turtlezhong/camera_lidar_calibration

A tool used for calibrate the extrinsic between 2D laser range finder (LRF) and camera. ROS Version: https://github.com/TurtleZhong/camera_lidar_calibration_v2

Camera Lidar Calibration Tool.

Author:xinliangzhong([email protected])

1.Description

The package is used to calibrate a 2D LiDAR or laser range finder (LRF) with a monocular camera. Specficially, Hokuyo UTM-30LX have been suscessfully calibrated against a mono camera.

But this approach is really a naive way, in a word, it just 3D-2D optimization problem. So we decide to develop the new approach to calibrate the extrinsic

2.Prerequisites

We have tested the library in 16.04, but it should be easy to compile in other platforms.

OpenCV

We use OpenCV to manipulate images and features. Dowload and install instructions can be found at: http://opencv.org. Required at leat 2.4.3. Tested with OpenCV 3.3.

Eigen3

Download and install instructions can be found at: http://eigen.tuxfamily.org. Required at least 3.1.0.

Ceres

Download and install instructions can be found at: http://www.ceres-solver.org/installation.html.

3.How to build

mkdir build
cd build
cmake ..
make

4.How to Prepare the calibration data

We need the 3d points in laser coordination and the same points in the image. The 3D points (actually z=0) you can put in the data/laser_points.txt; The 2D points in the image you can put in the data/image_points.txt; We combine the data in data.txt with 4 cols which represents x y u v respectively.

So how to fine the 3D-2D pairs correctly? Here we provide a simple way: Suppose your laser in the horizontal plane, and we just need to measure the height of the laser. Actually the red point in the following picture is the flag that we tested in our setups.

For the x and y, you can use the rviz (2D nav goal) tool to measure. The result will be show in the terminal. For the u and v in the image, we provide a simple tool to detect corners in the rectangle your mouse selected. and the data will automatically saved in data/image_points.txt.

5.Key Algorithm

Actually it just a least square problem.

6.Show the reprojection results

Tcl: which takes a vector from laser to camera.

7.Good luck to you!

camera_lidar_calibration's Issues

关于项目improvement的问题

嗨， 谢谢开源此项目。在readme内， 你提到了关于这方法十分naive，请问有相关new approach的信息吗？

But this approach is really a naive way, in a word, it just 3D-2D optimization problem. So we decide to develop the new approach to calibrate the extrinsic

Can you give the paper of this source?

Hello, thank you so much for the code
I really want to read your paper about the method in this code,
Would you send me?

Best regards,
Nam

Using with HLS-LFOM1 from hitachi

Hello! I was wondering if your calibration solution can work with any kind of lidar sensor? I'm planning on getting a HLS-LFOM1 sensor from hitachi. But in the end I want to have a colored pointcloud, so I'm exploring calibration techniques.

points number

Hi,

1. how many laser points in your projects for calibration R T ?
2. can I use this method to calibrate radar and camera ?