myOCR/base64crnn.py

import base64
import os
import re
from crnn.crnn import crnnOcr
from ocrapi import ocr
import numpy as np
import cv2

def get_image():
    image_list = []
    for i in os.listdir('./result/image'):
        if re.match('(\d+)-(\d+)-(\d+)-(\d+)',i):
            with open(os.path.join('./result/image',i), 'rb') as f:
                base64_data = base64.b64encode(f.read())
                s = base64_data.decode()
                y_min, y_max, x_min, x_max = re.match('(\d+)-(\d+)-(\d+)-(\d+)', i).groups()
                image_list.append({'left':x_min,'top':y_min,'b64':s})

    return image_list

def get_text(online=False):
    image_list = []
    for i in os.listdir('./result/text_img'):
        if re.match('(\d+)-(\d+)-(\d+)-(\d+)',i):
            with open(os.path.join('./result/text_img',i), 'rb') as f:


                if online:
                    s = ocr(os.path.join('./result/text_img',i))
                else:
                    s = crnnOcr(os.path.join('./result/text_img', i))

                y_min, y_max, x_min, x_max = re.match('(\d+)-(\d+)-(\d+)-(\d+)', i).groups()
                image_list.append({'left':x_min,'top':y_min,
                       'w':int(x_max)-int(x_min),
                       'h':int(y_max)-int(y_min),
                       'size':int((int(y_max)-int(y_min))*0.6),
                       's':s})

                # print(i)
                # print({'left':x_min,'top':y_min,
                #        'w':int(x_max)-int(x_min),
                #        'h':int(y_max)-int(y_min),
                #        'size':int((int(y_max)-int(y_min))*0.6),
                #        's':s})

    return image_list




def get_image_html():
    ...


def get_text_html():
    image_list = []
    for i in os.listdir('./result/text_img'):
        if re.match('(\d+)-(\d+)-(\d+)-(\d+)',i):
            with open(os.path.join('./result/text_img',i), 'rb') as f:


                s = ocr(os.path.join('./result/text_img',i))
                y_min, y_max, x_min, x_max = re.match('(\d+)-(\d+)-(\d+)-(\d+)', i).groups()
                image_list.append({'left':x_min,'top':y_min,
                       'w':int(x_max)-int(x_min),
                       'h':int(y_max)-int(y_min),
                       'size':int((int(y_max)-int(y_min))*0.6),
                       's':s})



import json
def base64ocr(base64_str):
    imgString = base64.b64decode(base64_str)
    array = np.fromstring(imgString, np.uint8)
    image = cv2.imdecode(array, cv2.IMREAD_COLOR)
    cv2.imshow('1',image)
    cv2.waitKey()
    string = crnnOcr(image)

    return string

if __name__ == '__main__':


    import pandas as pd
    data = pd.read_csv(r'F:\exam_segment_django_0330\x11.csv',encoding='utf8')
    for i in data['base64']:
        try:
            print(base64ocr(i))
        except Exception as e:
            print(e)

    # with open(r'F:\exam_segment_django_0330\2.txt') as f:
    #     for i in f:
    #         print(base64ocr(i))