remove some other colors for ocr, possible python package: imutils numpy opencv-python Pillow pytesseract

py-threshold-image.py

from PIL import Image
import pytesseract
import cv2

from os import listdir
from os.path import isfile, join

if __name__ == "__main__":
    path = r'C:\Users\c\Downloads\ocr'
    onlyfiles = [f for f in listdir(path) if isfile(join(path, f))]
    pytesseract.pytesseract.tesseract_cmd = r'C:\Program Files\Tesseract-OCR\tesseract.exe'
    for f in onlyfiles:
        file = join(path, f)
        img = cv2.imread(file,0)
        #ret , thresh1 = cv2.threshold(img,36,0,cv2.THRESH_BINARY)

        # Run canny edge detection on each channel
        #edges = cv2.Canny(img, 200, 250)

        #new_image = edges.copy()
        ret , new_image = cv2.threshold(img,170,255,cv2.THRESH_BINARY_INV)
        new_image = cv2.blur(new_image, (3, 3))
        #new_image = cv2.adaptiveThreshold(new_image,190,cv2.ADAPTIVE_THRESH_GAUSSIAN_C,cv2.THRESH_BINARY,11,2)
        #new_image = cv2.bitwise_not(new_image);
        file2 = join(path, 'out',f)
        cv2.imwrite(file2, new_image)
        text = pytesseract.image_to_string(new_image)
        print(text)
        #cv2.imshow(f, new_image)
        #cv2.waitKey(0)
        #break;