segment-region: clip polygons to parent...

(before annotating polygon coordinates, ensure they fit
 into their parent if it exists, or the page frame otherwise)

CHANGELOG.md

@@ -5,6 +5,10 @@ Versioned according to [Semantic Versioning](http://semver.org/).
 
 ## Unreleased
 
+Fixed:
+
+  * segment-region: ensure polygons are within page/Border
+
 Changed:
 
   * segment-region: in `sparse_text` mode, also add text lines

ocrd_tesserocr/segment_region.py

@@ -1,6 +1,7 @@
 from __future__ import absolute_import
 
 import os.path
+from shapely.geometry import Polygon
 from tesserocr import (
     PyTessBaseAPI,
     PSM, RIL, PT
@@ -20,6 +21,7 @@
     MetadataItemType,
     LabelsType, LabelType,
     CoordsType, AlternativeImageType,
+    PageType,
     OrderedGroupType,
     ReadingOrderType,
     RegionRefIndexedType,
@@ -215,6 +217,7 @@ def _process_page(self, it, page, page_image, page_coords, page_id):
             else:
                 polygon = polygon_from_x0y0x1y1(bbox)
             polygon = coordinates_for_segment(polygon, page_image, page_coords)
+            polygon = polygon_for_parent(polygon, page)
             points = points_from_polygon(polygon)
             coords = CoordsType(points=points)
             # if xywh['w'] < 30 or xywh['h'] < 30:
@@ -299,3 +302,28 @@ def _process_page(self, it, page, page_image, page_coords, page_id):
             not og.get_UnorderedGroupIndexed()):
             # schema forbids empty OrderedGroup
             ro.set_OrderedGroup(None)
+
+def polygon_for_parent(polygon, parent):
+    """Clip polygon to parent polygon range.
+    
+    (Should be moved to ocrd_utils.coordinates_for_segment.)
+    """
+    childp = Polygon(polygon)
+    if isinstance(parent, PageType):
+        if parent.get_Border():
+            parentp = Polygon(polygon_from_points(parent.get_Border().get_Coords().points))
+        else:
+            parentp = Polygon([[0,0], [0,parent.get_imageHeight()],
+                               [parent.get_imageWidth(),parent.get_imageHeight()],
+                               [parent.get_imageWidth(),0]])
+    else:
+        parentp = Polygon(polygon_from_points(parent.get_Coords().points))
+    if childp.within(parentp):
+        return polygon
+    interp = childp.intersection(parentp)
+    if interp.is_empty:
+        # FIXME: we need a better strategy against this
+        raise Exception("intersection of would-be segment with parent is empty")
+    if interp.type == 'MultiPolygon':
+        interp = interp.convex_hull
+    return interp.exterior.coords[:-1] # keep open