ag/pixelocr
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Implement correction for optical distance between letters.

Browse source
This commit is contained in:
Andrey Golovizin 2014-08-15 14:40:02 +02:00
parent ce4252e361
commit c4b170cc21
1 changed files with 20 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

24
pixelocr/page.py
View file

@ -112,6 +112,22 @@ class Line(PageObject):
letters = self._combine_diacritics(letters) letters = self._combine_diacritics(letters)
return self._insert_spaces(letters) return self._insert_spaces(letters)
def _optical_correction(self, letter1, letter2):
base = min(letter1.top, letter2.top)
height = max(letter1.bottom, letter2.bottom) - base
bitmap1 = np.hstack([np.ones((letter1.height, 1)), letter1.image.bitmap])
bitmap2 = np.hstack([letter2.image.bitmap, np.ones((letter2.height, 1))])
margin1 = np.zeros(height, np.int)
margin1.fill(letter1.width)
margin2 = np.zeros(height, np.int)
margin2.fill(letter2.width)
margin1[letter1.top - base: letter1.bottom - base] = np.fliplr(bitmap1).argmax(axis=1)
margin2[letter2.top - base: letter2.bottom - base] = bitmap2.argmax(axis=1)
margins = margin1 + margin2
return margins.min()
def _combine_diacritics(self, letters): def _combine_diacritics(self, letters):
def is_diacritic(glyph): def is_diacritic(glyph):
# XXX # XXX
@ -133,9 +149,10 @@ class Line(PageObject):
for letter, next_letter in pairwise(letters): for letter, next_letter in pairwise(letters):
yield letter yield letter
if next_letter is not None: if next_letter is not None:
distance = next_letter.left - letter.right correction = self._optical_correction(letter, next_letter)
distance = next_letter.left - letter.right + correction
if distance > 5: if distance > 5:
yield Space(self.image.space(letter.right, self.top, distance, self.baseline - self.top)) yield Space(self.image.space(letter.right, self.top, distance, self.height), self.baseline - self.top)
def _extract_blob(self, blob_slice, label, labels): def _extract_blob(self, blob_slice, label, labels):
image = self.image[blob_slice] image = self.image[blob_slice]
@ -150,5 +167,4 @@ class Letter(PageObject):
class Space(Letter): class Space(Letter):
def __init__(self, image): pass
super().__init__(image, elevation=0)