From: Daniel Axtens <dja@ucc.asn.au>
Date: Fri, 8 Apr 2011 07:32:14 +0000 (+0800)
Subject: Improved border artifact work; dealing with blank scans; 2up page detection
X-Git-Url: https://git.ucc.asn.au/?p=dja%2Fscandal.git;a=commitdiff_plain;h=13a56d0f2178dd64270ae0b9d6e2994de42e9c07

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

diff --git a/detect2pages.sh b/detect2pages.sh
new file mode 100755
index 0000000..4b8c189
--- /dev/null
+++ b/detect2pages.sh
@@ -0,0 +1,45 @@
+#!/bin/bash
+
+#should probably have given up on shell and gone over to python.
+
+
+infile=$1
+pg=$2
+subpg=$3
+
+dir=$(dirname $infile)
+
+convert -blur 10 -gamma 0.00001 -median 1 $infile -resize 1x601\! -gamma 0.3 $dir/strip-${pg}-${subpg}.png
+
+#There seems to be a definte bias in my reader scans for things to be in the top part rather than the bottom part
+testpoints=`jot - 270 320`
+
+sum=0
+prev=0
+longestrun=0
+
+
+for testpoint in $testpoints; do
+	#convert -verbose strip-${pg}-${subpg}.png[1x1+0+${testpoint}] -format "%[fx:s.intensity]" info:
+	value=$(convert $dir/strip-${pg}-${subpg}.png[1x1+0+${testpoint}] -format "%[fx:s.intensity>0.3]" info:)
+	sum=$(expr $value + $sum)
+
+	if [[ $prev = 1 && $value = 0 ]]; then
+		if [[ $sum > $longestrun ]]; then
+			longestrun=$sum;
+		fi;
+		sum=0
+	fi;
+
+	prev=$value
+
+	#echo "$testpoint: $value ($sum, $longestrun)"
+done;
+
+if [[ $sum > $longestrun ]]; then
+	longestrun=$sum;
+fi
+
+echo $longestrun
+
+exit $( expr $longestrun "<=" 10 )
diff --git a/scan2pages.sh b/scan2pages.sh
index 321d31c..b129963 100755
--- a/scan2pages.sh
+++ b/scan2pages.sh
@@ -160,21 +160,46 @@ for pg in `$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
 
+	#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}-?.pnm -resize 1200x1600 $dir/final-${pgn}-%01d.${extension} || exit 1
+		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}-?.pnm $dir/final-${pgn}-%01d.${extension} || exit 1
+		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
+mkdir -p $dir/pages
 mv $dir/final-*.${extension} $dir/pages