Should images be rectified as an input to your package? about extrinsic_lidar_camera_calibration HOT 2 CLOSED

Hi,

We only use one camera; therefore, there's no rectification process. However, in the equation (6), we only use a pin-hole model and assume the camera itself is calibrated, i.e., intrinsic matrix and distortion parameters are given. Thus, the given camera corners are calibrated; the projected LiDAR vertices are also calibrated. If you have an uncalibrated camera, I would suggest calibrating it before using this package to get better results!

Please let me know if you have more questions!

from extrinsic_lidar_camera_calibration.

Sorry for being imprecise (in ROS image_proc package a rectified image is the one without lens distortion and de-Bayered). What I wanted to ask is whether or not the input images should be undistorted (free from lens distortion). And based on your answer and the analysis of your code, I conclude that they definitely should be undistorted because of formula (6) which assumes pure perspective projection for lidar vertices and because there is nothing in the code that makes the images undistorted.
Thanks for your answer!

from extrinsic_lidar_camera_calibration.