| Camera calibration is a basic problem in computer vision.By calibrating the camera,the three-dimensional spatial information of the image can be further recovered.The accuracy of the calibration results determines whether the next work can be carried out smoothly.So it has great research significance.There are many kinds of existing calibration algorithms.Conic is widely studied because of its good stability and easy to obtain.In this thesis,based on analysis of the common tangent properties of confocal conics under pinhole model,two camera calibration methods are proposed.The algorithm is as follows:Under the pinhole camera,four common tangents of the central confocal conics intersect at six points,of which two groups are real focus and virtual focus,and the other group is circular points.By judging the position relationship between the real focus and the infinite line and the short axis,the infinite line and the circular points can be distinguished.From the properties of focus and principal axis,two mutually perpendicular principal axes and a group of mutually orthogonal infinite points can be obtained.For centerless confocal conics,their four common tangents are a pair of conjugate complex lines and two coincident infinite lines.The infinite line and the circular points can be easily distinguished.Moreover from harmonic conjugation of complete quadrangle and Laguer theorem,two pairs of orthogonal infinite points can be obtained.According to the projective invariance,the images of the circular points and the orthogonal vanishing points can be obtained by the images of any two confocal conics,and then the calibration can be completed.The image of the conic dual to the circular points can be solved by the vanishing line and two lines in the orthogonal direction,and then the external parameters of the camera can be recovered.The effectiveness and robustness of the algorithms are verified by a large number of simulation and real experiments on the computer platform. |
