Image registration is the process of transforming different images of one scene into the same coordinate system. You can find more about it in this wonderful article.
We will use AKAZE local features from OpenCV 2D Features framework to detect and match keypoints on two images. These matching keypoints will allow us to compute the homography matrix and warp the second image so that we will make a panorama.
Original images
We detect keypoints and compute descriptors using AKAZE algorithm
Use brute-force matcher to find 2-nn matches and select only the best of them by comparing distances between the first two matches
Compute the homography matrix and warp the second image. Combine the images to make a panorama.
