Homography Estimation

Homography

In order to complete this mapping exercise, you will need to transform the images taken from the robots perspective into a top-down view.

This task is most commonly solved with planar homography and image warping, as described in the videos below:

You will need to create a script to do this.

To find the homography transform, you will need to:

Capture an image from your robot on the calibration page, ensuring the robots axis of rotation is centred above the mark.

Record a set of correspondence points between the pixel locations in an image and real world distances from the robot (in millimetres).

Ensure that the x coordinate for the real world points begins at the centre of rotation, and the y coordinate extends to the left.

HINT: Two useful functions for this include cv2.findHomography() and cv2.warpPerspective().