OpenCV offers a fisheye library for calibrating wide-angle lenses. I have implemented the calibration and want to verify the reprojection error by using a new set of images. So is there a function to take a new set of chessboard corners points, undistort them, then distort them, and then compare the initial points with the set of redistorted points. When using cv2.fisheye.distortPoints and cv2.fisheye.undistortPoints I just get a perfect fit so this is not for error measurement. I want to get the reprojection error with new images. One of the reasons is that I want to be able to tell if the camera becomes uncalibrated after usage.