def getSimplifiedFromImage():
'''
Take image in the form of a JSON
object, returns the simplfied text
'''
result = request.get_json()
requires = ["img"]
if isValid(requires, result):
info = result["img"]
if info[1:21] == "data:application/pdf":
type = "PDF"
else:
type = "IMG"
base64Image = info[info.find(',') + 1:]
text = 'undefined'
if type == "PDF":
text = OCR.process_PDF(base64Image)
elif type == "IMG":
text = OCR.process_IMG(base64Image)
return text
else:
logging.info("Invalid request.")
return "Failed to process image"
return "OK"
def check():
if "blank" in OCR.resolve_url():
print("! RESET")
return this_func(selected_link)
else:
if OCR.check_current_url():
time.sleep(1)
cmd = Commands()
cmd.go_to_top()
def main():
print("Loading Images ...")
# KNN ML Model ################################################################################################################
blnKNNTrainingSuccessful = DetectChars.loadKNNDataAndTrainKNN(
CLASS_TXT_DIR, FLAT_IMG_DIR) # attempt KNN training
if blnKNNTrainingSuccessful == False:
print("\nerror: KNN traning was not successful\n")
return
# end if
# Loading Images ##############################################################################################################
BR_IMGS = os.listdir(DATASET_DIR)
BR_IMGS = [img for img in BR_IMGS if ".jpg" in img]
extractedPlates = []
numberPlateList = []
filename = []
# Write extracted plates images to path
print("Begin Pre-processing images ...")
for BR_IMG in BR_IMGS:
BR_IMG_path = DATASET_DIR + BR_IMG
imgOriginal = cv2.imread(BR_IMG_path)
listOfPossiblePlates = DetectPlates.detectPlates(imgOriginal)
listOfPossiblePlates = DetectChars.detectCharsInPlates(
listOfPossiblePlates)
# Sorting list of possible plates in DESC order (the one with most chars recognized)
listOfPossiblePlates.sort(
key=lambda possiblePlate: len(possiblePlate.strChars),
reverse=True)
licPlate = listOfPossiblePlates[0]
extractedPlates.append(licPlate.imgPlate)
numberPlateList.append(BR_IMG.rsplit('.', 1)[0])
cv2.imwrite(EXTRACTED_PLATES_DIR + BR_IMG, licPlate.imgPlate)
# Export list of number plate to csv
WriteNumberPlateList(numberPlateList)
print("Completed pre-processing and export list of number plates.")
# Pre-process extracted plates and export processed image file
print("Begin Image Pre-processing for OCR..")
ocr_preprocessed_plates = OCR.preprocess(extractedPlates, numberPlateList,
PREPROCESSED_PLATES_DIR)
# OCR the preprocessed extracted plates and export the OCR Output to csv
OCROutput = OCR.OCR(ocr_preprocessed_plates)
print("Completed Image Pre-processing for OCR and export to csv.")
# Accuracy result based on predicted number plate vs actual number plates
OCR_Result = OCR.CheckOCR(OCROutput, numberPlateList)
# Export Result to csv
ExportActualAndPredictedNumberPlate(numberPlateList, OCROutput)
return # end of main()
def setUp(self):
base_dir = os.path.abspath(os.getcwd())
with open('det_args.json') as f:
self.det_args = json.load(f)
path = os.environ.get("DET_MODEL_PATH")
self.det_args["trained_model"] = os.path.join(base_dir, path)
self.det_net = ocr.init_detection_model(self.det_args)
with open('rec_args.json') as f:
self.rec_args = json.load(f)
path = os.environ.get("REC_MODEL_PATH")
self.rec_args["trained_model"] = os.path.join(base_dir, path)
self.rec_net = ocr.init_rec_model(self.rec_args)
def index():
if request.method == 'POST':
image_file = request.files['file']
if image_file:
filepath = os.path.join('uploads', image_file.filename)
image_file.save(filepath)
language = detect_language(filepath)
result_file_path = 'recognized/' + image_file.filename + '.txt'
OCR.recognizeFile(filepath=filepath, resultFilePath=result_file_path,
language=language, outputFormat='txt')
return translate(filepath=result_file_path), 500
return render_template('main/index.html')
def toPDF():
result = request.get_json()
requires = ["img"]
if isValid(requires, result):
info = result["img"]
if info[1:21] == "data:application/pdf":
type = "PDF"
else:
type = "IMG"
if type == "PDF":
return OCR.PDFFromBase64(info[info.find(',') + 1:])
elif type == "IMG":
return OCR.PDFFromBase64(info[info.find(',') + 1:])
def payment_solver(self, selected_link, sensitive=False):
self.wait_for("Basarılı mı?",
f'assets/{selected_link}/success',
scrolldown_enabled=True,
sensitive=sensitive)
