def test_image(self):
"""Test SetImage and GetUTF8Text."""
self._api.Init()
self._api.SetImage(self._image)
text = self._api.GetUTF8Text()
self.assertIn('quick', text)
text2 = tesserocr.image_to_text(self._image)
self.assertEqual(text, text2)
def ocr_page(self, page_no):
if tesserocr is None:
return ''
with self.get_image(page_no, resolution=300) as img:
pil_image = PILImage.frombytes('RGB', img.size, img.make_blob('RGB'))
return tesserocr.image_to_text(
pil_image,
lang=TESSERACT_LANGUAGE[self.language],
path=self.config.get('TESSERACT_DATA_PATH', '')
)
def run_ocr(jid):
"""Perform OCR using Tesseract on the :class:`~toja.models.image.Image` with the id ``jid`` and the type ``'joke'``.
Creates a new :class:`~toja.models.transcription.Transcription` for the image with the OCRed text set as the
text and a ``status`` of ``'ocr'``. If such a :class:`~toja.models.transcription.Transcription` already exists,
then the text is updated instead.
This is a dramatiq Actor, so can be run in the background.
"""
storage_path = ConfigMiddleware.config_setting('app.images.storage.path')
if storage_path:
dbsession = DBSessionMiddleware.dbsession()
joke = dbsession.query(Image).filter(
and_(Image.id == jid, Image.type == 'joke')).first()
if joke:
transcription = dbsession.query(Transcription).filter(
and_(Transcription.status == 'ocr',
Transcription.source == joke)).first()
if not transcription:
transcription = Transcription(source=joke,
text={
'type':
'doc',
'content': [{
'type':
'paragraph',
'content': [{
'type': 'text',
'text': ''
}]
}]
},
status='ocr',
attributes={})
img = PILImage.open(os.path.join(storage_path, *joke.padded_id()))
raw_text = tesserocr.image_to_text(img).strip().replace('--', '—')
transcription.text = {
'type':
'doc',
'content': [{
'type':
'paragraph',
'content': [{
'type': 'text',
'text': para.replace('\n', ' ')
}]
} for para in raw_text.split('\n\n') if para.strip() != '']
}
dbsession.add(transcription)
else:
raise OCRException("Joke not found in the database")
else:
raise OCRException(
"Missing configuration setting app.images.storage.path")
def getTextTest():
image = cv2.imread("okey-check6.jpg", 0)
api = PyTessBaseAPI()
api.SetImage(Image.fromarray(image))
# print api.GetUTF8Text()
image = api.GetThresholdedImage()
image.save("rus.new.exp6.tiff")
text = image_to_text(image, lang="rus")
# text = image_to_string(image, lang="rus")
print text
return receiptParser.getJson(text)
def ocr_run(image_path):
ocr_image = Image.open(image_path)
value_text = tesserocr.image_to_text(ocr_image)
value_text = value_text.lower() # because letters are no fun
value_text = value_text.replace("l", "1")
value_text = value_text.replace("o", "0")
value_text = value_text.replace("s", "5")
value_text = value_text.strip()
value_text = "SO" + value_text
print("Value Text: " + str(value_text))
return value_text
def get_text(loc, typ):
im = ImageGrab.grab(bbox=loc)
#im.show()
txt = tesserocr.image_to_text(im)
#print(txt)
# print("="*20)
if typ == 'o':
txt = get_opt(txt)
else:
txt = get_query(txt)
return txt
def image2text(image_name: str) -> str:
image = Image.open(image_name)
image = image.convert('L')
"""
new_size = tuple(2 * x for x in image.size) # enlarge the image size
image = image.resize(new_size, Image.ANTIALIAS)
"""
# image.show()
return tesserocr.image_to_text(image,
lang='eng+chi_sim',
psm=tesserocr.PSM.SINGLE_BLOCK)
def same_old_solve():
"""
识别率依旧很低, 不提了...
