Method for monocular camera to estimate principal point and calibrate zoom lens

摘要

We present an improved calibration method to optimize and simplify the current principal point estimation model and the zoom lens calibration process. A Euclidean distance weighting model (EDWM) for the principal point is constructed, which can more accurately estimate the principal point, that is, the distortion center. In addition, the camera has been calibrated by Zhang's method in several zoom settings to get the intrinsic parameters, which can be used to simplify the distortion model in zoom processing. As a result, the discrepancy between the principal point determined by the EDWM and Zhang's approach is 2.7484 pixels on average, and the reprojection root-mean-square error in the image distortion correction experiment is 0.1157 pixels. The results demonstrate that the proposed method is precise and effective in identifying the principal point while simplifying the process of zoom lens calibration. Therefore, the approach is suitable for close-range vision measurement tasks that require large-scale zoom camera calibration.