6D Pose Estimation of Camera Based on the Fusion of Checkerboard and ICP Algorithm

Ziqiang Li, Siyv Fu, Wenxing Liao, Tao Wu
Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, PMLR 278:164-170, 2025.

Abstract

Nowadays, camera pose estimation, as a core technology in the field of 3D vision, plays a crucial role in various applications such as autonomous driving. This paper presents a camera pose estimation method that combines checkerboard pattern and ICP (Iterative Closest Point) algorithm. Based on the chessboard calibration plate captured by the mobile camera, by extracting two-dimensional feature points through pixel-level corner detection and then performing sub-pixel optimization on the feature points, and then precisely matched with the three-dimensional point cloud data obtained by the Gemini2 depth camera. A registration model based on the ICP algorithm is constructed to simultaneously solve the rotation matrix and translation vector of the camera. The experimental results demonstrate that this method exhibits high accuracy and achieves a root mean square error (RMSE) of 0.0109 and the fitness is 1. By merely utilizing 40 key feature points (derived from an 8$\times$5 checkerboard), this method reduces the computational load, enables ICP to converge within 40 iterations, and enhances the real-time efficiency.

Cite this Paper

BibTeX
@InProceedings{pmlr-v278-li25e, title = {6D Pose Estimation of Camera Based on the Fusion of Checkerboard and ICP Algorithm}, author = {Li, Ziqiang and Fu, Siyv and Liao, Wenxing and Wu, Tao}, booktitle = {Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing}, pages = {164--170}, year = {2025}, editor = {Zeng, Nianyin and Pachori, Ram Bilas and Wang, Dongshu}, volume = {278}, series = {Proceedings of Machine Learning Research}, month = {25--27 Apr}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v278/main/assets/li25e/li25e.pdf}, url = {https://proceedings.mlr.press/v278/li25e.html}, abstract = {Nowadays, camera pose estimation, as a core technology in the field of 3D vision, plays a crucial role in various applications such as autonomous driving. This paper presents a camera pose estimation method that combines checkerboard pattern and ICP (Iterative Closest Point) algorithm. Based on the chessboard calibration plate captured by the mobile camera, by extracting two-dimensional feature points through pixel-level corner detection and then performing sub-pixel optimization on the feature points, and then precisely matched with the three-dimensional point cloud data obtained by the Gemini2 depth camera. A registration model based on the ICP algorithm is constructed to simultaneously solve the rotation matrix and translation vector of the camera. The experimental results demonstrate that this method exhibits high accuracy and achieves a root mean square error (RMSE) of 0.0109 and the fitness is 1. By merely utilizing 40 key feature points (derived from an 8$\times$5 checkerboard), this method reduces the computational load, enables ICP to converge within 40 iterations, and enhances the real-time efficiency.} }
Endnote
%0 Conference Paper %T 6D Pose Estimation of Camera Based on the Fusion of Checkerboard and ICP Algorithm %A Ziqiang Li %A Siyv Fu %A Wenxing Liao %A Tao Wu %B Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing %C Proceedings of Machine Learning Research %D 2025 %E Nianyin Zeng %E Ram Bilas Pachori %E Dongshu Wang %F pmlr-v278-li25e %I PMLR %P 164--170 %U https://proceedings.mlr.press/v278/li25e.html %V 278 %X Nowadays, camera pose estimation, as a core technology in the field of 3D vision, plays a crucial role in various applications such as autonomous driving. This paper presents a camera pose estimation method that combines checkerboard pattern and ICP (Iterative Closest Point) algorithm. Based on the chessboard calibration plate captured by the mobile camera, by extracting two-dimensional feature points through pixel-level corner detection and then performing sub-pixel optimization on the feature points, and then precisely matched with the three-dimensional point cloud data obtained by the Gemini2 depth camera. A registration model based on the ICP algorithm is constructed to simultaneously solve the rotation matrix and translation vector of the camera. The experimental results demonstrate that this method exhibits high accuracy and achieves a root mean square error (RMSE) of 0.0109 and the fitness is 1. By merely utilizing 40 key feature points (derived from an 8$\times$5 checkerboard), this method reduces the computational load, enables ICP to converge within 40 iterations, and enhances the real-time efficiency.
APA
Li, Z., Fu, S., Liao, W. & Wu, T.. (2025). 6D Pose Estimation of Camera Based on the Fusion of Checkerboard and ICP Algorithm. Proceedings of 2025 2nd International Conference on Machine Learning and Intelligent Computing, in Proceedings of Machine Learning Research 278:164-170 Available from https://proceedings.mlr.press/v278/li25e.html.

Related Material