X-Git-Url: https://git.ucc.asn.au/?p=dja%2Fscandal.git;a=blobdiff_plain;f=scan2pages.sh;h=ccb464bf60d865bbdc497575c121abc92c6a4f4d;hp=9c08f6a0552a13c08328dee1419f846b67fbdb19;hb=8f4ebaf877ac3e1113b60cbdad00dbeb3c7bcaf7;hpb=17f116bd1a594a9c8c222757169360068e443ca4 diff --git a/scan2pages.sh b/scan2pages.sh index 9c08f6a..ccb464b 100755 --- a/scan2pages.sh +++ b/scan2pages.sh @@ -7,7 +7,7 @@ Convert pdffile - a pdf of scanned facing pages - to a set of images in outdir, each with only one page. OPTIONS: -v: Be verbose. - -d depth: Use given depth. Default is 1 for black and white, 8 for colour. + -d depth: Use given depth. Default is 1 for black and white, 8 for colour/gray. Unpaper can only handle up to 8. -D dpi: Use given DPI. Colour scans on library scanners are usually 300, bw 400. -F extension: use the output format signified by the extension. Default is png @@ -15,7 +15,7 @@ OPTIONS: -f: Forceably redo everything. -p: Forceably redo pdf conversion. Implies options below, equivalent to -f. -m: Forceably redo masking/trimming and other preprocessing. Impiles options below. - -u: Forceably redo unpaper processing. Implies option below. IGNORED + -u: Forceably redo unpaper processing. Implies option below. -t: Forceably redo final trimming and cleaning. IGNORED -b: Optimize for BeBook One. __EOF__ @@ -131,27 +131,27 @@ for pg in `$my_seq 1 $pages`; do pgn=$(printf '%03d' $pg) # convert from pdf - origpnm=$dir/pg-${pgn}.pnm - if [[ ! $([ -e $origpnm ]) || $forcepdf ]]; then + origfile=$dir/pg-${pgn}.png + if [ ! -e $origfile ] || [[ $forcepdf ]]; then convert $convertflags -depth $depth -density $dpi $file[$(expr $pg - 1)] \ - $origpnm || exit 1 + $origfile || exit 1 fi; # preprocess it! preppnm=$dir/pg-pp-${pgn}.pnm - if [[ ! $skipmask && ( ! $([ -e $preppnm ]) || $forcemask ) ]]; then + if [[ ! $skipmask ]] && ( [ ! -e $preppnm ] || [[ $forcemask ]] ); then # create mask: # ... downscale, blur, - convert $convertflags -resize 10% -depth 8 -blur 10 -median 2 $origpnm $dir/pg-mask-${pgn}.pnm ||exit 1 + convert $convertflags -resize 10% -depth 8 -blur 10 -median 2 $origfile $dir/pg-mask-${pgn}.png ||exit 1 # ... get crop co-ords. They're off by ~2 as I don't know how to # properly correct for the border. - cropcords=$(convert -border 1x1 -bordercolor '#000' -resize 1000% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.pnm info: || exit 1) + cropcords=$(convert -border 1x1 -bordercolor '#000' -resize 1000% -trim -fuzz 90% -format "%wx%h%O" $dir/pg-mask-${pgn}.png info: || exit 1) # ... crop and despeckle? the final pre-prepared image - convert $convertflags -crop $cropcords $origpnm $preppnm || exit 1 + convert $convertflags -crop $cropcords $origfile $preppnm || exit 1 elif [[ $skipmask ]]; then - cp $origpnm $preppnm + cp $origfile $preppnm fi; # check it hasn't mostly disappeared - e.g. if the scan was all black @@ -163,33 +163,49 @@ for pg in `$my_seq 1 $pages`; do #unpaper it #names go a bit funny here - #also, ignore flags starting here - unppnm=$dir/upg-${pgn}-%01d.pnm - unpaper $unpaperflags --layout double --overwrite --no-blackfilter -ni 10 -op 2 $preppnm $unppnm || exit 1 - - #detect if the page is 2-up + unppnmbase=$dir/upg-${pgn} + if [ ! -e ${unppnmbase}-1.pnm ] || [ ! -e ${unppnmbase}-2.pnm ] || [[ $forceunpaper ]]; then + unpaper $unpaperflags --layout double --overwrite -ni 10 -op 2 $preppnm ${unppnmbase}-%01d.pnm || exit 1 + fi; + for subpg in $($my_seq 1 2); do echo Processing subpg ${subpg}. - if $(dirname $0)/detect2pages.sh $dir/upg-${pgn}-${subpg}.pnm ${pgn} ${subpg}; then + + subpgfile=${unppnmbase}-${subpg}.pnm #$unppnmbase has $dir/ already + + prepsubpnm=$dir/upg-pp-${pgn}-${subpg}.pnm + + # remask and retrim. FIXME: may throw off 2page detection + convert $convertflags -resize 10% -depth 8 -gamma 0.01 -median 2 $subpgfile $dir/spg-mask-${pgn}-${subpg}.png ||exit 1 + + # Trim #-border 1x1 -bordercolor '#fff' -trim -fuzz 30% + cropcords=$(convert -trim -fuzz 90%\ + -resize 1000% -format "%wx%h%O" $dir/spg-mask-${pgn}-${subpg}.png info: || exit 1) + + # ... crop and despeckle? the final pre-prepared image + convert $convertflags -crop $cropcords $subpgfile $prepsubpnm || exit 1 + + #detect if the page is 2-up + + if $(dirname $0)/detect2pages.sh ${prepsubpnm} ${pgn} ${subpg}; then if [[ $verbose == 1 ]]; then echo "Resplitting subpg ${subpg}." fi - unpaper $unpaperflags --pre-rotate 90 --layout double --overwrite --no-blackfilter -op 2 $dir/upg-${pgn}-${subpg}.pnm $dir/upg-${pgn}-${subpg}-%01d.pnm + unpaper $unpaperflags --pre-rotate 90 --layout double --overwrite -op 2 $prepsubpnm $dir/upg-${pgn}-${subpg}-%01d.pnm || exit 1 else - cp $dir/upg-${pgn}-${subpg}.pnm $dir/upg-${pgn}-${subpg}-1.pnm + cp $prepsubpnm $dir/upg-${pgn}-${subpg}-1.pnm fi; + + #final convert and clean w/ bebook optimisation + if [[ $bebook ]]; then + convert $convertflags -colorspace Gray -median 1 $dir/upg-${pgn}-${subpg}-?.pnm -trim -fuzz 80% -resize 1200x1600 $dir/final-${pgn}-${subpg}-%01d.${extension} || exit 1 + else + convert $convertflags $dir/upg-${pgn}-${subpg}-?.pnm $dir/final-${pgn}-${subpg}-%01d.${extension} || exit 1 + fi + done; - #final convert and clean w/ bebook optimisation - if [[ $bebook ]]; then - convert $convertflags -colorspace Gray -median 1 $dir/upg-${pgn}-1-?.pnm -trim -resize 1800x2400 $dir/final-${pgn}-1-%01d.${extension} || exit 1 - convert $convertflags -colorspace Gray -median 1 $dir/upg-${pgn}-2-?.pnm -trim -resize 1800x2400 $dir/final-${pgn}-2-%01d.${extension} || exit 1 - else - convert $convertflags $dir/upg-${pgn}-1-?.pnm $dir/final-${pgn}-1-%01d.${extension} || exit 1 - convert $convertflags $dir/upg-${pgn}-2-?.pnm $dir/final-${pgn}-2-%01d.${extension} || exit 1 - fi - done mkdir -p $dir/pages