"""
for i in range(10001):
picture_path = "cap1/im{}.png".format(i)
picture = Image.open(picture_path)
result1 = tesserocr.image_to_text(picture.convert('L'), lang='eng')
result2 = tesserocr.file_to_text(picture_path, psm=tesserocr.PSM.AUTO)
print(result1)
print(result2)
def img_ocr(self, context, to_key, *args):
"""Optical Character Recognition(OCR) using tesseract
Args:
context(dict): dictionary of context info
to_key(str): key to store ocr results
*args:
Returns:
dict
"""
# set PSM (Page Segmentation Mode) to 6 to handle
# images containing only numerals
psm = 6
with open_image(context["uuid"], context["key"], self.redis_conn,
self.binary_r) as img:
logger.info(image_to_text(img, psm=psm))
# r redis conn basically global
self.redis_conn.hset(context["uuid"], to_key,
image_to_text(img, psm=psm).strip())
return context
def lambda_handler(event, context):
supported_formats = ['jpeg', 'png', 'tiff']
for record in event['Records']:
size = record['s3']['object']['size']
s3_name = record['s3']['bucket']['name']
s3_key = record['s3']['object']['key']
s3_path = s3_name + '/' + s3_key
file_local = '/tmp/' + os.path.basename(s3_key)
if size > 15728640:
print('S3 object s3://' + s3_path +
' larger than 50MB, hence did not process')
try:
s3_vid = record['s3']['object']['versionId']
except KeyError:
s3_vid = ''
try:
if s3_vid:
s3.download_file(s3_name,
s3_key,
file_local,
ExtraArgs={'VersionId': s3_vid})
else:
s3.download_file(s3_name, s3_key, file_local)
except:
print('Failed to download the S3 object: s3://' + s3_path)
traceback.print_exc()
continue
try:
ftype = imghdr.what(file_local)
if ftype not in supported_formats:
print('S3 object s3://' + s3_path +
' is not one of the supported image format.')
continue
print('S3 object s3://' + s3_path + ' is of file type ' + ftype)
except:
print('Failed to check the format of the S3 object: s3://' +
s3_path)
traceback.print_exc()
continue
try:
print('Going to recognize ' + file_local)
rec_text = tesserocr.image_to_text(PIL.Image.open(file_local))
update_status(s3_path, s3_vid, rec_text)
except:
print('OCR failed to recognize the S3 object: s3://' + s3_path)
traceback.print_exc()
continue
def dist_captcha_code():
img = Image.open('code.png')
img_l = img.convert('L')
threshold = 180
table = []
for j in range(256):
if j < threshold:
table.append(0)
else:
table.append(1)
image = img_l.point(table, '1')
return tesserocr.image_to_text(image)
def getImage(filepath):
image = Image.open(filepath)
image = image.convert('L')
image = image.point(lambda x: 0 if x < 159 else 255, '1')
image.save("result.png")
f = open('extractedTestFromRec.txt', 'w')
a = tesserocr.image_to_text(image)
print a
#print type(a)
f.write(a.encode("UTF-8"))
f.close()
def get_market_change(self):
area = serpent.cv.extract_region_from_image(
self.current_frame.grayscale_frame,
self.game.screen_regions["MARKET_CHANGE"])
text = tesserocr.image_to_text(Image.fromarray(area))
matches = re.findall("\d*", text)
for match in matches:
if len(match) > 0:
return int(match.strip())
return -1
def plate_recognition(plate, visualize=True):
"""Performs OCR with Tesseract to a plate image.
Args:
plate (nparray): License plate cropped from an image (OpenCV).
visualize (bool): Processing images displayed or not (default True).
Returns:
plate_text (str): Text that Tesseract's been able to detect.
"""
# Process the current plate crop to make it easier to recognize
plate_processed = cv2.cvtColor(plate, cv2.COLOR_BGR2GRAY)
plate_processed = cv2.bilateralFilter(plate_processed, 11, 17, 17)
plate_processed = cv2.threshold(plate_processed, 177, 255,
cv2.THRESH_BINARY | cv2.THRESH_OTSU)[1]
if visualize:
cv2.imshow('Plate crop', plate)
cv2.moveWindow('Plate crop', 0, 0)
cv2.imshow('Processed plate', plate_processed)
cv2.moveWindow('Processed plate', 0, plate.shape[1])
cv2.waitKey(1000)
# Pass the plate crop through an OCR system
print("Without preprocessing: ")
print("Pytesseract: {}".format(pytesseract.image_to_string(plate)))
img = Image.fromarray(plate)
print("OCR: {}".format(tesserocr.image_to_text(img)))
print("With preprocessing: ")
plate_text = pytesseract.image_to_string(plate_processed)
print("Pytesseract: {}".format(plate_text))
img = Image.fromarray(plate)
print("OCR: {}".format(tesserocr.image_to_text(img)))
#cv2.waitKey(0)
plate_text = ''.join(ch for ch in plate_text if ch.isalnum())
print('Recognized text into the license plate: {}'.format(plate_text))
