3D points in lidar frame calculation correct？

[narutojxl]

Hi, when reading the code about calculating 3D points in lidar frame, defined in Corners.cpp, I am confused.The input param retval fo function getCorners(), has been transformed into camera frame with initial guessed transform matrix,

point_cloud = transform(point_cloud,  config.initialTra[0], config.initialTra[1], config.initialTra[2], 
                                      config.initialRot[0], config.initialRot[1], config.initialRot[2]);

But the author finally calculates 8 corners of two calibration board in lidar using point cloud in camera frame, as you will see in Corners.cpp, I have no idea about why?

When i did the calibration experiment, the result transform is wrong. I transform the above points into lidar frame with the initial TF, the rusult is close to actual TF. This calibration repo dependents on initial TF heavily, not very perfect. Any disscussion will be welcomed, appreciate for your time, thanks a lot!