About

This code is to porject the LiDAR point cloud to the image & generate the point cloud with color

Input

• Images (.png files)
• Point cloud files (.bin files)

• Calib files
  Support two kinds of calib files’ type.

  • All parameters stored in one file. e.g. KITTI

    • R_rect: shape 3*3
    • P_rect: shape 3*4
    • Tr: shape 3*4

  • The parameters are stored in two files. One is for the camera, the other is for LiDAR2Cam. e.g. KITTI raw

    • R_rect: shape 3*3

    • P_rect: shape 3*4

    • R: shape 3*3

    • T: shape 3*1

Steps

• Put all the images into img, and put all the point cloud files into lidar(Notice: The filename of the image and the corresponding point cloud file should be the same.). Put the calib file into calib

• Set the parameters in main.py.

  • CALIB_TYPE: 0:All parameters in one file; 1:Parameters are stored in two files

    If CALIB_TYPE == 0:

    • CALIB: Path of the calib file

    If CALIB_TYPE == 1:

    • CAM2CAM: Path of the calib file for camera
    • LIDAR2CAM: Path of the calib file for LiDAR2Cam

  • IMG_PATH: The path of images
  • LIDAR_PATH: The path of point cloud files
  • SIMG_PATH: The path to stored the result images
  • SPC_PATH: The path to stored the result point cloud with color
  • SBEV_PATH: The path to stored the result BEV
  • SFV_PATH: The path to stored the result FV
  • CAM_ID: The NO. of the camera.

    For the multi-camera sensor system like KITTI, you need to set it to the ID of the camera. For the system which only has one camera, it is set to be 0.

• Run main.py to get the result.

    python main.py

Visualization

Image

The result image is saved in the path you set

Point cloud with color

• Go to the path you set to save the result .pcd files
• Use Open3d 0.7.0.0 to show the point cloud

    python pcd_vis.py

BEV & FV

The BEV & FV is saved in the path you set