cv2.destroyAllWindows()
return plate_text
def login(self):
# 创建会话
session = requests.Session()
# 访问登录界面
login_url = 'https://sso.ckcest.cn/portal/signin?locale=zh_CN&device=pc&type=account&service=http%3A%2F%2Fwww.ckcest.cn%2Fcas'
login_view = session.get(url=login_url)
res = Selector(text=login_view.text)
# 获取请求数据
app_key = res.xpath('//input[@name="app_key"]/@value').get()
# 获取验证码
code = token = ''
while len(code) != 4:
self.logger.info('********* 识别验证码 *********')
url_image_json = 'https://sso.ckcest.cn/api/auth_web/captcha?width=120&height=51&type=char&app_key={}×tamp={}'
response_imgjson = session.get(
url_image_json.format(app_key,
time.time() * 1000)).json()
token = response_imgjson.get('data').get('token')
img_str = response_imgjson.get('data').get('base64Image').split(
',', 1)[-1]
img_byte = base64.b64decode(img_str)
img_file = io.BytesIO(img_byte)
# 识别验证码
img = Image.open(img_file)
code = tesserocr.image_to_text(img).strip()
# self.logger.info(f'********* {code}*********')
# 访问登录API
login_api = 'https://sso.ckcest.cn/api/auth_web/login_code_token'
data = dict()
data['account'] = '17109324122'
data['password'] = '******'
data['code'] = code
data['app_key'] = app_key
data[
'service'] = 'https://sso.ckcest.cn/portal/signin?locale=zh_CN&device=pc&type=account&service=http%3A%2F%2Fwww.ckcest.cn%2Fcas&_s=http%3A%2F%2Fwww.ckcest.cn%2Fcas'
data['_eventId'] = 'submit'
data['token'] = token
login = session.post(url=login_api, data=data)
res = Selector(text=login.text)
result = res.xpath('//title[contains(text(), "登录-中国工程科技知识")]').get()
if result:
# 循环登录
return False
# 请求表格数据 获取JSESSIONID和ASP.NET_SessionId 字段
session.get(url=self.table_url)
self.JSESSIONID = session.cookies.get('JSESSIONID',
domain='zhcy.app.ckcest.cn')
return True
def main():
image = Image.open('test.jpg')
image = image.convert('L')
threshold = 120
table = []
for i in range(256):
if i < threshold:
table.append(0)
else:
table.append(1)
image = image.point(table, '1')
result = tesserocr.image_to_text(image)
print(result)
def ImageRecognition(self):
pointArr = self.__RotatePictures()
plt.imshow(pointArr)
plt.show()
# 使用tesserocr进行识别，lang选择自己训练的语言库
result = tesserocr.image_to_text(pointArr, lang='fontyp')
# 格式化输出结果
result = [
re.findall('[A-Za-z0-9\u4e00-\u9fa5]', result[i])
for i in range(len(result))
]
result = ''.join([i[0] if i else '' for i in result])
return result
def decode_img(self, img_file):
im = Image.open(img_file)
reenforced_im = self.transform_img(im)
expression = tesserocr.image_to_text(reenforced_im)
code_value = None
try:
code_value = self.calculate(expression)
except:
pass
return code_value
def getText(rotation, image):
ImageFile.LOAD_TRUNCATED_IMAGES = True
image = Image.open(image)
image.save("loaded.jpg", dpi=(600, 600))
image = cv2.imread("loaded.jpg", 0)
if rotation != 0:
image = np.rot90(image, (360 - rotation) / 90)
# cv2.imwrite("rotated.jpg", image)
api = PyTessBaseAPI()
api.SetImage(Image.fromarray(image))
image = api.GetThresholdedImage()
text = image_to_text(image, lang="rus")
return receiptParser.getJson(text)
def img_ocr_key(context, key, to_key, *args):
"""Optical Character Recognition(OCR) using tesseract
Args:
context(dict): dictionary of context info
key: key to use to ocr
to_key(str): key to store ocr results
*args:
Returns:
dict
"""
# set PSM (Page Segmentation Mode) to 6 to handle
# images containing only numerals
psm = 6
with open_image(context['uuid'], key) as img:
logger.info(image_to_text(img, psm=psm))
# r redis conn basically global
# set PSM (Page Segmentation Mode) to 6 to handle
# images containing only numerals
r.hset(context['uuid'], to_key, image_to_text(img, psm=psm).strip())
return context
def solve():
"""
先旋转再识别
:return:
"""
for i in range(10001):
picture_path = "cap1/im{}.png".format(i)
picture = Image.open(picture_path)
# picture = rotate_picture(picture)
picture = online_rotate_picture(picture)
result1 = tesserocr.image_to_text(picture.convert("L"), lang="eng")
# picture.show()
print(result1.strip())
def get_captcha():
image = Image.open("1.jpg")
image = image.convert('L')
threshold = 220
table = []
for i in range(256):
if i < threshold:
table.append(0)
else:
table.append(1)
image = image.point(table, '1')
result = tesserocr.image_to_text(image)
print(result)
def test_image_orc():
folder = './examples'
for name in os.listdir(folder):
