Camera Calibration

中文 | English

This repository calibrates cameras using ChArUco markers, streamlining the camera calibration process.

Installation

camera_calibration requires opencv-contrib-python and opencv-python version 4.7.0 or higher. Installing this package locally will automatically fetch other required dependencies.

git clone https://github.com/Onicc/camera_calibration.git
cd camera_calibration
pip install -r requirements.txt

Usage

Generate and Print Calibration Board

Run the following command to generate the calibration board. The path for the calibration board image is set to ./calibration_board.png. You can also use the provided calibration_board.png. Print it out.

python3 generate_calibration_board.py

Camera Calibration

Measure the side length of the pure black squares on the calibration board in meters. Fill in the square_length on line 9 of camera_calibration.py with the measured value, for example, 0.024.

board, aruco_dict, board_name = ChArUcoBoard(width=12, height=8, square_length=0.024)

Capture images of the calibration board from different angles, collecting more than 20 images in PNG format, and store them in ./ChArUcoData. Run the following command to calibrate. The calibration images will be displayed, press any key to switch to the next image.

python3 camera_calibration.py

After calibration, the command line will print camera parameters and distortion coefficients. The parameters will also be saved in ./params/camera_params.yaml.

camera_matrix:
 [[3.03536832e+03 0.00000000e+00 2.01985797e+03]
 [0.00000000e+00 3.03628900e+03 1.48595051e+03]
 [0.00000000e+00 0.00000000e+00 1.00000000e+00]]
distortion_coefficients:
 [[ 1.98927651e-01 -8.93785973e-01 -9.71016301e-05 -6.80081320e-04
   1.60146884e+00]]
Calibration successful. Calibration file used: ./params/camera_params.yaml

Estimate Camera Pose Using Calibration Parameters

Modify the image file path on line 59 of detect_marker.py. Run the following command, and it will automatically read the camera parameters from the YAML file, draw coordinate axes on the ChArUco markers, and print the rotation and translation vectors in the terminal.

python3 detect_marker.py

rvec:  [[ 1.85860902]
 [ 1.83634612]
 [-0.55535779]]
tvec:  [[-0.11929498]
 [-0.04856779]
 [ 0.42148201]]

Reference

https://github.com/Jcparkyn/dpoint