Improved border artifact work; dealing with blank scans; 2up page detection

[dja/scandal.git] / scan2pages.sh

diff --git a/scan2pages.sh b/scan2pages.sh
index c9a10c8..b129963 100755 (executable)
--- a/scan2pages.sh
+++ b/scan2pages.sh
@@ -2,20 +2,22 @@
 
 function usage {
 cat >&2 << __EOF__ 
-Usage: $0 [-vfpsmut] [-d depth] pdffile outdir
+Usage: $0 [-vfpsmutb] [-D DPI] [-d depth] [-F format] pdffile outdir
 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.
                Unpaper can only handle up to 8.
-       -D dpi: Use given DPI. Colour scans on library scanners are usually 300, bw 400. 
+       -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
        -s: skip masking/trimming. Overrides -m.
        -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
        -t: Forceably redo final trimming and cleaning. IGNORED
+       -b: Optimize for BeBook One.
 __EOF__
 }
 
@@ -40,7 +42,9 @@ depth=
 dpi=
 skipmask=
 verbose=
-while getopts 'vd:sfpmuc' OPTION
+extension="png"
+bebook=
+while getopts 'vd:D:sfpmucF:b' OPTION
 do
        case $OPTION in
        v)      verbose=1
@@ -63,9 +67,10 @@ do
                ;;
        t)      forceclean=1
                ;;
-#      b)      bflag=1
-#              bval="$OPTARG"
-#              ;;
+       F)      extension="$OPTARG"
+               ;;
+       b)      bebook=1
+               ;;
        ?)      usage
                exit 2
                ;;
@@ -143,11 +148,11 @@ for pg in `$my_seq 1 $pages`; do
        if [[ ! $skipmask && ( ! $([ -e $preppnm ]) || $forcemask ) ]]; then
                # create mask: 
                # ... downscale, blur,
-               convert $convertflags -resize 25% -depth 8 -blur 10 $origpnm $dir/pg-mask-${pgn}.pnm ||exit 1
+               convert $convertflags -resize 10% -depth 8 -blur 10 -median 2 $origpnm $dir/pg-mask-${pgn}.pnm ||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 400% -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}.pnm info: || exit 1)
                
                # ... crop and despeckle? the final pre-prepared image
                convert $convertflags -crop $cropcords $origpnm $preppnm || exit 1
@@ -155,18 +160,46 @@ for pg in `$my_seq 1 $pages`; do
                cp $origpnm $preppnm
        fi;
        
+       # check it hasn't mostly disappeared - e.g. if the scan was all black
+       # (e.g. forgot to put the book down when you first hit scan)
+       # ... not sure why such a convoluted form is required. expr doesn't like
+       # processing a convert -format "%w * %h" or any variant thereof
+       if [[ $(convert  $preppnm -format '%[fx:s.w*s.h>1000]' info:) = "0" ]]; then
+               [[ $verbose == 1 ]] && echo "Discarding pg ${pgn}: not enough remains after masking."
+               continue; 
+       fi;
+       
        #unpaper it
        #names go a bit funny here
        #also, ignore flags starting here
        unppnm=$dir/upg-${pgn}-%01d.pnm
        unpaper $unpaperflags --layout double --overwrite --no-blackfilter -ni 10 -op 2 $preppnm $unppnm || exit 1
 
-       # final convert and clean
-       convert $convertflags $dir/upg-${pgn}-?.pnm $dir/upg-${pgn}-%01d.tiff || exit 1
-       
+       #detect if the page is 2-up
+       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
+                       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
+               else
+                       cp $dir/upg-${pgn}-${subpg}.pnm $dir/upg-${pgn}-${subpg}-1.pnm
+               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 $dir/pages
-mv $dir/upg-*.tiff $dir/pages
+mkdir -p $dir/pages
+mv $dir/final-*.${extension} $dir/pages