search

HKUST-Aerial-Robotics / OmniNxt

[IROS'24 Oral] A Fully Open-source and Compact Aerial Robot with Omnidirectional Visual Perception
https://hkust-aerial-robotics.github.io/OmniNxt/
GNU General Public License v3.0
256 stars 13 forks source link

About cylinder undistortion #2

Closed Eunseong17 closed 4 months ago

Eunseong17 commented 5 months ago

Thank you for your excellent work.

In your paper, the cylinder undistortion method references [25], where the distortion parameter is based on Davide Scaramuzza's radial polynomial parameter model (k1, k2, k3, k4). They even use k1 to the focal length in the code(https://plaut.github.io/fisheye_tutorial/#cylindrical-images-to-the-rescue, https://github.com/valeoai/WoodScape/blob/master/scripts/calibration/example.py) .

Does OmniNxt use this parameter for undistortion? if yes, do I need to use MATLAB to obtain these parameters, or can they be applied to the projection: MEI and radtan model?

Peize-Liu commented 4 months ago

Yes, the radtan distortion parameters are involved in fisheye image rectification. "or can they be applied to the projection: MEI and radtan model?" I am not very clear about this question. We calibrate these parameters by utilizing tartan-calib with MEI model. Hope this information helps. We can further discuss via email.

Eunseong17 commented 4 months ago

Thank you for your response.

I have sent an email regarding related questions