imageboxer.py

import re
import cv2
import pytesseract
from pytesseract import Output

class imageboxer:
    
    # Main image conversion process
    def box_image(self, image):
        output_data = pytesseract.image_to_data(image, output_type=Output.DICT)
        n_boxes = len(output_data['text'])
        for i in range(n_boxes):
            if int(output_data['conf'][i]) > 60:
                (x, y, w, h) = (output_data['left'][i], output_data['top'][i], output_data['width'][i], output_data['height'][i])
                image = cv2.rectangle(image, (x, y), (x + w, y + h), (0, 255, 0), 2)
        return image

    # Only box dates
    def box_image_date(self, image):
        # Format for a date
        date_pattern = '^(0[1-9]|[12][0-9]|3[01])/(0[1-9]|1[012])/(19|20)\d\d$'

        output_data = pytesseract.image_to_data(image, output_type=Output.DICT)
        n_boxes = len(output_data['text'])
        for i in range(n_boxes):
            if int(output_data['conf'][i]) > 60:
                if re.match(date_pattern, output_data['text'][i]):
                    (x, y, w, h) = (output_data['left'][i], output_data['top'][i], output_data['width'][i], output_data['height'][i])
                    image = cv2.rectangle(image, (x, y), (x + w, y + h), (0, 255, 0), 2)
        return image