From d21a9b6df4cd3a47e540c3cca8203b8f0f53a029 Mon Sep 17 00:00:00 2001 From: Florian Pritz Date: Sun, 12 Apr 2015 10:45:32 +0200 Subject: add new scripts Signed-off-by: Florian Pritz --- ocr-scan | 20 ++++++++++++++++++++ 1 file changed, 20 insertions(+) create mode 100755 ocr-scan (limited to 'ocr-scan') diff --git a/ocr-scan b/ocr-scan new file mode 100755 index 0000000..5e7423f --- /dev/null +++ b/ocr-scan @@ -0,0 +1,20 @@ +#!/bin/bash + +jpeg_quality=70 +jpeg_size=1500x1500 + +time=$(date +%Y-%m-%d_%H%M%S) +scanimage --resolution 300 -b -x 215 -y 297 +i=1 +while [[ -e "out$i.pnm" ]]; do + img="out$i.pnm" + newname="$(printf "%s_%03d" "$time" "$i")" + unpaper "$img" "unpaper-$img" + tesseract "unpaper-$img" "$newname.txt" + convert "$img" -normalize -resize $jpeg_size -quality $jpeg_quality "$newname.jpg" && rm "$img" "unpaper-$img" + ((i++)) +done + +if [[ $1 = "--pdf" ]]; then + convert "${time}"_*.jpg -resize $jpeg_size -compress jpeg -quality $jpeg_quality -density 72x72 -repage 595x842 "${time}.pdf" && rm "${time}"_*.jpg +fi -- cgit v1.2.3-24-g4f1b