Week 5 - Investigating Camera Calibration

Hi there! Welcome to my blog post for the fifth week. It was a fairly short week due to the Labour Day holiday (I hope you enjoyed it). For this week, I focused on researching camera calibration.

Note: It is highly recommended to view this blog post on my personal blog site: https://www.jessexu.me/basis-senior-project/week5/

See it yourself!

If you followed my blog posts before, you know it, here’s the demo that you can try by yourself: https://senior-project.jessexu.me/?file_selection=Week+5

What is camera calibration?

Camera calibration allows us to get the intrinsic parameters, extrinsic parameters, and distortion coefficients of a camera. With these parameters, we may remove camera lens distortion, measure lengths, and more. We usually use a checkerboard pattern or a symmetrical circles pattern to calibrate the camera, since these shapes can be easily recognized by computer algorithms. In my case, I will use a checkerboard pattern for camera calibration because it’s the most used method.

Common-patterns-in-the-camera-calibration-process

Common patterns in camera calibration process (Image source)

If you want to learn more about camera calibration, MathWorks has a decent explanation of it: https://www.mathworks.com/help/vision/ug/camera-calibration.html

Preliminary work I did

I did not do much since it was a short week. Since I don’t have easy access to the microscopic camera that will be used in production and find an appropriately sized checkerboard for the microscopic camera easily, I decided to experiment with my phone’s camera. I printed off a 7*9 checkerboard pattern with each square being 20*20 mm, and took a picture of the checkerboard with my phone. I then put it into OpenCV to see if the checkerboard pattern can be detected. It turns out that the checkerboard got detected perfectly.

Detection result

Checkerboard detection result

Summary

This week, I started investigating on camera calibration and ran a test on recognizing the checkerboard pattern. Next week, I will try to get the parameters of the camera through camera calibration.