I wanted to carry out a small personal project that consists of extracting text from an image and translating it into other languages, all using python.But I have problems when I want to extract the texts from the images. For example, I want to extract the text from this image:

Image used as proof

And this is what it throws at me:

yO

DONT You KNOW
WHO I AM? I’M ED
ROTHSTAYLOR!

THE HORRIBLE
GOOD-FOR-NOTHING,
‘ED ROTHSTAYLOR!

bL—) wyscre |
AROUND ME?!

This is the code:

import cv2
import pytesseract


pytesseract.pytesseract.tesseract_cmd = r'C:Program FilesTesseract-OCRtesseract'


img = cv2.imread('webtoon13.png')
gris = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)


print(pytesseract.image_to_string(gris))

Is there any way to improve the result?

Thanks in advance.

I have already tried image smoothing and used the .threshold() method of the cv2 library.

I expected improvements in the result, but no, it simply got worse, it didn’t even detect the text well anymore. The best result I have had so far is converting the image to a grayscale, but it still does not detect the text well (or it simply does not detect all the text).