A high-accuracy monocular self-calibration method based on the essential matrix and bundle adjustment

摘要

Accurate camera calibration is necessary in photogrammetry systems. While traditional calibration methods need precise calibration pattern, this paper introduces a high-accuracy self-calibration method that doesn't need any metric information of the scene. The proposed method combines epipolar geometry and bundle adjustment, and both intrinsic and extrinsic parameters are obtained. Synthetic data as well as real data have been used to test the method, and the simulation result shows that the mean absolute relative calibration errors of the focal lengths and the principal point are about 4.5e-4 and 8e-4, respectively, under zero-mean Gaussian noise with 0.1 pixels standard deviation. Compared with existing calibration methods, the proposed method is cost-effective and simple with comparable accuracy.