Publication Date: 2023/10/05
Abstract: With the mass production of non-metric cameras with unstable internal properties, the camera calibration task has become more and more sophisticated and difficult to perform. Several approaches are available in the literature with good performance. However, some of these techniques are laborious to perform and require the use of some distortion model specific algorithms, which cannot accommodate other types of models/lens distortion characteristics. Moreover, some of the proposed techniques require known coordinates of 3D points and can only handle small amounts of distortions. In this paper, we propose a new distortion correction model, which takes advantage of the properties of the hyperbolic curves and those of abstract geometry to estimate camera parameters without prior knowledge of the coordinates of the distortion center and 3D coordinates measures. Tested on an uneven building wall as calibration surface, the approach dealt with severe barrel distortion profile. The technique offers an analytical solution for the distortion coefficient in opposition to some of the existing approaches which suffers the limitations of iteration and Least Squares solutions which requires good initial estimates of the unknown parameters.
Keywords: Fisheye Lenses, FOV Model, Barrel Distortions, Radial Distortions, Camera Calibration, Abstract Geometry.
DOI: https://doi.org/10.5281/zenodo.8409235
PDF: https://ijirst.demo4.arinfotech.co/assets/upload/files/IJISRT23SEP1435.pdf
REFERENCES