Strip vertical whitespace from letters.

pixelocr/image.py

@@ -7,6 +7,11 @@ from skimage.color import rgb2gray
 from .utils import cached_property
 
 
+def _is_nonblank(bitmap):
+    """Return True if bitmap contains at least one black (=1) pixel."""
+    return bitmap.any()
+
+
 class Image(object):
     """Basic image class."""
 
@@ -58,6 +63,10 @@ class Image(object):
         grayscale = rgb2gray(self._data)
         return (grayscale < 1).astype('b')
 
+    @cached_property
+    def isspace(self):
+        return not _is_nonblank(self.bitmap)
+
     @property
     def key(self):
         """Return a byte string uniquely representing the image."""
@@ -79,6 +88,25 @@ class Image(object):
     def unframe(self, width=2):
         return self[width:-width,width:-width]
 
+    def strip(self):
+        """Strip top and bottom blank space.
+        
+        All-whitespace images are not stripped.
+        """
+
+        if self.isspace:
+            return self
+
+        def _get_margin_height(rows):
+            for i, row in enumerate(rows):
+                if _is_nonblank(row):
+                    return i
+            return 0
+
+        top_margin = _get_margin_height(self.bitmap)
+        bottom_margin = _get_margin_height(reversed(self.bitmap))
+        return self[top_margin:self.height - bottom_margin, :]
+
     def _iter_children(self, min_space):
         if self.child_cls is None:
             raise NotImplementedError
@@ -86,11 +114,8 @@ class Image(object):
         line_start = None
         prev_line_end = 0
 
-        def is_nonblank(row):
-            return row.any()
-
         for i, row in enumerate(self.bitmap):
-            if is_nonblank(row):
+            if _is_nonblank(row):
                 if line_start is None:
                     line_start = i
                     height = line_start - prev_line_end
@@ -112,7 +137,7 @@ class Line(Image):
 
     def __iter__(self):
         for rotated_letter in self.T._iter_children(min_space=10):
-            yield rotated_letter.T
+            yield rotated_letter.T.strip()
 
 
 class Page(Image):