[ocrfeeder] Fix wrong font size calculations when there are initial characters
- From: Joaquim Manuel Pereira Rocha <jrocha src gnome org>
- To: commits-list gnome org
- Cc:
- Subject: [ocrfeeder] Fix wrong font size calculations when there are initial characters
- Date: Tue, 27 Jul 2010 22:07:57 +0000 (UTC)
commit 990ec303f61ed8f27796a4399b0306be055080a4
Author: Joaquim Rocha <jrocha igalia com>
Date: Thu Jul 8 11:19:12 2010 +0200
Fix wrong font size calculations when there are initial characters
layoutAnalysis.LayoutAnalysis: Use only the right half of an image when
calculating the font size, this will prevent measuring eventual initial
characters which create the "illusion" of a much larger line.
feeder/layoutAnalysis.py | 9 +++++++--
1 files changed, 7 insertions(+), 2 deletions(-)
---
diff --git a/feeder/layoutAnalysis.py b/feeder/layoutAnalysis.py
index 5da0498..63b827e 100644
--- a/feeder/layoutAnalysis.py
+++ b/feeder/layoutAnalysis.py
@@ -440,13 +440,18 @@ class LayoutAnalysis(object):
return data_box
def getTextSizeFromImage(self, image, page_resolution):
- text_size = graphics.getTextSizeFromImage(image)
+ width, height = image.size
+ # We get the right half of the image only because this
+ # way we avoid measuring eventual "initial chars" which
+ # leads to false text sizes (obviously this will fail
+ # for right-to-left languages)
+ image_right_half = image.crop((width / 2, 0, width, height))
+ text_size = graphics.getTextSizeFromImage(image_right_half)
if not text_size:
return None
y_resolution = float(page_resolution)
text_size /= y_resolution
text_size *= DTP
- print text_size
return round(text_size)
def readImage(self, image):
[
Date Prev][
Date Next] [
Thread Prev][
Thread Next]
[
Thread Index]
[
Date Index]
[
Author Index]