A few years ago I struggled a lot to study and understand OpenCV, especially the calibration process.
Now that I have just a little bit more the hang of it, I thought It might be helpful for people that are going to approach this incredible computer vision library starting from zero.
The process of installing OpenCV on some single board computers is not as straightforward as "apt-get install...." and you have no other choice rather than compiling it from zero.
The steps I followed are pretty much the same you find on pyimagsearch and on the Ardupilot Companion PC setup page
Hope it will be of help
Comments
Tiziano: thanks for the kind words! It's all just basic pin hole camera model math. :-) When you find the time to work on your project, please let me know how I can help. I can point you towards source code and answer questions. My scripts understand a couple different flight data log formats and I'm happy to do whatever I can to expand on that. Some of my code has been developed as part of my day job at the University of Minnesota, so as much as possible I have placed it all under the MIT open-source license.
Curt, that video of your is simply amazing! I have to setup something similar for a project of mine I have been working on.
Well done! Thanks for taking the time to present a clear explanation. Camera calibration is the doorway into an very cool world! I have been using opencv + python (in post processing) to add an augmented reality HUD to my flight videos. (ex: https://www.youtube.com/watch?v=1fgVt5OetlI) It's really helpful for visualizing the flight controller performance (and I think a little bit cool.) None of that happens without basic camera calibration first.
Love this! Thanks for posting