Function get_captcha_fields

recognizer/components/image_processor.py:18–69

(
    img: cv2.typing.MatLike,
)

def get_captcha_fields(
    img: cv2.typing.MatLike,
) -> Tuple[List[bytes], List[Tuple[int, int]]]:
    captcha_fields_with_sizes: List[Tuple[bytes, int, int, int]] = []
    captcha_fields: List[Tuple[bytes, int, int]] = []

    # Turn image to grayscale and Apply a white threshold to it
    gray = cv2.cvtColor(img, cv2.COLOR_RGB2GRAY)
    ret, thresh = cv2.threshold(gray, 254, 255, cv2.CHAIN_APPROX_NONE)
    # Find Countours of the white threshold
    try:
        image, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE)  # type: ignore
    except ValueError:
        contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE)  # type: ignore

    for contour in contours:
        # Checking Image area to only get large captcha areas (no image details)
        area = cv2.contourArea(contour)
        if area > 1000:
            # IDK what this does i copy pasted it :)
            approx = cv2.approxPolyDP(contour, 0.01 * cv2.arcLength(contour, True), True)

            # Calculating x, y, width, height, aspectRatio
            x, y, w, h = cv2.boundingRect(approx)
            aspectRatio = float(w) / h

            if 0.95 <= aspectRatio <= 1.05:
                # Cropping Image area to Captcha Field
                crop_img = img[y:y+h, x:x+w]  # fmt: skip
                # Cv2 to Image Bytes
                image_bytes = cv2.imencode(".jpg", crop_img)[1].tobytes()
                image_size: int = w * h
                # Getting Center of Captcha Field
                center_x, center_y = x + (w // 2), y + (h // 2)
                captcha_fields_with_sizes.append((image_bytes, center_x, center_y, image_size))

    if len(captcha_fields_with_sizes) >= 9:
        # Dont use captcha fields that are too big
        size_median = median([sizes[3] for sizes in captcha_fields_with_sizes])
        for i, (image_bytes, center_x, center_y, image_size) in enumerate(captcha_fields_with_sizes):
            if int(image_size) == int(size_median):
                captcha_fields.append((image_bytes, center_x, center_y))
    else:
        for image_bytes, center_x, center_y, image_size in captcha_fields_with_sizes:
            captcha_fields.append((image_bytes, center_x, center_y))

    sorted_captcha_fields: List[Tuple[bytes, int, int]] = sorted(captcha_fields, key=lambda element: [element[2], element[1]])
    # return sorted_captcha_fields
    return (
        [field[0] for field in sorted_captcha_fields],
        [(field[1], field[2]) for field in sorted_captcha_fields],
    )


def split_image_into_tiles(img: cv2.typing.MatLike, tile_count: int) -> List[bytes]: