search

ankitdhall / lidar_camera_calibration

ROS package to find a rigid-body transformation between a LiDAR and a camera for "LiDAR-Camera Calibration using 3D-3D Point correspondences"
http://arxiv.org/abs/1705.09785
GNU General Public License v3.0
1.49k stars 460 forks source link

How could I do 2D-3D correspondence calibration? #34

Closed xiaoshuliu closed 6 years ago

xiaoshuliu commented 6 years ago

Hi, I have read your paper, and I wonder could I just do the 2D-3D correspondence calibration without ArUco markers? Is the code for 2D-3D in this repo? Thanks.

karnikram commented 6 years ago

The code has not been included in this repo.

And I don't think the 2D-3D approach worked as well as this one.

OmarBarakat1995 commented 6 years ago

@karnikram I've read the paper but couldn't get how 3D features are obtained in case of monocular . Is it because the boards are planar, thus all edge points are having same depth?

karnikram commented 6 years ago

Aruco gives us the 3d pose of the marker relative to the camera directly. And the coordinates of the board's corner points relative to the aruco marker are known since the board's dimensions are known. So a simple matrix multiplication can be done to obtain the coordinates of the corner points in the camera's frame.

OmarBarakat1995 commented 6 years ago

@karnikram Thank you .. But I am considered with the depth (3rd dimension) in case of monocular camera not stereo. let me thank you again and if I am misunderstanding something please refer me to a read .

xiaoshuliu commented 6 years ago

@karnikram How does ArUco give us the 3d pose? Is it the same principle as the chess board, but just use a different library?

karnikram commented 6 years ago

@OmarBarakat1995 Stereo images are not being used for pose estimation here, only monocular. Aruco markers or any fiducial markers can be used for monocular pose estimation.

@xiaoshuliu Refer to this for an explanation of the arcuo library.