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

Kabsch or ICP #28

Closed blutjens closed 6 years ago

blutjens commented 6 years ago

In your paper, you mention the results for ICP (iterative approximation) and Kabsch (closed-form solution) algorithm.

  1. Why does ICP outperform Kabsch in the RMSE? Is the error in Kabsch caused by averaging the rotation matrices?

  2. Does the code use ICP or Kabsch algorithm? Is there a way of switching in between both?

Thanks!

karnikram commented 6 years ago

A1) I'm unsure about it myself. Maybe @ankitdhall could help. A2) This program uses Kabsch. The ICP implementation hasn't been included in this repository. But I thnk PCL's implementation of ICP was used!

ankitdhall commented 6 years ago

Kabsch is used because point correspondences are known. A reason for error could be error in measurements of point coordinates. The current code does not use ICP.

lkqnaruto commented 1 year ago

Kabsch is used because point correspondences are known. A reason for error could be error in measurements of point coordinates. The current code does not use ICP.

Can you explain what does "point correspondences" mean? Thank you