def extract_image_data(data, languages=None):
"""Extract text from a binary string of data."""
tessdata_prefix = get_config('TESSDATA_PREFIX')
if tessdata_prefix is None:
raise IngestorException("TESSDATA_PREFIX is not set, OCR won't work.")
languages = get_languages_iso3(languages)
text = Cache.get_ocr(data, languages)
if text is not None:
return text
try:
img = Image.open(StringIO(data))
except DecompressionBombWarning as dce:
log.debug("Image too large: %", dce)
return None
except IOError as ioe:
log.info("Unknown image format: %r", ioe)
return None
# TODO: play with contrast and sharpening the images.
extractor = Tesseract(tessdata_prefix, lang=languages)
extractor.set_page_seg_mode(PageSegMode.PSM_AUTO_OSD)
text = extractor.ocr_image(img)
extractor.clear()
log.debug('OCR done: %s, %s characters extracted',
languages, len(text))
Cache.set_ocr(data, languages, text)
return text
def ocr_text(img):
'''Perform OCR on the image.'''
tr = Tesseract(lang='eng')
tr.clear()
pil_image = pil.Image.fromarray(img)
tr.set_image(pil_image)
utf8_text = tr.get_text()
return utf8_text
def ocr(img,idioma):
ocr_img = Image.fromarray(img)
ocr = Tesseract(lang=idioma)
ocr.set_image(ocr_img)
pattern = re.compile('[a-zA-Z0-9]')
text = ocr.get_utf8_text()
text = text.splitlines()
text = [x for x in text if x != '']
text = [x for x in text if pattern.search(x)]
ocr.clear()
return (text)
def ocr_text(img):
tr = Tesseract(lang='eng')
tr.clear()
pil_image = pil.Image.fromarray(img)
# Turn off OCR word dictionaries
tr.set_variable('load_system_dawg', "F")
tr.set_variable('load_freq_dawg', "F")
tr.set_variable('-psm', "7") # treat image as single line
tr.set_variable('tessedit_char_whitelist', "ABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789")
tr.set_image(pil_image)
utf8_text = tr.get_text()
return unicode(utf8_text)
