__init__.py

from cv2 import (
    imread,
    cvtColor,
    COLOR_BGR2GRAY,
    GaussianBlur,
    threshold,
    THRESH_BINARY,
    )
from pytesseract import image_to_string


def ocr(image, lang='eng', tess_opt='--psm 6'):
    if isinstance(image, str):
        image = imread(image)  # Color
    gray = cvtColor(image, COLOR_BGR2GRAY)  # Grayscale
    blurred = GaussianBlur(gray, (3, 3), 0)  # Smoothing
    ret, thresh = threshold(blurred, 127, 255, THRESH_BINARY)  # Black & white
    height, width, _ = image.shape
    width = int(width*7/10)
    cropped = thresh[0:height, 0:width]  # Buang tanda tangan
    return image_to_string((cropped), lang=lang, config=tess_opt)