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

Add Line.page and Glyph.line.

Browse source
This commit is contained in:
Andrey Golovizin 2014-08-30 20:50:59 +02:00
parent f3808df0cd
commit a373765d3d
1 changed files with 12 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

19
pixelocr/page.py
View file

@ -124,11 +124,11 @@ class Page(PageObject):
line_start = i line_start = i
else: else:
if line_start is not None: if line_start is not None:
yield Line(self.image[line_start:i,:]) yield Line(self, self.image[line_start:i,:])
line_start = None line_start = None
prev_line_end = i prev_line_end = i
if line_start is not None: if line_start is not None:
yield Line(self.image[line_start:,:]) yield Line(self, self.image[line_start:,:])
def _merge_lines(self, lines, min_space=2, min_height=10): def _merge_lines(self, lines, min_space=2, min_height=10):
prev_line = None prev_line = None
@ -144,7 +144,7 @@ class Page(PageObject):
) )
not_high_enough = prev_line.height < min_height not_high_enough = prev_line.height < min_height
if too_close or not_high_enough: if too_close or not_high_enough:
prev_line = Line(self.image[prev_line.top:line.bottom]) prev_line = Line(self, self.image[prev_line.top:line.bottom])
else: else:
yield prev_line yield prev_line
prev_line = line prev_line = line
@ -153,6 +153,10 @@ class Page(PageObject):
class Line(PageObject): class Line(PageObject):
def __init__(self, page, image):
super().__init__(image)
self.page = page
def __iter__(self): def __iter__(self):
return iter(self.glyphs) return iter(self.glyphs)
@ -180,7 +184,7 @@ class Line(PageObject):
for (label, blob_slice) in blob_slices for (label, blob_slice) in blob_slices
) )
glyphs = ( glyphs = (
Glyph(image, self.baseline - image.bottom) Glyph(self, image, self.baseline - image.bottom)
for image in glyph_images for image in glyph_images
) )
glyphs = sorted(glyphs, key=lambda glyph: (glyph.left, -glyph.bottom)) glyphs = sorted(glyphs, key=lambda glyph: (glyph.left, -glyph.bottom))
@ -214,7 +218,7 @@ class Line(PageObject):
if next_glyph is not None: if next_glyph is not None:
distance = glyph.optical_distance(next_glyph) distance = glyph.optical_distance(next_glyph)
if distance >= min_distance: if distance >= min_distance:
yield Space(self.image.space(glyph.right, self.top, distance, self.height), self.baseline - self.top) yield Space(self, self.image.space(glyph.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]
@ -229,9 +233,10 @@ class Line(PageObject):
class Glyph(PageObject): class Glyph(PageObject):
MIN_BODY_HEIGHT = 10 MIN_BODY_HEIGHT = 10
def __init__(self, image, elevation): def __init__(self, line, image, elevation):
super().__init__(image) super().__init__(image)
self.elevation = elevation self.elevation = elevation
self.line = line
@property @property
def key(self): def key(self):
@ -287,7 +292,7 @@ class Glyph(PageObject):
if not diacritics: if not diacritics:
return self return self
diacritic_images = (diacritic.image for diacritic in diacritics) diacritic_images = (diacritic.image for diacritic in diacritics)
return Glyph(combine(self.image, *diacritic_images), self.elevation) return Glyph(self.line, combine(self.image, *diacritic_images), self.elevation)
class Space(Glyph): class Space(Glyph):