offset = self.config["payment_solver"][selected_link]["region_offset"]
pos = self.find(f'assets/{selected_link}/resolve_payment',
sensitive=sensitive)
if pos:
pos = list(pos)
pos[0] += offset[0]
pos[1] += offset[1]
pos[2] = offset[2]
pos[3] = offset[3]
save_path = f"payments/{selected_link}"
if not os.path.isdir(save_path):
os.makedirs(save_path, exist_ok=True)
payment_solver_text = OCR.resolve_region(
pos,
show=True,
save=
f"{save_path}/{str(datetime.datetime.now()).split('.')[0].replace(':','-')}.png"
)
print("Payment Image Pos:", pos)
payment_amount = re.match("\d+", payment_solver_text)
if payment_amount:
with open(f"payments/{selected_link}.txt", 'a+') as f:
str_log = f"#Time: {str(datetime.datetime.now()).split('.')[0]}, #Payment: {payment_amount[0]}"
print(str_log)
f.write(str_log + "\n")
else:
print("Could not detect payment solver image")
def main():
#Clone testimages from cloud storage
subprocess.call("gsutil cp -r -n gs://smartreviewdata/testimages .",
shell=True)
# Run Object Detection and create image parts
print("\nObject Detection Running and Exporting into imageparts...")
OD.predict_boxes()
#Push xmls to cloud storage
print("\nPushing XMLs to cloud storage ...")
subprocess.call("gsutil mv testxmls/* gs://smartreviewdata/testxmls/",
shell=True)
# Calculate Rows data using OCR
print("\nRows being calculated for the imageparts...")
sno = database.query(
'SELECT count(*) from `demo.{}`'.format(table_id))[0][0]
ROWS = OCR.calculate_data(sno,
'imageparts',
train_flag='False',
source='testimages')
print("\nMatching with BIGQUERY records and pushing...")
match_query(ROWS)
#Push imageparts to cloud storage
print("\nPushing imageparts to global_imageparts ...")
subprocess.call(
"gsutil mv imageparts/* gs://smartreviewdata/global_imageparts/",
shell=True)
#Clear testimages
subprocess.call("rm -r testimages", shell=True)
def upload_file():
# check if the post request has the file part
if 'file' not in request.files:
resp = jsonify({'message': 'No file part in the request'})
resp.status_code = 400
return resp
file = request.files['file']
if file.filename == '':
resp = jsonify({'message': 'No file selected for uploading'})
resp.status_code = 400
return resp
if file and allowed_file(file.filename):
filename = secure_filename(file.filename)
file_path = os.path.join(app.config['UPLOAD_FOLDER'], filename)
file.save(file_path)
resp = jsonify({'message': 'File successfully uploaded'})
dpi = 300
documentText, fname = OCR.Convert(file_path, dpi, str(1))
data = Extract.Info(fname)
data = Align.restructure(data)
return str(data)
else:
resp = jsonify({'message': 'Allowed file type is pdf'})
resp.status_code = 400
return resp
def get(self):
url = request.args.get('url')
token = request.args.get('token')
state = request.args.get('state')
url = url + '&token=' + token
url_response = urllib.request.urlopen(url)
img_array = np.array(bytearray(url_response.read()), dtype=np.uint8)
plateImg = cv2.imdecode(img_array, -1)
height, width = plateImg.shape[:2]
warped = four_point_transform(
plateImg,
np.array([(0, 0), (width, 0), (width, height), (0, height)]))
h, w = warped.shape[:2]
ratio = w / h
height = int(300 / ratio)
#imgPlate = cv2.resize(warped, (300, height))
#cv2.imshow('lol',imgPlate)
#cv2.waitKey(0)
read1, read2 = OCR.readRC(warped)
dic = parseRC.parseToJSON(read1, read2, state)
print(dic)
return jsonify(dic)
def img_convert_str():
global result, img_path
appid_str = appid.get()
apikey_str = apikey.get()
secretkey_str = secretkey.get()
result = OCR.img_to_str(img_path, appid_str, apikey_str, secretkey_str)
result_text.insert(INSERT, result)
def get(self):
url = request.args.get('url')
token = request.args.get('token')
correctorOn = True
CNN_OCR = True
url = url + '&token=' + token
url_response = urllib.request.urlopen(url)
img_array = np.array(bytearray(url_response.read()), dtype=np.uint8)
plateImg = cv2.imdecode(img_array, -1)
height, width = plateImg.shape[:2]
warped = four_point_transform(
plateImg,
np.array([(0, 0), (width, 0), (width, height), (0, height)]))
validChars = Segmentation.startSegment(warped)
print(len(validChars))
plateText = OCR.readPlate(validChars, correctorOn, CNN_OCR)
print(plateText)
if plateText != 'ignore':
return jsonify(plate=plateText)
else:
return jsonify(plate='Try Again!')
