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

Add Word.baseline and Letter.elevation.

Browse source
This commit is contained in:
Andrey Golovizin 2014-08-13 19:55:31 +02:00
parent e3217d28b8
commit a15cd7c4a2
1 changed files with 12 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

15
pixelocr/page.py
View file

@ -80,7 +80,7 @@ class Line(PageObject):
@collect_iterable @collect_iterable
def words(self): def words(self):
for rotated_word_img in self.image.T._iter_lines(min_space=10): for rotated_word_img in self.image.T._iter_lines(min_space=10):
yield Word(rotated_word_img.T) yield Word(rotated_word_img.T, baseline=self.baseline)
@property @property
def letters(self): def letters(self):
@ -99,6 +99,10 @@ class Line(PageObject):
class Word(PageObject): class Word(PageObject):
def __init__(self, image, baseline):
super().__init__(image)
self.baseline = baseline
def __iter__(self): def __iter__(self):
return iter(self.letters) return iter(self.letters)
@ -108,10 +112,15 @@ class Word(PageObject):
labels, max_label = ndimage.label(self.image.bitmap, CONNECTIVITY8) labels, max_label = ndimage.label(self.image.bitmap, CONNECTIVITY8)
obj_indices = ndimage.find_objects(labels, max_label) obj_indices = ndimage.find_objects(labels, max_label)
letter_images = (self.image[obj_index] for obj_index in obj_indices) letter_images = (self.image[obj_index] for obj_index in obj_indices)
letters = (Letter(image) for image in letter_images) letters = (
Letter(image, self.y1 + self.baseline - image.y2)
for image in letter_images
)
sorted_letters = sorted(letters, key=lambda letter: (letter.x1, -letter.y1)) sorted_letters = sorted(letters, key=lambda letter: (letter.x1, -letter.y1))
return iter(sorted_letters) return iter(sorted_letters)
class Letter(PageObject): class Letter(PageObject):
pass def __init__(self, image, elevation):
super().__init__(image)
self.elevation = elevation