I'm trying to get the data out of this image: and no matter what I try I can't get a good result. I have tried ImageEnhance and cv2 I got the most promising result using cv2 and adaptive Treshold: gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY) inverted_image = cv2.bitwise_not(gray) #_, thresh = cv2.threshold(inverted_image ,95, 255, cv2.THRESH_BINARY) Test = cv2.adaptiveThreshold(inverted_image,120,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,cv2.THRESH_BINARY,11,2) Which gives me this picture: But tesseract doesn't seem to be able to "read it" I get the following output: text = pytesseract.image_to_string((Test), lang='fra') 845 Vitalité 75 lnteligence + 35 Sagesse 1Poñge 15 Dommages Feu AaTace à Teva eM text = pytesseract.image_to_string((Test), lang='eng') Extracted Text: 345 Vitalité ‘75 Inteligence ~ ‘35 Sagesse “onée. 1S Dommages Feu WBTaole 2s 7 Reve OM. Doesn't look much better. I also tried bluring it with cv2.GaussianBlur(image, (3, 3), 0) but that just makes it unreadable. I am missing something or doing something wrong? Whole project for better understanding: I want to create a kind of bot who reads the information on a part of my screen does some calculations and does a simple action. There exist around 40 different words that can be in the screenshot. Would it be a solution to create a Database with all words and do a pixel comparison? But how would I manage the numbers? I don't want to save all the numbers from 1-500. Any help is appreciated! Continue reading...