def swipe(start, end, t):
global count
cmd = "adb shell input swipe 500 %d 500 %d 2000"%(start, end)
subprocess.Popen(str(cmd), shell=True)
time.sleep(3)
for i in range(6):
click(600, 445 + i * 270, 2+random.random())
click(540, 550, 4+random.random())
count = count + 1
screenshot(r"./picture/", count)
time.sleep(2+random.random())
OCR.ocr_function(("picture/%d.png"%(count)), count)
back()
time.sleep(1+random.random())
back()
time.sleep(1+random.random())
time.sleep(1+random.random())
def rowdownmid(team):
time.sleep(5)
T = time.clock()
# assume loaded
# click in screen
BTcoords = None
if (team == "blue"):
while BTcoords is None:
BTcoords = openCVLocate.locateCenter(imgpath + '/blueTeam.PNG')
else:
BTcoords = None
clickInGame(BTcoords)
# press y to lock camera
keyHelper.PandRKey(keyHelper.cToHex('y'))
# buy dorans blade
screenshotArray = np.array(ImageGrab.grab().convert('RGB'))
buyUpgradeAndBack.buy(buyorder, screenshotArray, BTcoords)
cloop = 0
lastHealth = 0
time.sleep(40)
# main loop
# move back on big damage
# remove slow move
# make the loop[ faster for back and forth
# buy on back
# back on low health
# level all abilities
# use all abilities in a forward direction
while True:
cloop += 1
ctime = time.clock()
screenshotArray = np.array(ImageGrab.grab().convert('RGB'))
try:
health = int(OCR.getWhiteText(screenshotArray, math.floor(BTcoords[0] - 545), 26, math.floor(BTcoords[1] + 120), 12, lower=150).replace('.', '').replace(' ', ''))
if (health == 0):
dead(screenshotArray, BTcoords)
continue
elif health < 200:
keyHelper.PandRKey(keyHelper.cToHex('f'))
buyUpgradeAndBack.back(buyorder, screenshotArray, BTcoords)
continue
if(health - lastHealth < -130):
moveBackwards(1, BTcoords)
autoDownMid(BTcoords, time.clock() - T)
if(openCVLocate.locateCenter(imgpath + '/upgrade.PNG') is not None):
buyUpgradeAndBack.upgrade()
if cloop % 10 == 0:
autoDownMid(BTcoords, time.clock() - T)
print(health)
lastHealth = health
except ValueError:
print("health got messed up")
finally:
time.sleep(max(.7 - (time.clock() - ctime), 0))
def welcome():
uid = uuid.uuid4()
f =request.files['file']
file_name = uid.hex+f.filename
f.save(secure_filename(file_name))
img = cv2.imread(file_name,0)
os.remove(file_name)
return OCR.detect(img)
def test_recognition_1(self):
img_name = 'paper-title.jpg'
img = loadImage(img_name)
ocr.text_detection_results = ocr.text_detect(img, self.det_net, self.det_args)
# check that 23 boxes are found in this input image
self.assertEqual(len(ocr.text_detection_results['bboxes']), 23)
# get crops from bounding boxes and input image
bboxes = ocr.text_detection_results['bboxes']
crops = {}
for k, v in bboxes.items():
bbox_coords = v.split(',')
min_x = (int)(bbox_coords[0])
min_y = (int)(bbox_coords[1])
max_x = (int)(bbox_coords[4])
max_y = (int)(bbox_coords[5])
crops[k] = img[min_y:max_y, min_x:max_x, :]
recognition_results = ocr.recognize(crops, self.rec_args, self.rec_net)
self.assertEqual(len(recognition_results), 23)
self.assertEqual(recognition_results['crop0']['result'], 'endtoend')
def main():
plate_no = OCR.predict()
#plate_no = OcrCamera.predict()
