shallowdream-x / traffic-LiDAR-camera-calibration

A simultaneous intrinsic and extrinsic calibration method between LiDAR and camera in complex outdoor scenes like traffic scenes

Intrinsic and Extrinsic Calibration of Roadside LiDAR and Camera

Method

Requirements

• ＞=cmake 3.10
• pcl 1.8
• Eigen3
• Ceres 2.0.0
• OpenCV 3.4.5

Usage

• We provide tests in real world traffic scenes. There are two examples in real world traffic scenes in feature folder. The feature/feature folder stores the selected 2d features. The feature/feature_bit is the bit image for distance transform(This is auto generated). The feature/3d_feature folder stores the selected 3d features.

• If you want to extract features by hand, refer to doc/extract_feature.md

• Run the example

  • modify the initial intrinsic and extrinsic paramter in config/param.yaml like:

    # image0
    image_path: "/home/jingxin/traffic-LiDAR-camera-calibration/feature/image2/image2.jpg"
    feature_num: 15
    fx: 2000
    fy: 2000
    cx: 960
    cy: 540
    extrinsic: [9.88910941e-01,  1.48500658e-01,  1.64494210e-03,  2.20000000e+01,
                5.18266263e-02, -3.34706924e-01, -9.40895996e-01, -1.50000000e+01,
                -1.39173101e-01,  9.30547597e-01, -3.38691627e-01,  8.00000000e+01,
                0.00000000e+00,  0.00000000e+00,  0.00000000e+00,  1.00000000e+00]
    

  • build and run main.cpp

    cd traffic-LiDAR-camera_calibration
    mkdir build
    cd build
    cmake ..
    make
    ./main
    

• Results:

  After the convergence, you will see the results like this:

  

  The intrinsic and extrinsic results will be printed on the console. And the initial reproject and final reproject will be shown like this.

  initial:

  

  final:

  

Demo Gif

TODO

• rough calibration(coming soon)
• simulation code(coming soon)