Hi everyone,
I want to know more about cv2:: calibrateCamera() in python and whether it is using Tsai, or zhang method or …? Any help would be appreciated as I couldn’t find much in the Documentation!
I looked in the code briefly and didn’t find any references to the specific algorithm. I don’t think it uses the Tsai method (I think that requires a 3D calibration target). Also I think it is generally agreed that, while the Tsai algorithm was great for the time, it has long been eclipsed by newer algorithms.
I had thought that OpenCV used the Zhang algorithm, but I recall reading a thread (maybe on this forum) that said it did not use the Zhang method for patent reasons. This is just based on memory and I don’t have any references to back this up. Maybe someone else knows?
-Steve
2 Likes
If you are using a recent version of OpenCV, you are using a method implemented from this paper: OpenCV: Bibliography
See more information from this pull request: https://github.com/opencv/opencv/pull/12772
Before, the implementation was based on:
4 Likes
Thank you for providing those references, Eduardo!
Hi Eduardo Thanks for your answer. Do you have any previous experience with using this method for non-planar objects and the accuracy it could give? I have problems with providing an initial value for its intrinsic Matrix that the method asks for