Take optical distance into account when splitting lines.

pixelocr/page.py

@@ -133,7 +133,12 @@ class Page(PageObject):
             if prev_line is None:
                 prev_line = line
             else:
-                if line.top - prev_line.bottom < min_space:
+                distance = line.top - prev_line.bottom
+                if (
+                    # avoid unnecessary calling optical_distance() which may be expensive
+                    distance < min_space
+                    and prev_line.optical_distance(line) < min_space
+                ):
                     prev_line = Line(self.image[prev_line.top:line.bottom])
                 else:
                     yield prev_line