Camera Calibration For The Double Plane Mirrors Catadioptric Camera Based On Coaxial Parallel Circles

Camera calibration is an essential step in computer vision,the main purpose of which is to establish the relationship between 2D image and 3D space.The advantage of a catadioptric camera system composed of pinhole camera and two plane mirrors is not only that it is easy to make,but also that it can expand the field of vision of imaging,and can obtain multiple views of objects in one image.In this system,the point set composed of a point on the object and its reflection points is called a reflection point group.A reflection point group is on the same circle,and all reflection points group circles are coplanar coaxial parallel circles.In this thesis,four camera calibration algorithms are proposed based on coaxial parallel circle in the double mirrors catadioptric imaging system.The internal parameters of the camera are solved by obtaining orthogonal vanishing points,the image of circular points or a pair of the pole-polar with respect to the image of the absolute conic.Two infinite points and an infinite line are obtained by using any two reflection point groups and their circles.Two groups of infinite points in orthogonal direction can be deduced from the property of infinite points about the polar line of a circle.According to Projective Invariance,two groups of orthogonal vanishing points can be derived.By solving the intersection point of circular image and vanishing line,a pair of images of circular points can be obtained.In addition,according to the inference of Laguerre's theorem,a pair of the pole-polar with respect to the image of the absolute conic can be obtained.Using any a reflection point group circle and their circles,in which any three reflection points and a circular point can form a complete quadrangle on a circle.According to the projective invariance of the harmonic conjugate property on the opposite edge of a complete quadrupole,the image of a circular point can be obtained by the constraint of the harmonic conjugate property on the image plane.Several simulation experiments,real experiments and 3D reconstruction results were carried out to verify the feasibility and effectiveness of the proposed calibration algorithm in this thesis.