Document Scanning incl OCR - Pastebin.com

1 min read Original article ↗

Guest User

a guest

Aug 19th, 2013

380

0

Never

Not a member of Pastebin yet? Sign Up, it unlocks many cool features!

  1. #!/bin/bash

  3. set -x

  4. set -e

  6. target="${1}.pdf"

  7. num_pages=${num_pages:-1}

  9. if [[ -z "$target" ]]; then

  10. echo >&2 "Usage: $0 DESTINATION"

  11. exit 1

  12. fi

  14. echo "Scanning $num_pages page(s) to $target..."

  16. tempdir=$(mktemp -d)

  17. scansource=$(LC_ALL=C scanimage -L \

  18. | grep 'HP 7650 Document scanner' \

  19. | sed "s,^device \`\([^']\+\)'.*,\1,")

  21. (

  22. cd "$tempdir"

  23.     scanimage -d"$scansource" -B64 --format=tiff -l0 -t0 -x215 -y297 \

  24. --batch-start=0 \

  25. --batch-count="$num_pages" \

  26. --batch-increment=1 \

  27. -b \

  28. --mode Color \

  29. --resolution 300 \

  30. --source ADF

  31. )

  33. convert="convert ${tempdir}/out0.tif"

  34. convert "${tempdir}/out0.tif" "${tempdir}/out0.pdf"

  35. gs="gs \

  36.    -dBATCH \

  37.    -q \

  38.    -dNOPAUSE \

  39.    -sDEVICE=pdfwrite \

  40.    -sOutputFile=- \

  41. ${tempdir}/out0.pdf"

  43. for ((i=1; i < num_pages; ++i)) do

  44.     convert "${tempdir}/out${i}.tif" "${tempdir}/out${i}.pdf"

  45. convert="$convert -append ${tempdir}/out${i}.tif"

  46. gs="$gs ${tempdir}/out${i}.pdf"

  47. done

  49. $convert "${tempdir}/scanimage.pnm"

  51. gocr -i "${tempdir}/scanimage.pnm" -o "${tempdir}/${1}.txt" -f UTF8

  52. a2ps -Xutf-8 -i "${tempdir}/${1}.txt" -o "${tempdir}/${1}.ps"

  53. $gs "${tempdir}/${1}.ps" > "$target"

  55. ls -lh "$tempdir"

  56. ls -lh "$target"

  57. rm -r "$tempdir"