def get_preprocessed_image(self, color_threshold=225):
im = self.im
if im.mode != 'L':
im = im.convert('L')
im = self.im.filter(ImageFilter.SHARPEN)
im = threshold(im, color_threshold)
im = im.filter(ImageFilter.FIND_EDGES)
im = ImageOps.invert(im)
return im
def split(self, output_dir, **kwargs):
"""Split an image that contains a table into per-cell images"""
for box in self.get_boxes(**kwargs):
col, row, left, upper, right, lower = box
region = self.im.crop((left, upper, right, lower))
region = self.transform(region, col, row)
# Increase contrast. Whiteish regions become absolute white
region = threshold(region, 200)
output_filename = cell_basename(self.filename, col, row) + ".tiff"
output_path = os.path.join(output_dir, output_filename)
region.save(output_path)
md5 = md5sum(output_path)
split_image.send(self, input_filename=self.filename,
input_md5=self.md5, filename=output_filename, md5=md5,
column=col, row=row, left=left, upper=upper, right=right,
lower=lower)
