如何使用回调 function 避免全局变量

Question

I want to have the label_image() code into a function that is initialized with an image, ie , label_image(image) . Right now, if I do not declare the image as global variable, it does not properly update because of the callback function having image as global vs label_image(image) having a local copy of image . Also, despite the callback function click() being initialized inside label_image() , it cannot see the variable image unless it is declared globally. I guess my problem would be easily solved if I could pass image to click() which then would be able to update the same copy that label_image(image) received. So far, I didn't find anyway to do so as the callback function expects the 5 parameters event, x, y, flags, param ...

def click(event, x, y, flags, param):
        global click_pts, image_storage, image
        if event == cv2.EVENT_LBUTTONDOWN:
            click_pts.append((x, y, 1))
            cv2.circle(image, (x, y), 5, (255, 0, 0), -1)
            image_storage = lshift(image_storage, image)
            cv2.imshow("image", image_storage[-1])
            print('added %(n)s, size %(s)s, type %(t)s' % {'n': (x, y), 's': len(click_pts), 't': 1})
        elif event == cv2.EVENT_RBUTTONDOWN:
            click_pts.append((x, y, 2))
            cv2.circle(image, (x, y), 5, (0, 255, 255), -1)
            image_storage = lshift(image_storage, image)
            cv2.imshow("image", image_storage[-1])
            print('added %(n)s, size %(s)s, type %(t)s' % {'n': (x, y), 's': len(click_pts), 't': 2})


def label_image() -> list:
    global click_pts, image_storage, image
    click_pts = []
    image_storage = np.zeros((10,) + image.shape, np.uint8)
    image_storage[:] = np.ndarray.copy(image)
    cv2.namedWindow('image', cv2.WINDOW_NORMAL)
    cv2.resizeWindow('image', 1400, 1400)
    cv2.setMouseCallback("image", click)

    # Loop until 'q' is pressed
    while True:
        cv2.imshow("image", image_storage[-1])
        key = cv2.waitKey(1) & 0xFF
        if key == ord('q'):
            break
        elif key == ord('b'):
            try:
                print('removed: %(n)s, size %(s)s' % {'n': click_pts[-1], 's': len(click_pts)})
                click_pts.pop()
            except IndexError:
                print('the array is empty')
            image_storage = rshift(image_storage)
            image = np.ndarray.copy(image_storage[-1])
            cv2.imshow("image", image_storage[-1])

    cv2.destroyAllWindows()
    return click_pts

image = cv2.imread('someimage.png')
label_image()

Answer 1

Given that you want to end up calling this:

def click(image, event, x, y, flags, param):

so that we are passing in image as the first parameter, we can make a partial function which has the image parameter on the front already, leaving the callback code to apply the normal event, x, y, flags, param params just as normal.

This means making a partial for the callback:

def label_image():
    # all the usual stuff elided ...

    partial_click = partial(click, image)  # here image is applied first
    cv2.setMouseCallback("image", partial_click)