HYTOS

    <Compile Include="azure_handwrite_ocr_module.py">

      <SubType>Code</SubType>

    </Compile>

4) azure_handwrite_ocr_module.py

- Azure Computer Vision API �� Handwrite OCR�� ����ϱ� ���� ���

- ���� ������ �ʴ� ���

- azure_ocr_module �� ū ���̴� ������, azure_ocr_module �� ��� ������ �ؽ�Ʈ ������ �ƴ� �ؽ�Ʈ ���� ����� ����Ǿ� ����� �ڵ尡 ������

  Delete "$INSTDIR\azure_handwrite_ocr_module.py"

#import httplib

import http.client

import urllib

import base64

import json

import cv2

import numpy as np

import matplotlib.pyplot as plt

import matplotlib.image as mpimg

from PIL import Image

from io import BytesIO

def url_to_image(url):

	# download the image, convert it to a NumPy array, and then read

	# it into OpenCV format

	resp = urllib.request.urlopen(url)

	#resp = urllib.urlopen(url)

	image = np.asarray(bytearray(resp.read()), dtype="uint8")

	image = cv2.imdecode(image, cv2.IMREAD_COLOR)

	# return the image

	return image

def getRollbackRotateCodeByOrientation(orientation):

    rotatedAngle = None

    if orientation == "Left":

        rotatedAngle = cv2.ROTATE_90_COUNTERCLOCKWISE

    elif orientation == "Right":

        rotatedAngle = cv2.ROTATE_90_CLOCKWISE

    else:

        rotatedAngle = cv2.ROTATE_180

    return rotatedAngle

def getRotateCodeByOrientation(orientation):

    rotatedAngle = None

    if orientation == "Left":

        rotatedAngle = cv2.ROTATE_90_CLOCKWISE

    elif orientation == "Right":

        rotatedAngle = cv2.ROTATE_90_COUNTERCLOCKWISE

    else:

        rotatedAngle = cv2.ROTATE_180

    return rotatedAngle

####################################

###############################################

#### Update or verify the following values.  ###

###############################################

# Replace the subscription_key string value with your valid subscription key.

#subscription_key = '4774f949a51e4728b436d05a0d0415c2' # zombie612

#subscription_key = 'ef94d01a99854539a72b5696118b4ac9' # kyusu99

subscription_key = '828669ef78ae40959882ff4a897fb1a4' # master

# Replace or verify the region.

#

# You must use the same region in your REST API call as you used to obtain your

# subscription keys.

# For example, if you obtained your subscription keys from the westus region,

# replace

# "westcentralus" in the URI below with "westus".

#

# NOTE: Free trial subscription keys are generated in the westcentralus region,

# so if you are using

# a free trial subscription key, you should not need to change this region.

#uri_base = 'westcentralus.api.cognitive.microsoft.com' # USA

#uri_base = 'eastasia.api.cognitive.microsoft.com' # east asia for master

uri_base = 'southeastasia.api.cognitive.microsoft.com' # southeast asia for master

headers = {

    # Request headers.

    #'Content-Type': 'application/json',

    'Content-Type': 'application/octet-stream',

    'Ocp-Apim-Subscription-Key': subscription_key,

}

#params = urllib.urlencode({

params = urllib.parse.urlencode({

    # Request parameters.  The language setting "unk" means automatically

    'handwriting': 'false'

})

# The URL of a JPEG image containing text.

#body = "{'url':'D:/Visual Studio Project/PyOpenCv/PyOpenCv/res/symbol1.png'}"

#body =

#"{'url':'https://upload.wikimedia.org/wikipedia/commons/thumb/a/af/Atomist_quote_from_Democritus.png/338px-Atomist_quote_from_Democritus.png'}"

#body = "{'url':'http://cfile9.uf.tistory.com/image/99AFE1505A322DBD114994'}"

def detectAndRemoveText(image):

    tmp = image.copy()

    img = np.array(tmp)

    try:

        tmp = image.copy()

        bin_img = BytesIO()

        tmp.save(bin_img, format='PNG')

        tmp.close()

        img_data = bin_img.getvalue()

        bin_img.close()

        #region REST API with raw file

        # Execute the REST API call and get the response.

        conn = http.client.HTTPSConnection(uri_base)

        #conn = httplib.HTTPSConnection(uri_base)

        conn.request("POST", "/vision/v1.0/recognizeText?%s" % params, img_data, headers)

        response = conn.getresponse()

        data = response.read()

        #endregion

        operationLocation = response.info().get('Operation-Location')

        operationId = operationLocation.split("/")[-1]

        print("oID : " + operationId)

        #operationId = "319d55a2-fe52-482c-a961-f94e81840173"

        isFinished = False

        while isFinished == False:

            conn2 = http.client.HTTPSConnection(uri_base)

            conn2.request("GET", "/vision/v1.0/textOperations/"+operationId+"?%s" % params, "{body}", headers)

            resp = conn2.getresponse()

            data2 = resp.read()

            tempRes = json.loads(data2)

            if tempRes['status'] != 'Succeeded':

                isFinished = True

            else:

                conn2.close()

        #region REST API with url

        ## Execute the REST API call and get the response.

        #conn = httplib.HTTPSConnection(uri_base)

        #conn.request("POST", "/vision/v1.0/ocr?%s" % params, body, headers)

        #response = conn.getresponse()

        #data = response.read()

        #endregion

        # 'data' contains the JSON data.  The following formats the JSON data

        # for display.

        parsed = json.loads(data2)

        #print ("Response:")

        #print (json.dumps(parsed, sort_keys=True, indent=2))

        conn.close()

        conn2.close()

        #chan, w, h = img.shape[::-1]

        w, h = img.shape[::-1]

        orientation = parsed['orientation']

        if orientation != "Up":

            img = cv2.rotate(img, getRotateCodeByOrientation(orientation))

        count = 0

        if parsed['status'] != 'Succeeded':

            print('Is Running')

            return

        for re in cs['lines']:

            #print(re)

            #print(re['words'])

            #print(re['boundingBox'])

            for words in re['words']:

                print(words)

                strArr = words['boundingBox'].split(',')

                pt = (int(strArr[0]), int(strArr[1]))

                ep = (int(strArr[4]), int(strArr[5]))

                width = ep[0]-pt[0]

                height = ep[1]-pt[1]

                #cv2.rectangle(img, pt, (pt[0] + width, pt[1] + height), (255, 255, 255), -1)

                roi = img[pt[1]:pt[1]+height, pt[0]:pt[0]+width]

                temp = roi.copy()

                #_ret, tempBin = cv2.threshold(temp, 250, 255, cv2.THRESH_BINARY_INV) 

                tempBin = cv2.bitwise_not(temp)

                #temp = np.zeros((height, width, 3), np.uint8)

                #temp[::] = (255, 255, 255)

                roi = cv2.bitwise_xor(roi, tempBin, roi)

                #dilate = cv2.dilate(roi, (5, 5))

                #cv2.destroyAllWindows()

                #cv2.imshow('temp', temp)

                #cv2.imshow('tempbin', tempBin)

                #cv2.imshow('ocr', roi)

                ##cv2.imshow('dilate', dilate)

                #cv2.waitKey(0)

                count = count + 1

            #count = count + len(re['words'])

        print("OCR Result count : " + str(count))

        if orientation != "Up":

            img = cv2.rotate(img, getRollbackRotateCodeByOrientation(orientation))

        #resizedRet = cv2.resize(img, None, fx = 0.25, fy = 0.25)

        #cv2.imshow('text removed', resizedRet)

