OCR

SeedDMS itself has no support for optical character recognition (OCR) because it does not care about the content of file. Though, external OCR software can be used to convert an image into text and index it by the full text search engine. From SeedDMS point of view, it would be sufficient to have a conversion service which converts an image into text. This can be implemented in any possible way, but most likely as a SeedDMS extension.

The following script can be use to convert a pdf with scanned images into a text. The script converts any page into a image, runs it through tesseract, which creates a pdf again containing a text layer. All those pdf documents will be united into a single pdf and through pdftotext again. It was published in the SeedDMS forum https://sourceforge.net/p/seeddms/discussion/general/thread/4ec5973d/

#!/bin/bash
inputpdf=$1
temp_folder=/tmp/seedinput/$(date +"%Y_%m_%d_%H%M%S")/
lockfile=/tmp/seed
protokolldatei=./tesser_syslog
cores=2

mkdir -p $lockfile

while [ -e "$lockfile"/"`basename $0`" ];
do
    sleep 5
done

if ( set -o noclobber; echo "locked" > "$lockfile"/"`basename $0`"); then
  trap 'rm -f "$lockfile"/"`basename $0`";  echo $(date) " Lock file will be deleted: " $lockfile"/"`basename $0` Aufrufparameter: $* >> $protokolldatei ;rm -r $temp_folder;  exit $?'  INT TERM KILL EXIT
  # write date and script name into log file
  echo $(date) " Lock file created: " $lockfile"/"`basename $0` >> $protokolldatei
else
  # Exit script if lock file could not be created
  echo $(date) " Script will exit, because lock file could not be created:  $lockfile"/"`basename $0` Aufrufparameter: $* " >> $protokolldatei
  exit 1
fi

mkdir -p $temp_folder

$(pdftotext -raw $1 - 1> $temp_folder''tmp.txt )
pdf_contents=`cat $temp_folder''tmp.txt`
pdf_contents=`echo "$pdf_contents" | tr -dc '[:print:]'`
if [ -z "$pdf_contents" ]; then
  convert -density 300 -quality 95 $inputpdf +adjoin $temp_folder''image%03d.jpg
  find $temp_folder -name '*.jpg'| parallel --gnu -j $cores tesseract -l deu --psm 6 {} {} pdf

  num=`find $temp_folder -name '*.pdf'| wc -l`
  if [ "$num" -gt "1" ]; then
    pdfunite $temp_folder*.pdf $temp_folder''tmp.pdf
  else
    mv $temp_folder*.pdf $temp_folder''tmp.pdf
  fi
  pdftotext $temp_folder''tmp.pdf $temp_folder''tmp.txt
  mv $temp_folder''tmp.pdf $1
fi
cat $temp_folder''tmp.txt

2.3 KiB Raw Blame History

OCR

2.3 KiB

Raw Blame History