Add Image.color and PageObject.color.

pixelocr/image.py

@@ -18,6 +18,7 @@ import itertools
 from io import BytesIO
 
 import numpy as np
+from scipy.stats.mstats import mode
 from skimage.io import imread, imsave
 
 from .utils import cached_property, pairwise
@@ -162,6 +163,16 @@ class Image(object):
     def isspace(self):
         return not is_nonblank(self.bitmap)
 
+    @cached_property
+    def color(self):
+        """Return the most frequent foreground color."""
+        if self.isspace:
+            return None
+        mask3 = np.dstack([~self.bitmap] * 3)
+        colors = np.ma.MaskedArray(self.data, mask3).reshape(-1, 3)
+        modes, counts = mode(colors)
+        return tuple(modes[0])
+
     def serialize(self):
         """Serialize the image as some hashable object."""
         bitmap = self.data.astype(np.uint8).tostring()

pixelocr/page.py

@@ -80,6 +80,10 @@ class PageObject(object):
     def ycenter(self):
         return (self.bottom - self.top) / 2
 
+    @property
+    def color(self):
+        return self.image.color
+
     def fits(self, left, top, right, bottom):
         """Return True if the glyph fits into the given bounding box."""