        #cv2.waitKey(0)

        return img

    except Exception as ex:

        print('error occured({}) in {}:{}'.format(ex, sys.exc_info()[-1].tb_frame.f_code.co_filename, sys.exc_info()[-1].tb_lineno))

        return img

def removeText(imagePath):

    image = Image.open(imagePath)

    ret = np.array(image)

    for i in range(4):

        ret = detectAndRemoveText(Image.fromarray(ret))

        ret = cv2.rotate(ret, cv2.ROTATE_90_CLOCKWISE)

        #b, g, r, a = cv2.split(ret)

        #pltSrc = cv2.merge([r, g, b])

        #plt.imshow(pltSrc)

        #plt.show()

        #cv2.waitKey(0)

    ret = detectAndRemoveText(Image.fromarray(ret))

    resizedRet = cv2.resize(ret, None, fx = 0.25, fy = 0.25)

    #cv2.imshow('text removed result', resizedRet)

    #cv2.imwrite("res/p15_2_xor_word.png", ret)

    #cv2.imwrite("res/p10_____text_white_box.png", ret)

    #plt.imshow(ret)

    #plt.show()

    #cv2.waitKey(0)

    image.close()

    return ret

def removeTextFromNpArray(img):

    ret = img.copy()

    for i in range(4):

        ret = detectAndRemoveText(Image.fromarray(ret))

        ret = cv2.rotate(ret, cv2.ROTATE_90_CLOCKWISE)

        #b, g, r, a = cv2.split(ret)

        #pltSrc = cv2.merge([r, g, b])

        #plt.imshow(pltSrc)

        #plt.show()

        #cv2.waitKey(0)

    ret = detectAndRemoveText(Image.fromarray(ret))

    #resizedRet = cv2.resize(ret, None, fx = 0.25, fy = 0.25)

    #cv2.imshow('text removed result', resizedRet)

    #cv2.imwrite("res/result.png", ret)

    #plt.imshow(ret)

    #plt.show()

    #cv2.waitKey(0)

    return ret

def getTextInfoFromNpArray(image):

    try:

        image = cv2.resize(image, None, fx = 2.0, fy = 2.0)

        tmp = Image.fromarray(image)

        bin_img = BytesIO()

        tmp.save(bin_img, format='PNG')

        tmp.close()

        img_data = bin_img.getvalue()

        bin_img.close()

        #region REST API with raw file

        # Execute the REST API call and get the response.

        conn = http.client.HTTPSConnection(uri_base)

        #conn = httplib.HTTPSConnection(uri_base)

        conn.request("POST", "/vision/v1.0/recognizeText?%s" % params, img_data, headers)

        response = conn.getresponse()

        data = response.read()

        #endregion

        operationLocation = response.info().get('Operation-Location')

        operationId = operationLocation.split("/")[-1]

        print("oID : " + operationId)

        #operationId = "319d55a2-fe52-482c-a961-f94e81840173"

        isFinished = False

        while isFinished == False:

            conn2 = http.client.HTTPSConnection(uri_base)

            conn2.request("GET", "/vision/v1.0/textOperations/"+operationId+"?%s" % params, "{body}", headers)

            resp = conn2.getresponse()

            data2 = resp.read()

            tempRes = json.loads(data2)

            if tempRes['status'] == 'Succeeded':

                isFinished = True

            else:

                isFinished = False

                conn2.close()

        #region REST API with url

        ## Execute the REST API call and get the response.

        #conn = httplib.HTTPSConnection(uri_base)

        #conn.request("POST", "/vision/v1.0/ocr?%s" % params, body, headers)

        #response = conn.getresponse()

        #data = response.read()

        #endregion

        # 'data' contains the JSON data.  The following formats the JSON data

        # for display.

        parsed = json.loads(data2)

        #print ("Response:")

        #print (json.dumps(parsed, sort_keys=True, indent=2))

        conn.close()

        conn2.close()

        #chan, w, h = img.shape[::-1]

        chan, w, h = image.shape[::-1]

        #orientation = parsed['orientation']

