ocropus-gpageseg: Enable usage of masks to specify column separators/ ignore areas of scan

ocropus-gpageseg

@@ -171,7 +171,7 @@ def DSAVE(title,image):
 ### This attempts to find column separators, either as extended
 ### vertical black lines or extended vertical whitespace.
 ### It will work fairly well in simple cases, but for unusual
-### documents, you need to tune the parameters.
+### documents, you need to tune the parameters or use a mask.
 ################################################################
 
 def compute_separators_morph(binary,scale):
@@ -262,8 +262,19 @@ def compute_colseps(binary,scale):
         #colseps = compute_colseps_morph(binary,scale)
         colseps = np.maximum(colseps,seps)
         binary = np.minimum(binary,1-seps)
+    binary,colseps = apply_mask(binary,colseps)
     return colseps,binary
 
+def apply_mask(binary,colseps):
+    try:
+        mask = ocrolib.read_image_binary(base+".mask.png")
+    except IOError:
+        return binary,colseps
+    masked_seps = np.maximum(colseps,mask)
+    binary = np.minimum(binary,1-masked_seps)
+    DSAVE("masked_seps", masked_seps)
+    return binary,masked_seps
+
 
 ################################################################
 ### Text Line Finding.

run-test-ci

@@ -77,6 +77,8 @@ test_gpageseg() {
     $RUNNER $BASE/ocropus-gpageseg temp/$TESTIMAGE -n --maxseps 3
     $RUNNER $BASE/ocropus-gpageseg temp/$TESTIMAGE -n -b
     $RUNNER $BASE/ocropus-gpageseg temp/$TESTIMAGE -n --usegauss
+    cp $BASE/tests/table.bin.png $BASE/tests/table.mask.png temp
+    $RUNNER $BASE/ocropus-gpageseg temp/table.bin.png --debug -n --minscale 7 --maxcolseps 0
 }
 
 test_rpred() {