Topic: Image registration is the process of transforming different sets of data into one coordinate system. Data may be multiple photographs, data from different sensors, times, depths, or viewpoints. It is used in computer vision, medical imaging, military automatic target recognition, and compiling and analyzing images and data from satellites. Registration is necessary in order to be able to compare or integrate the data obtained from these different measurements.


Exercise: Create a Python file and perform the following tasks:

  • Import OpenCV library.
  • Load to images to perform registration.
  • Convert them to grey scale and find their feature points
    – You can implement a feature detection algorithm from scratch (e.g. Harris corner detector) or you can use a built-in algorithm from OpenCV (e.g. ORB).
  • Compute the feature descriptors from both images.
  • Match the features and take the good matches with a ratio test.
  • Calculate the homography matrix with your own algorithm and warp the images using cv2.warpPerspective() function (you must have at least two examples for image mosaicking/stitching and one for subtractive radiography).
  • Show the resulting image.
    Bonus: Blend the images for better results in image mosaicking.

Instructions and Material for solution: Click Figures to download the exercise material.

Lecture for better understanding: https://icarus.csd.auth.gr/image-registration-lecture/

This exercise was developed by I. Karavitsis.

————————————————————————————————————–

For the solutions to the exercises, please contact koroniioanna@csd.auth.gr