print("License plate is :{}".format(plate_no))
result = (searchnp.mysql_searchdata(plate_no))
if result == 1:
ardino.main()
print("gates open")
else:
print("gates will not open")
def crop(boxes, image):
img = image
x, y, w, h = boxes[
0] #get the coordinates of the part the portion where the number plate is found
crop_img = img[y:y + h, x:x + w] #cropping the image
# cv2.imwrite('crop.jpg',crop_img)
x = OCR.get_number(
crop_img
) #sending the image for extraction of number using OCR from OCR space api
return x
def main():
# Run Object Detection and create image parts
print("Object Detection Running and Exporting into imageparts...")
OD.predict_boxes()
# Calculate Rows data using OCR
print("Rows being calculated for the imageparts...")
ROWS = OCR.calculate_data(36, 'imageparts')
print("Matching with BIGQUERY records and pushing...")
match_query(ROWS)
def chooseLesson(lesson):
conn = httplib.HTTPConnection(host)
# get captcha
conn.request("GET", "/xk/input.jsp", None, headers)
res = conn.getresponse()
parser = MyHTMLParser()
parser.feed(res.read())
token = parser.get_token()
if token is None:
print "have bugs in my programme"
return False
conn.request("GET", "/xk/image.do?token={}".format(token), None, headers)
res = conn.getresponse()
if res.getheader("Set-Cookie") is not None:
headers["Cookie"] = res.getheader("Set-Cookie").split(";")[0]
with open("captcha.jpg", "wb") as F:
F.write(res.read())
# OCR
OCRHdl = OCR.CaptchaHandler()
rand = OCRHdl.captcha_from_image("captcha.jpg")
# POST
data = urllib.urlencode({"token": token,
"selectionId": lesson,
"xklb": "ss",
"rand": rand})
try:
time.sleep(3)
conn.request("POST", "/xk/doSelectServlet", data, headers)
res = conn.getresponse()
F = open("out.html", "w")
F.write(res.read())
F.close()
print u"返回结果写在out.html中，请查看"
except Exception as e:
print u"请求失败"
return False
# check if lesson is selected
conn.request("GET", "/xk/courseTableServlet", None, headers)
res = conn.getresponse()
parser.set_lesson(lesson)
parser.feed(res.read())
conn.close()
return parser.has_lesson()
def sendKeyword(self):
inputwords = self.leKeyword.text()
self.pbOCR.setValue(self.__i)
# to ocrProgress
self.stackedWidget.setCurrentIndex(6)
QApplication.processEvents()
self.__Ocr = OCR.CoverCheck(self.__DC.IMG_path, self.__DC.Crop_path, self.__DC.Improvement_path, inputwords)
self.__Ocr.setOnstepListener(self)
crop_list = self.__Ocr.crop(self.__coverList)
improvement_list = self.__Ocr.improve(crop_list)
detail_list = self.__Ocr.comparison(improvement_list)
self.__DC.CROP_clear()
self.__DC.IMPROVEMENT_clear()
self.detailResult(detail_list)
self.stackedWidget.setCurrentIndex(7)
def click_l(self, count=1, close_after_click=True):
'''
if new tab is opened return [True] or return [False]
'''
self.mouse.click(Mouse.Button.left, count)
time.sleep(2)
if not close_after_click:
return False
# Check URL
if OCR.check_current_url():
time.sleep(1)
self.go_to_top()
return True
return False
async def image_translate(ctx, lang):
# получение отправителя команды
author = ctx.message.author
if lang.lower() == "ангглийский":
await ctx.send(
f"{author.mention}, зачем переводить на один и тот же язык?")