if not name.endswith('.png'):
continue
path = os.path.join(folder, name)
print(path)
image = Image.open(path)
result = tesserocr.image_to_text(image)
print(result)
def recognize_captcha(img_path):
im = Image.open(img_path)
# threshold = 140
# table = []
# for i in range(256):
# if i < threshold:
# table.append(0)
# else:
# table.append(1)
#
# out = im.point(table, '1')
num = tesserocr.image_to_text(im)
return num
def main(input_path):
try:
image = Image.open(input_path)
text = tesserocr.image_to_text(
image,
lang=
'eng+rus+bul+grc+srp+srp_latn+Georgian+Greek+spa+Cyrillic+pol+nld+hun+fin+deu+tur'
)
# text = " ".join(text1.split())
except:
text = ""
return text
def zhihulogin(username, password):
# 获取页面
driver = webdriver.Chrome()
driver.get('https://www.zhihu.com/#signin')
driver.find_element_by_xpath(
'//*[@class="SignContainer-switch"]/span').click()
driver.find_element_by_name('username').send_keys(username)
driver.find_element_by_name('password').send_keys(password)
driver.find_element_by_class_name('SignFlow-submitButton').click()
# 截取当前页面
driver.save_screenshot("1.png")
time.sleep(3)
try:
# 验证码所在的位置
element = driver.find_element_by_class_name('Captcha-englishImg')
location = element.location
print(location)
size = element.size
print(size)
# 计算出元素上下左右位置
left = element.location['x']
top = element.location['y']
right = element.location['x'] + element.size['width']
bottom = element.location['y'] + element.size['height']
im = Image.open('1.png')
im = im.crop((left, top, right, bottom))
im.save('2.png')
# 打开图片
image = Image.open('2.png')
image = image.convert('L')
threshod = 127
table = []
for i in range(256):
if i < threshod:
table.append(0)
else:
table.append(1)
image = image.point(table, '1')
result = tesserocr.image_to_text(image)
print('验证码是：', result)
driver.find_element_by_class_name('SignFlowInput-errorMask').send_keys(
result)
driver.find_element_by_class_name('SignFlow-submitButton').click()
time.sleep(10)
except Exception as e:
print('出现错误了～')
print(e)
def get_working_buys(self):
area = serpent.cv.extract_region_from_image(
self.current_frame.grayscale_frame,
self.game.screen_regions["WORKING_BUYS"])
self.visual_debugger.store_image_data(area, area.shape, 3)
text = tesserocr.image_to_text(Image.fromarray(area))
matches = re.findall("\(\d\)", text)
for match in matches:
if len(match) > 0:
return int(match.strip(' ()'))
return -1
def getText(rotation, image):
ImageFile.LOAD_TRUNCATED_IMAGES = True
image = Image.open(image).convert("RGB")
b, g, r = image.split()
image = Image.merge("RGB", (r, g, b))
image = cv2.cvtColor(np.array(image), cv2.COLOR_BGR2GRAY)
if rotation != 0:
image = np.rot90(image, (360 - rotation) / 90)
api = PyTessBaseAPI()
api.SetImage(Image.fromarray(image))
image = api.GetThresholdedImage()
text = image_to_text(image, lang="rus")
# print text
return receiptParser.getJson(text)
def image_test():
image = Image.open('CheckCode.jpg')
image = image.convert('L')
threshold = 140
table = []
for i in range(256):
if i < threshold:
table.append(0)
else:
table.append(1)
image = image.point(table, '1')
image.show()
result = tesserocr.image_to_text(image)
print(result)
def check_code(self, img_url):
img = Image.open(img_url)
threshold = 80
img = img.convert('L')
table = []
for i in range(256):
if i < threshold:
table.append(0)
else:
table.append(1)
img = img.point(table, '1')
img.show()
result = tesserocr.image_to_text(img)
print(result)
def book_clear(image, threshold):
image = image.convert("L")
table = []
for i in range(256):
if i < threshold:
table.append(0)
else:
table.append(1)
img = image.point(table, "1")
# img.save("./photo/img1.png")
# img.show()
result = tesserocr.image_to_text(img)
print('灰度二值化之后：' + result)
return img
def get_code(url):
flag = url.split("/")[-2]
fn = flag + '.png'
with open(fn, 'wb+') as sw:
sw.write(requests.get(url).content)
img = Image.open(fn)
img = img.convert('L')
# table = [0 if i < 127 else 1 for i in range(256)]
# img = img.point(table, '1')
result = tesserocr.image_to_text(img).strip()
print(flag, result)
if re.match('^[A-Za-z0-9]{4}$', result):
return flag, result
import tesserocr
from PIL import Image
image = Image.open('code2.jpg')
image = image.convert('L')
threshold = 127
table = []
for i in range(256):
if i < threshold:
table.append(0)
else:
table.append(1)
image = image.point(table, '1')
image.show()
result = tesserocr.image_to_text(image)
print(result)
