X-Git-Url: https://git.ucc.asn.au/?a=blobdiff_plain;ds=sidebyside;f=scan2pages.sh;h=42b3bfbf8a6d091394e881876c35f2f98767abe1;hb=aeb663e7da50cdad5b60dcb5d7a1f78fc83f36b4;hp=501fbe3b150e498e003d8e0c8f1f7832adce17ed;hpb=2a745fdb131ca5159d1ec9ad94ee532f2cd0f1bc;p=dja%2Fscandal.git diff --git a/scan2pages.sh b/scan2pages.sh index 501fbe3..42b3bfb 100755 --- a/scan2pages.sh +++ b/scan2pages.sh @@ -125,7 +125,19 @@ for scanpgnum in `$my_seq 1 $pages`; do $scanpg || exit 1 fi; - # preprocess it! +done; + +# do ocr binarise +[ -e $outdir/scanpgs ] && rm -r $outdir/scanpgs +ocropus book2pages $outdir/scanpgs $outdir/scanpg-*.png || exit 1 + +for scanpgnum in `$my_seq 1 $pages`; do + + scanpgnum=$(printf '%03d' $scanpgnum) + scanpg=$outdir/scanpg-${scanpgnum}.png + binscanpg=$outdir/scanpgs/$(printf '%04d' $scanpgnum).bin.png + + # preprocess scanned page cleanscanpg=$outdir/scanpg-clean-${scanpgnum}.pnm if [ ! -e $cleanscanpg ]; then # create mask: @@ -137,9 +149,7 @@ for scanpgnum in `$my_seq 1 $pages`; do cropcords=$(convert -border 1x1 -bordercolor '#000' -resize 1000% -trim -fuzz 90% -format "%wx%h%O" $outdir/scanpg-mask-${scanpgnum}.png info: || exit 1) # ... crop and despeckle? the final pre-prepared image - convert $convertflags -crop $cropcords $scanpg $cleanscanpg || exit 1 - elif [[ $skipmask ]]; then - cp $origfile $preppnm + convert $convertflags -crop $cropcords $binscanpg $cleanscanpg || exit 1 fi; # check it hasn't mostly disappeared - e.g. if the scan was all black @@ -182,7 +192,7 @@ for scanpgnum in `$my_seq 1 $pages`; do fi unpaper $unpaperflags --pre-rotate 90 --layout double --overwrite \ -op 2 --no-blackfilter --no-grayfilter --no-noisefilter \ - -no-blurfilter $cleanphyspg $outdir/logpg-${scanpgnum}-${physpgnum}-%01d.pnm || exit 1 + --no-blurfilter $cleanphyspg $outdir/logpg-${scanpgnum}-${physpgnum}-%01d.pnm || exit 1 else cp $cleanphyspg $outdir/logpg-${scanpgnum}-${physpgnum}-1.pnm fi;