return
# обработка ошибок
try:
# доставание url из вложения
url = str(ctx.message.attachments[0])
url = url[url.find("url") + 5:url.rfind("'")]
# если ошибка
except:
# отправка сообщения ботом
await ctx.send(f"{author.mention}, извините, я не могу найти картинку")
# если нет ошибок
else:
# запрос к функции перевода изображения в текст
text = OCR.image_to_text(url)
if not text:
await ctx.send(
f"{author.mention}, извините, я не смог найти текста на английском на картинке"
)
else:
print(f"{author} использует команду image_translate",
datetime.datetime.now())
# отправка текста в функции перевода с автоопределением языка
await auto_translate(ctx, lang, text)
def refresh_coordinates_thread():
time.sleep(2)
Upload.new_ship(0, 0, 0, settings.Variables.layerName,
settings.Variables.layerID)
nextGetTime = 0
while True:
if time.time() > nextGetTime:
nextGetTime = time.time() + 0.2
threading.Thread(target=OCR.refresh_coordinates()).start()
if OCR.Coordinates.success:
OCR.Coordinates.success = False
# noinspection SpellCheckingInspection
threading.Thread(target=Upload.move_ship,
args=(OCR.Coordinates.x, OCR.Coordinates.y,
OCR.Coordinates.z,
settings.Variables.layerName,
settings.Variables.layerID)).start()
coordinates_display = ""
coordinates_display = coordinates_display + "X: " + str(
Upload.Vars.latestOnlineX)
for i in range(0, 10 - len(str(Upload.Vars.latestOnlineX))):
coordinates_display = coordinates_display + " "
coordinates_display = coordinates_display + "\n"
coordinates_display = coordinates_display + "Y: " + str(
Upload.Vars.latestOnlineY)
for i in range(0, 10 - len(str(Upload.Vars.latestOnlineY))):
coordinates_display = coordinates_display + " "
coordinates_display = coordinates_display + "\n"
coordinates_display = coordinates_display + "Z: " + str(
Upload.Vars.latestOnlineZ)
for i in range(0, 10 - len(str(Upload.Vars.latestOnlineZ))):
coordinates_display = coordinates_display + " "
Graphics.GUI.set_online_coordinates(coordinates_display)
def main():
correctorOn = True
oldSeg = False
CNN_OCR = True
plateImg = cv2.imread('./data/input/test5.png')
height, width = plateImg.shape[:2]
warped = four_point_transform(
plateImg, np.array([(0, 0), (width, 0), (width, height), (0, height)]))
if oldSeg:
validChars = OldSegmentation.segment(plateImg)
else:
validChars = Segmentation.startSegment(plateImg)
print(str(len(validChars)))
plateText = OCR.readPlate(validChars, correctorOn, CNN_OCR)
if plateText != 'ignore':
print(plateText)
else:
print('Try Again!')
def login():
conn = httplib.HTTPConnection(host)
conn.request("GET", "/xk")
res = conn.getresponse()
res.read()
conn.request("GET", "/xk/image.do")
res = conn.getresponse()
if res.getheader("Set-Cookie") is not None:
headers['Cookie'] = res.getheader("Set-Cookie").split(";")[0]
with open("captcha.jpg", "wb") as F:
F.write(res.read())
# OCR
OCRHdl = OCR.CaptchaHandler()
rand = OCRHdl.captcha_from_image("captcha.jpg")
# POST form
data = urllib.urlencode({
"studentId": studentId,
"password": password,
"rand": rand,
"Submit2": r"提交"
})
try:
time.sleep(0.2)
conn.request("POST", "/xk/loginServlet", data, headers)
res = conn.getresponse()
F = open("out.html", "w")
F.write(res.read())
F.close()
except Exception as e:
print 'Exception'
return False
conn.close()
result = res.reason == "Found"
if result:
OCRHdl.save_attribute_codes([i for i in rand])
return result
def read(self):
self.view.t1.delete(0,len(self.view.t1.get()))
self.view.t2.delete(0,len(self.view.t2.get()))
self.view.t3.delete(0,len(self.view.t3.get()))
self.view.t4.delete(0,len(self.view.t4.get()))
self.view.t5.delete(0,len(self.view.t5.get()))
self.view.t6.delete(0,len(self.view.t6.get()))
self.view.t7.delete(0,len(self.view.t7.get()))
self.view.t8.delete(0,len(self.view.t8.get()))
self.view.t9.delete(0,len(self.view.t9.get()))
file_name = tk.filedialog.askopenfilename()
OCR.OCR(file_name)
data = NLP.NLP()
self.view.t1.insert(0,data[3])
self.view.t2.insert(0,data[4])
self.view.t3.insert(0,data[0])
self.view.t4.insert(0,data[5])
self.view.t5.insert(0,data[7])
self.view.t6.insert(0,data[1])
self.view.t7.insert(0,data[2])
self.view.t8.insert(0,data[6])
self.view.t9.insert(0,data[8])
def get_patient_info(diff_imgs):
'''Gets a unique hash of a patient name, the date that the x-ray was taken
and the gender of the patient.
