How to use pytesseract to read text from this image with simple numbers?

Question

image_processed variable is the attached image.

    custom_config = r'--oem 3 --psm 7 -c tessedit_char_whitelist= 0123456789/'
    result = pytesseract.image_to_string(image_processed, lang='eng', config=custom_config)

The output:

43659 [44 38

The application takes a screenshot of the screen and crops the specified coordinates with the numbers, then applies inverted threshold to get black numbers over a white board. I am trying to read the cropped numbers with pytesseract but it does not output reliable text outputs.

How to use pytesseract to read text from this image with simple numbers?

Answer 1

The main problems of recognizing numbers in this picture are solved by adding 10 pixels of white space above and below the image and setting a black border along the contour, as recommended in the documentation.

Solution using the Pillow library.

import pytesseract
from PIL import Image, ImageOps

pytesseract.pytesseract.tesseract_cmd = r"C:\Program Files\Tesseract-OCR\tesseract.exe"

image_processed = Image.open(r"jiyYb.jpg")

image_processed = ImageOps.expand(image_processed, border=10, fill='#ffffff')
image_processed = ImageOps.expand(image_processed, border=1)

custom_config = r'--psm 7 -c tessedit_char_whitelist=" /0123456789"'
result = pytesseract.image_to_string(image_processed, config=custom_config)
print(result)

-------------

5367 /5438

Solution using opencv-python.

import cv2
import pytesseract

pytesseract.pytesseract.tesseract_cmd = r"C:\Program Files\Tesseract-OCR\tesseract.exe"

image_processed = cv2.imread('jiyYb.jpg')

image_processed = cv2.copyMakeBorder(src=image_processed, top=10, bottom=10, left=0, right=0,
                                     borderType=cv2.BORDER_CONSTANT, value=[255, 255, 255])
image_processed = cv2.copyMakeBorder(src=image_processed, top=1, bottom=1, left=1, right=1,
                                     borderType=cv2.BORDER_CONSTANT)

custom_config = r'--psm 7 -c tessedit_char_whitelist=" /0123456789"'
data = pytesseract.image_to_string(image_processed, config=custom_config)
print(data)

cv2.imshow('image_processed', image_processed)
cv2.waitKey(0)