        #if orientation != "Up":

        #    image = cv2.rotate(image, getRotateCodeByOrientation(orientation))

        count = 0

        if parsed['status'] != 'Succeeded':

            print('Is Running')

            return

        textInfoList = []

        cs = parsed['recognitionResult']

        for re in cs['lines']:

            #print(re)

            #print(re['words'])

            #print(re['boundingBox'])

            #for words in re['words']:

            #    print(words)

            #    text = words['text']

            #    strArr = words['boundingBox']

            #    pt = (int(strArr[0])//2, int(strArr[1])//2)

            #    ep = (int(strArr[4])//2, int(strArr[5])//2)

            #    width = ep[0]-pt[0]

            #    height = ep[1]-pt[1]

            #    textInfoList.append((text, pt, width, height))

            print(re)

            text = re['text']

            strArr = re['boundingBox']

            pt = (int(strArr[0])//2, int(strArr[1])//2)

            ep = (int(strArr[4])//2, int(strArr[5])//2)

            width = ep[0]-pt[0]

            height = ep[1]-pt[1]

            textInfoList.append((text, pt, width, height))

        #if orientation != "Up":

        #    image = cv2.rotate(image, getRollbackRotateCodeByOrientation(orientation))

        return textInfoList

    except Exception as ex:

        print('error occured({}) in {}:{}'.format(ex, sys.exc_info()[-1].tb_frame.f_code.co_filename, sys.exc_info()[-1].tb_lineno))

        return None

    #    image = cv2.resize(image, None, fx=40.0, fy=40.0)

    #    tmp = Image.fromarray(image)

    #    bin_img = BytesIO()

    #    tmp.save(bin_img, format='PNG')

    #    tmp.close()

    #    #cv2.imshow('image', image)

    #    #cv2.waitKey(0)

    #    img_data = bin_img.getvalue()

    #    bin_img.close()

    #    #region REST API with raw file

    #    # Execute the REST API call and get the response.

    #    conn = http.client.HTTPSConnection(uri_base)

    #    #conn = httplib.HTTPSConnection(uri_base)

    #    conn.request("POST", "/vision/v1.0/recognizeText?%s" % params, img_data, headers)

    #    response = conn.getresponse()

    #    data = response.read()

    #    #endregion

    #    #region REST API with url

    #    ## Execute the REST API call and get the response.

    #    #conn = httplib.HTTPSConnection(uri_base)

    #    #conn.request("POST", "/vision/v1.0/ocr?%s" % params, body, headers)

    #    #response = conn.getresponse()

    #    #data = response.read()

    #    #endregion

    #    # 'data' contains the JSON data.  The following formats the JSON data

    #    # for display.

    #    parsed = json.loads(data)

    #    #print ("Response:")

    #    #print (json.dumps(parsed, sort_keys=True, indent=2))

    #    conn.close()

    #    #chan, w, h = img.shape[::-1]

    #    chan, w, h = image.shape[::-1]

    #    orientation = parsed['orientation']

    #    if orientation != "Up":

    #        image = cv2.rotate(image, getRotateCodeByOrientation(orientation))

    #    count = 0

    #    textInfoList = []

    #    for cs in parsed['regions']:

    #        for re in cs['lines']:

    #            #print(re)

    #            #print(re['words'])

    #            #print(re['boundingBox'])

    #            for words in re['words']:

    #                print(words)

    #                text = words['text']

    #                strArr = words['boundingBox'].split(',')

    #                pt = (int(strArr[0])//40, int(strArr[1])//40)

    #                width = int(strArr[2])//40

    #                height = int(strArr[3])//40

    #                textInfoList.append((text, pt, width, height))

    #    if orientation != "Up":

    #        image = cv2.rotate(image, getRollbackRotateCodeByOrientation(orientation))

    #    return textInfoList

    #except Exception as e:

    #    print('Error:')

    #    print(e)

    #    return None