Arguments:
diff_imgs: Array of multiple copies of the same x-ray, each copy has a different type
of image processing applied to it to affect the text in the image differently.
Returns:
patient_id: Unique hash of the patient name.
xray_datetime: Date and time that the x-ray was taken.
gender: Sex of the patient [M, F].
'''
patient_names = []
patient_births = []
xray_datetimes = []
genders = []
for img in diff_imgs:
# run OCR
patient_name, patient_birth, xray_datetime, gender = ocr.get_text_from_tesseract(
img)
patient_names += [patient_name]
patient_births += [patient_birth]
xray_datetimes += [xray_datetime]
genders += [gender]
# get the most likely string from all images
patient_name = get_popular_string(patient_names)
patient_birth = get_popular_string(patient_births)
translator = str.maketrans('', '', string.punctuation)
patient_birth = str(patient_birth).translate(translator)
xray_datetime = get_popular_string(xray_datetimes)
gender = get_popular_string(genders)
# encrypt patient name
to_encrypt = patient_name + patient_birth
to_encrypt = to_encrypt.encode('utf-8')
patient_id = hashlib.sha256(to_encrypt).hexdigest()[:20]
return patient_id, xray_datetime, gender
def api_root():
if request.method == "GET":
vidid = request.args.get('id').strip('https://www.youtube.com/embed/')
seconds = request.args.get('time')
url = request.args.get('id').replace('embed/', 'watch?v=')
fname = vidid + ".mp4"
folder = "./videos"
path = os.path.join(folder, fname)
print(url, path, seconds)
resp = jsonify(ocr.process_time(url, path, seconds))
print(resp)
else:
if request.headers['Content-Type'] != 'application/json':
return "Please post a JSON"
data = json.loads(json.dumps(request.json))
vidid = data['id'].strip('https://www.youtube.com/embed/')
url = data['id'].replace('embed/', 'watch?v=')
# TODO check if in DB
fname = vidid + ".mp4"
folder = "./videos"
path = os.path.join(folder, fname)
if not os.path.isfile(path):
vp.downloadYt(url, path)
resp = jsonify({})
return resp
def startGUI():
cap = capture.Capture(0)
print('before run')
cap.start()
t = time.time()
while cap.img is None:
print('waiting for camera to connect',time.time()-t)
time.sleep(.25)
proc = preprocess.Processor()
proc.start(cap)
while proc.img is None:
print('waiting for processed image',time.time()-t)
time.sleep(.25)
ocr = OCR.OCR()
ocr.start(proc)
while ocr.txt is None:
print('waiting for OCR',time.time()-t)
time.sleep(.25)
root.cap = cap
root.proc = proc
root.mainloop()
#!/usr/bin/env python
import OCR
OCR.parse_args(__file__)
prog = OCR.OCR_prog()
blocks = 2 #4
xes_per_blk = 4 #8
db_size = 1024
num_edts = blocks * xes_per_blk * 1 #0
main_dbs = [OCR.DB("db_" + str(n), db_size) for n in range(0, num_edts)]
child_edts = [OCR.EDT("child" + str(i)) for i in range(0, xes_per_blk)]
child_joiner = OCR.EDT("child_join", deps=child_edts)
comp_edts = [
OCR.EDT("comp_" + str(n),
deps=[main_dbs[n]],
scope=[child_joiner, child_edts]) for n in range(0, num_edts)
]
joiner_edt = OCR.EDT("joiner", deps=comp_edts, finish=True)
main_edt = OCR.EDT("main", scope=[main_dbs, comp_edts, joiner_edt])
print prog.to_str()
y += 3
