At present, many blogs have written about using MATLAB to complete the calibration of cameras and lidars, but when doing it, they found that it was different from what was written in the official and other blogs. MATLAB can already realize the whole process of directly outputting external parameters directly from images and point clouds (no need to manually import camera internal parameters).
Unlike all previous versions, this version does not need to leave no blanks as before (of course it is still recommended to leave no blanks), padding can be used on top, bottom, left, and right (as shown in the figure below). And no need to input longer than wide (before 6*8)
Regarding the production of the calibration board, I will not go into details here. You can customize it according to your needs. It is worth noting that the deformation should not be too large.
For how to extract images and point clouds and convert them into pcd format, please refer to:
https://github.com/young147/capture_pcd_img
Compile it in your catkin_ws/src, it is worth noting that in the above code:
Delete the endl after .jpg, otherwise the format error will occur when importing to the computer.
After importing the data, you can choose to manually import the camera internal reference or run it automatically.
Eliminate the outlier, re-calibrate to get the reprojection effect, and export the external parameters
It can be seen that the unaccepted data also has better results:
