UDC 004.021
ALGORITHM FOR TWO-STAGE CALIBRATION OF CAMERAS WITH SMALL ANGULAR DIMENSIONS OF FIELDS OF VIEW INTERSECTION
I. A. Kudinov, post-graduate student, Department of Electronic Computing Machines, RSREU, Ryazan, Russia;
orcid.org/0000-0003-1795-097X, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
M. B. Nikiforov, Ph.D. (Tech.), assistant professor, Department of Electronic Computing Machines, RSREU, Ryazan, Russia;
orcid.org/0000-0002-4796-0776, This email address is being protected from spambots. You need JavaScript enabled to view it.
I. S. Kholopov, Ph.D. (Tech.), Dean of the Faculty of Radio Engineering and Telecommunications, RSREU, Ryazan, Russia;
orcid.org/0000-0001-5220-0811, e-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
The article discusses an algorithm for two-stage photogrammetric calibration of a distributed aperture panoramic vision system cameras with small (less than 10 %) relative angular dimensions of its fields of view intersection. The aim of the work is to develop a calibration algorithm that provides a stitching error of the order of pixel units when forming a panoramic frame without taking into account information about the corresponding pairs of special points (i.e., only by information about external and internal parameters of cameras). The idea of the algorithm is to evaluate at the first stage of calibration only the matrices of internal parameters of cameras and the coefficients of distortion introduced by their lenses, using a planar test object with a pattern such as «checkerboard» or «radial-ring marks», and at the second stage evaluating only their external parameters using a volumetric test object of «reference cluster» type, one of linear dimensions of which is much larger than the other two. The experiment carried out by the authors showed that in contrast to the well-known calibration algorithm with an auxiliary camera, the developed algorithm provides up to five times less absolute panoramic frame stitching error.
Key words: camera calibration, Levenberg – Marquardt optimization algorithm, projective camera, matrix of camera internal parameters, distortion, Brown – Conrady model, Perspective-n-Point problem, panoramic image.