if key & 0xFF == ord('d'):
x += 3
if key & 0xFF == ord('a'):
s -= 3
if key & 0xFF == ord('q'):
break
if key & 0xFF == ord('t'):
t += 2
if key & 0xFF == ord('g'):
t -= 2
if key & 0xFF == ord('y'):
thresh = -thresh
if key & 0xFF == ord('p'):
cv2.imwrite('capture.png', croppedTop)
textTop = OCR.giveMeText()
print "Read Text From Top of Card:" + textTop
mostSimilarNameTop, errorDistanceTop = findMostSimilar(textTop, nameList)
print "Most Similar :" + mostSimilarNameTop + ", with error distance :" + str(errorDistanceTop)
cv2.imwrite('capture.png', croppedBottom)
textBottom = OCR.giveMeText()
print "Read Text From Bottom of Card:" + textBottom
mostSimilarNameBottom, errorDistanceBottom = findMostSimilar(textBottom, nameList)
print "Most Similar :" + mostSimilarNameBottom + ", with error distance :" + str(errorDistanceBottom)
if (mostSimilarNameTop != ''):
if (mostSimilarNameBottom != ''):
if (errorDistanceTop <= errorDistanceBottom):
print "Top is closer to a real magic card name"
else:
print "Bottom is closer to a real magic card name"
dataset[contador][8]=float(dataset[contador][8])
dataset[contador][9]=float(dataset[contador][9])
dataset[contador][10]=float(dataset[contador][10])
dataset[contador][11]=float(dataset[contador][11])
dataset[contador][12]=float(dataset[contador][12])
dataset[contador][13]=float(dataset[contador][13])
dataset[contador][14]=float(dataset[contador][14])
dataset[contador][15]=float(dataset[contador][15])
dataset[contador][16]=float(dataset[contador][16])
contador+=1
Matrix=dataset
[fila,columna]=img.shape
dato1=OCR.Relacion(img)
dato2=OCR.Area(img)
[dato3,negro1]=OCR.lineaVertical(fila,columna,img)
[dato4,negro2]=OCR.lineaHorizontal(fila,columna,img)
[x1,x2,x3,x4]=OCR.lineasVerticalesyHorizontales2(img)
dato5=OCR.lineasHorizontales5(img)
dato6=OCR.lineasVerticales5(img)
dato7=OCR.lineasHorizontales3(img)
dato8=OCR.lineasVerticales3(img)
dato9=x1+x2 #cuantos "1" encontre en los dos cortes
dato10=x3+x4# cuantos cambios de valor encontre en los dos cortes
dato11=negro1
dato12=negro2
dato13=(dato3+dato4)/fila # relacion entre los cambios quedetecte en una cruz en la imagen entre las filas
[dato14,dato15]=OCR.pixelesNegrosYBlancos(fila,columna,img)
prediccion=[dato1,dato2,dato3,dato4,dato5,dato6,dato7,dato8,dato9,dato10,dato11,dato12,dato13,dato14,dato15]
def process(self, image):
scan = OCR.recognize(image)
self.code.add_scan(scan)
return self.code
print("[6]Correcting defects")
CONTOUR, HIER = cv2.findContours(RES, cv2.RETR_LIST, cv2.CHAIN_APPROX_SIMPLE)
POINTS = []
for CNT in CONTOUR:
mom = cv2.moments(CNT)
(x, y) = int(mom['m10']/mom['m00']), int(mom['m01']/mom['m00'])
cv2.circle(IMG, (x, y), 4, (0, 255, 0), -1)
POINTS.append((x, y))
if len(POINTS) != 100:
print "[!]Centroids: " + str(len(POINTS))
print "[!]Exiting"
sys.exit()
C_POINTS = [sorted(i, key = lambda x: x[1]) for i in OCR.split_len(sorted(POINTS), 10)]
R_POINTS = [ list(i) for i in zip(*C_POINTS)]
R_POINTS = [x for sublist in R_POINTS for x in sublist]
###OCR Stage
print "[7]OCR Stage: "
RESULT = [[0]*9 for i in range(9)]
for row in range(9):
print "\t[+]Row: " + str(row + 1)
