Images at two or more correction levels are output as corrected images for a given image, and a user is allowed to select a preferred corrected image. Based on the selection result of the user, an image to be corrected which is different from the given image is corrected.