Research Article
A Data-Driven Algorithm for Large-Scale Multi-camera Calibration
@INPROCEEDINGS{10.1007/978-3-031-30237-4_7, author={Zijun Wang and Yuhui Wan and Kunlin Zhong and Yixin Zhang and Jian Wang}, title={A Data-Driven Algorithm for Large-Scale Multi-camera Calibration}, proceedings={Machine Learning and Intelligent Communication. 7th EAI International Conference, MLICOM 2022, Virtual Event, October 23-24, 2022, Proceedings}, proceedings_a={MLICOM}, year={2023}, month={4}, keywords={Multi-camera calibration Large-scale Data-driven Neural networks}, doi={10.1007/978-3-031-30237-4_7} }- Zijun Wang
Yuhui Wan
Kunlin Zhong
Yixin Zhang
Jian Wang
Year: 2023
A Data-Driven Algorithm for Large-Scale Multi-camera Calibration
MLICOM
Springer
DOI: 10.1007/978-3-031-30237-4_7
Abstract
Multi-camera calibration, which is the establishment of a mapping relationship between the 2D coordinates of individual cameras and the 3D world coordinates, has been a major challenge in computer vision technology. The model-driven multi-camera calibration, which starts from the imaging model of the camera, is computationally complex and difficult to consider the imaging distortion comprehensively, while the data-driven multi-camera calibration is often difficult to meet the needs of large-scale scenes in terms of the calibration range. To solve the above two common problems, this paper designs a multi-camera image acquisition system, which collects millions of point cloud coordinate samples in a large-scale space and uses neural networks to infinitely approximate the transformation model of 2D images and 3D world. After the calibration experiments, the scheme is simple and effective, and it can accomplish the requirement of high precision calibration for large-scale space in both spatial positioning and reprojection.