for column in range(9):
## top left corner = x1:y1 | down right corner = x2:y2
x1, x2, y1, y2 = OCR.getcorners((row, column), R_POINTS)
contour, hier = cv2.findContours(res,cv2.RETR_LIST,cv2.CHAIN_APPROX_SIMPLE)
##create a list with the coordinates sorted
centroids = []
for cnt in contour:
mom = cv2.moments(cnt)
(x,y) = int(mom['m10']/mom['m00']), int(mom['m01']/mom['m00'])
centroids.append((x,y))
if len(centroids) != 100:
print "Centroids: " + str(len(centroids))
exit()
c_points = [sorted(i, key = lambda x: x[1]) for i in OCR.split_len(sorted(centroids),10)]
r_points = [ list(i) for i in zip(*c_points)]
r_points = [x for sublist in r_points for x in sublist]
###OCR Stage
print "7) OCR Stage: "
result = [[0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0, 0, 0]]
for row in range(9):
print "\tRow: " + str(row+1)
for column in range(9):
## top left corner = x1:y1 | down right corner = x2:y2
###Final Stage###
###
result=""
size=630
size=(size,)*2
number=size[0]/9
print("OCR Stage")
import cv2.cv as cv
import OCR
for i,j in enumerate(b):
ri = i/10
ci = i%10
if ci != 9 and ri!=9:
src = bm[ri:ri+2, ci:ci+2 , :].reshape((4,2))
dst = np.array( [ [ci*number,ri*number],[(ci+1)*number-1,ri*number],[ci*number,(ri+1)*number-1],[(ci+1)*number-1,(ri+1)*number-1] ], np.float32)
retval = cv2.getPerspectiveTransform(src,dst)
warp = cv2.warpPerspective(res2,retval,(size))
whole_digit= warp[ri*number:(ri+1)*number-1 , ci*number:(ci+1)*number-1].copy()
noborder_digit= whole_digit[5:64,5:64]
gray_digit = cv2.cvtColor(noborder_digit,cv2.COLOR_BGR2GRAY)
thresh_digit = cv2.adaptiveThreshold(gray_digit,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,0,151,80)
image = cv.CreateImageHeader((thresh_digit.shape[1],thresh_digit.shape[0]), cv.IPL_DEPTH_8U, 1)
cv.SetData(image, thresh_digit.tostring(), thresh_digit.dtype.itemsize*thresh_digit.shape[1])
result+=OCR.ocr_singledigit(image)
rows = OCR.split_len(result,9)
for row in rows:
print row
print time.time()-st
for row in range(9):
print "Row:", row
for column in range(9):
src = bm[row:row+2, column:column+2 , :].reshape((4,2))
dst = np.array( [ [column*sqaure_size,row*sqaure_size],[(column+1)*sqaure_size-1,row*sqaure_size],[column*sqaure_size,(row+1)*sqaure_size-1],[(column+1)*sqaure_size-1,(row+1)*sqaure_size-1] ], np.float32)
retval = cv2.getPerspectiveTransform(src,dst)
warp = cv2.warpPerspective(res2,retval,(size))
whole_digit = warp[row*sqaure_size:(row+1)*sqaure_size-1 , column*sqaure_size:(column+1)*sqaure_size-1].copy()
noborder_digit = whole_digit[5:64,5:64]
gray_digit = cv2.cvtColor(noborder_digit,cv2.COLOR_BGR2GRAY)
thresh_digit = cv2.adaptiveThreshold(gray_digit,255,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,0,151,80)
image = cv2.cv.CreateImageHeader((thresh_digit.shape[1],thresh_digit.shape[0]), cv2.cv.IPL_DEPTH_8U, 1)
cv2.cv.SetData(image, thresh_digit.tostring(), thresh_digit.dtype.itemsize*thresh_digit.shape[1])
result[row][column] = OCR.ocr_singledigit(image)
for row in row:
row_string = ""
for digit in row:
string += digit + " "
print string
print
print "Total time: " + str(time.time()-timer_start)