search

koide3 / direct_visual_lidar_calibration

A toolbox for target-less LiDAR-camera calibration [ROS1/ROS2]
https://koide3.github.io/direct_visual_lidar_calibration/
772 stars 121 forks source link

Details On Lidar Image #52

Open JiSuanJiDaWang opened 1 year ago

JiSuanJiDaWang commented 1 year ago

Hi, I have some questions about the process of the generation of lidar- image. When generating lidar intensities image with lidar fov larger than 150 degrees, why use extrinsic like this: T_lidar_camera.linear()= (Eigen::AngleAxisd(-M_PI_2, Eigen::Vector3d::UnitX())).toRotationMatrix(); why not use the same extrinsic as the one that lidar fov smaller than 150 degrees. T_lidar_camera.linear() = (Eigen::AngleAxisd(M_PI_2, Eigen::Vector3d::UnitY()) * Eigen::AngleAxisd(-M_PI_2, Eigen::Vector3d::UnitZ())).toRotationMatrix(); Thanks!

koide3 commented 1 year ago

The difference comes from the difference of canonical poses of these camera models. The transformation for each camera model is created such that LiDAR XYZ axes align with the front-left-top axes of each camera model.