Do less with unpaper.
[dja/scandal.git] / scan2pages.sh
index 42b3bfb..069b421 100755 (executable)
@@ -133,9 +133,9 @@ ocropus book2pages $outdir/scanpgs $outdir/scanpg-*.png || exit 1
 
 for scanpgnum in  `$my_seq 1 $pages`; do
 
+       binscanpg=$outdir/scanpgs/$(printf '%04d' $scanpgnum).bin.png
        scanpgnum=$(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
@@ -162,7 +162,11 @@ for scanpgnum in  `$my_seq 1 $pages`; do
        #unpaper it
        physpgbase=$outdir/physpg-${scanpgnum}
        if [ ! -e ${physpgbase}-1.pnm ] || [ ! -e ${physpgbase}-2.pnm ]; then
-               unpaper $unpaperflags --layout double --overwrite -ni 10 -op 2 $cleanscanpg ${physpgbase}-%01d.pnm || exit 1
+               # most of unpaper's processing is redundant given ocropus, and somewhat too agressive
+               unpaper $unpaperflags --layout double --overwrite -op 2 \
+                       --no-processing \
+                       $cleanscanpg ${physpgbase}-%01d.pnm || exit 1
+
        fi;
        
        for physpgnum in $($my_seq 1 2); do
@@ -197,18 +201,28 @@ for scanpgnum in  `$my_seq 1 $pages`; do
                        cp $cleanphyspg $outdir/logpg-${scanpgnum}-${physpgnum}-1.pnm
                fi;
 
-       
+               #prepare for ocr
+               convert $convertflags $outdir/logpg-${scanpgnum}-${physpgnum}-?.pnm $outdir/logpg-${scanpgnum}-${physpgnum}-%01d.png || exit 1
+
                #final convert and clean w/ bebook optimisation
                if [[ $bebook ]]; then
-                       convert $convertflags -colorspace Gray -median 1 $outdir/logpg-${scanpgnum}-${physpgnum}-?.pnm -trim -fuzz 80% -resize 1200x1600 $outdir/final-${scanpgnum}-${physpgnum}-%01d.${extension} || exit 1
+                       convert $convertflags $outdir/logpg-${scanpgnum}-${physpgnum}-?.pnm -trim -fuzz 80% -resize 1200x1600 $outdir/final-${scanpgnum}-${physpgnum}-%01d.${extension} || exit 1
                else
                        convert $convertflags $outdir/logpg-${scanpgnum}-${physpgnum}-?.pnm $outdir/final-${scanpgnum}-${physpgnum}-%01d.${extension} || exit 1
                fi
-       
+
        done;
 
 done
 
+#try full ocr
+rm -rf $outdir/logpgs
+ocropus book2pages $outdir/logpgs $outdir/logpg-*.png
+ocropus pages2lines $outdir/logpgs
+ocropus lines2fsts $outdir/logpgs
+ocropus fsts2bestpaths $outdir/logpgs
+ocropus buildhtml $outdir/logpgs > $outdir/out.html
+
 mkdir -p $outdir/pages
 mv $outdir/final-*.${extension} $outdir/pages
        

UCC git Repository :: git.ucc.asn.au