session open until the feeder empties. Pull the latest scansnap-scan.

"Document feeder out of documents"

No paper in the ADF. This is also the normal, expected message at the end of a run.

A page came out rotated by a wild amount

Not this script — it clamps corrections to ±3°. If you added --swdeskew to the scanimage call yourself, remove it. See Automatic deskew.

"Invalid argument"

Wrong device string. Take the exact value from scanimage -L and set SCANSNAP_DEVICE.

Permission denied

Check that sane is installed and the device carries libsane_matched=yes (see Permissions). If the session isn't a seat session, uaccess won't apply and you need the group:

sudo usermod -a -G scanner $USER
# log out and back in

Scanner not detected

1. Check the USB cable and that the scanner's LED is on.
2. lsusb | grep -i fujitsu
3. sudo udevadm control --reload-rules && sudo udevadm trigger
4. Unplug and replug.

convert: command not found

ImageMagick 7 dropped the convert name. This script uses magick. Install imagemagick.

Slow scanning on USB 3.0

Older scanners can struggle on USB 3.0 ports:

export SANE_USB_WORKAROUND=1

Or use a USB 2.0 port.

---

Choosing settings

Use case	Resolution
Email / archive	150-200 DPI
OCR	300 DPI
High quality	400-600 DPI

• Lineart: text documents and forms, smallest files
• Grayscale: documents with photos, and a good default for pages with signatures
• Color: full-color documents and photographs, largest files

---

Filing scans without Paperless

If you don't run Paperless-ngx, a pdfgrep rule file gets you most of the way. Put the rules in ~/.config/scansnap/rules.conf:

# PATTERN -> DESTINATION
Belastingdienst          -> ~/Documents/Filed/Tax
"ING Bank"               -> ~/Documents/Filed/Banking/ING
Factuur                  -> ~/Documents/Filed/Invoices
Invoice                  -> ~/Documents/Filed/Invoices
Contract                 -> ~/Documents/Filed/Contracts

And a companion script, ~/.local/bin/scansnap-organize:

#!/bin/bash
# Move OCR'd PDFs into folders based on their content.
set -uo pipefail

RULES_FILE="${HOME}/.config/scansnap/rules.conf"
SCAN_DIR="${SCANSNAP_OUTPUT_DIR:-${HOME}/Documents/Scans}"

command -v pdfgrep >/dev/null || { echo "Install pdfgrep: yay -S pdfgrep"; exit 1; }

for pdf in "$SCAN_DIR"/*.pdf; do
  [ -f "$pdf" ] || continue
  matched=false

  while IFS= read -r line; do
    [[ "$line" =~ ^[[:space:]]*# ]] && continue
    [[ -z "${line// }" ]] && continue
    [[ "$line" != *"->"* ]] && continue

    pattern=$(echo "${line%%->*}" | xargs)
    destination=$(echo "${line##*->}" | xargs)
    destination="${destination/#\~/$HOME}"

    if pdfgrep -q -i "$pattern" "$pdf" 2>/dev/null; then
      mkdir -p "$destination"
      mv "$pdf" "$destination/"
      echo "$(basename "$pdf"): matched '$pattern' -> $destination"
      matched=true
      break
    fi
  done < "$RULES_FILE"

  [ "$matched" = false ] && echo "$(basename "$pdf"): no match, left in $SCAN_DIR"
done

chmod +x ~/.local/bin/scansnap-organize
scansnap-scan --ocr && scansnap-organize

This needs --ocr, since matching reads the PDF's text layer. Paperless-ngx does the same job with proper classification and is the better answer if you're willing to run a server.

---

Reference

Path	Purpose
/usr/lib/udev/hwdb.d/20-sane.hwdb	Hardware database, contains the iX500 entry
/usr/lib/udev/rules.d/65-sane.rules	udev rules for scanners
/etc/sane.d/fujitsu.conf	Fujitsu backend configuration
/etc/sane.d/dll.conf	Enabled SANE backends
~/Documents/Scans/	Default output directory
~/.local/bin/scansnap-scan	The script

Further reading:

• SANE project and its supported devices list
• Arch Wiki: SANE
• Paperless-ngx documentation

---

Verified against Omarchy 4.0.0, ImageMagick 7.1.2, August 2026. Scanner: Fujitsu ScanSnap iX500 / iX500EE (USB IDs 04c5:132b / 04c5:13f3).