Tesseract 无法读取图像中的数字（验证码）

Question

I have this image: Unprocessed Image From the code below, I was able to convert it to this: Processed Image

The image has the number: 8276 But my code reads it as 776

How can I successfully make my code be able to read it as 8276 ? I am very new at this image processing/cv2/pytesseract and upon too much searching was able to get this far.

import cv2
import pytesseract
from PIL import Image

pytesseract.pytesseract.tesseract_cmd = r'C:\Users\hamza.rana\AppData\Local\Tesseract-OCR\tesseract.exe'

image = cv2.imread('captcha.jpg')
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
gray = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY | cv2.THRESH_OTSU)[1]
gray = cv2.medianBlur(gray, 3)
gray = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C, cv2.THRESH_BINARY, 11, 2)
filename = "{}.png".format("temp")
cv2.imwrite(filename, gray)
text = pytesseract.image_to_string(Image.open('temp.png'),config='--psm 13 --oem 3 -c tessedit_char_whitelist=0123456789')
print(text)

Answer 1

The out-of-the-box training on Tesseract works best for typefaces, and (in my experience) poorly on hand printing, and forget it on long hand script.

On thing that helps slightly when things get tight is to expand the border by a few pixels. But starting with a messy capcha... That's something you might have to train a model for.