Here, you'll get a chance to try camera calibration and distortion correction for yourself!

There are two main steps to this process: use chessboard images to obtain image points and object points, and then use the OpenCV functions cv2.calibrateCamera() and cv2.undistort() to compute the calibration and undistortion.

Unfortunately, we can't perform the extraction of object points and image points in the browser quiz editor, so we provide these for you in the quiz below.

Try computing the calibration and undistortion in the exercise below, and if you want to play with extracting object points and image points yourself, fork the Jupyter notebook and images in this repository . You can also download the distortion pickle file and test image used in the below quiz if you'd like to run the below code on your own machine.

If you run into any errors as you run your code, please refer to the Examples of Useful Code section in the previous video and make sure that your code syntax matches up!

Distortion Correction

Start Quiz:

distort_correct.py

import pickle
import cv2
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.image as mpimg

# Read in the saved objpoints and imgpoints
dist_pickle = pickle.load( open( "wide_dist_pickle.p", "rb" ) )
objpoints = dist_pickle["objpoints"]
imgpoints = dist_pickle["imgpoints"]

# Read in an image
img = cv2.imread('test_image.png')

# TODO: Write a function that takes an image, object points, and image points
# performs the camera calibration, image distortion correction and 
# returns the undistorted image
def cal_undistort(img, objpoints, imgpoints):
    # Use cv2.calibrateCamera() and cv2.undistort()
    undist = np.copy(img)  # Delete this line
    return undist

undistorted = cal_undistort(img, objpoints, imgpoints)

f, (ax1, ax2) = plt.subplots(1, 2, figsize=(24, 9))
f.tight_layout()
ax1.imshow(img)
ax1.set_title('Original Image', fontsize=50)
ax2.imshow(undistorted)
ax2.set_title('Undistorted Image', fontsize=50)
plt.subplots_adjust(left=0., right=1, top=0.9, bottom=0.)

Here is one example answer:

def cal_undistort(img, objpoints, imgpoints):
    ret, mtx, dist, rvecs, tvecs = cv2.calibrateCamera(objpoints, imgpoints, img.shape[1:], None, None)
    undist = cv2.undistort(img, mtx, dist, None, mtx)
    return undist

Note that the quiz is loading this image as channels first, so we use `img.shape[1:]` to get the height and width. This should be adjusted as necessary, such as is seen on the previous page for a grayscale image.

Challenge: Calibrate your own camera!

If you're up for a challenge, go ahead and try these steps on your own camera images. Just print out a chessboard pattern , stick it to a flat surface, and take 20 or 30 pictures of it. Make sure to take pictures of the pattern over the entire field of view of your camera, particularly near the edges.

To extract object points and image points you can check out the Jupyter notebook in this repository . If you're following along with the code in the notebook, be sure you set the right size for your chessboard, the link above is to a 9 x 6 pattern, while before we were using an 8 x 6.