Calibration Method For Central Catadioptric Camera Using The Pole-Polar Relationship Of The Images Of Spheres And Lines

In the researches of computer vision,camera calibration is crucial to the accuracy of 3D reconstruction.As common geometric elements,the projective properties of lines and spheres have been widely studied in camera calibration.The central catadioptric camera is becoming more and more popular in daily life because of larger imaging field.In this thesis,a unified representation of the images of lines or spheres is discussed under the unified model of the central catadioptric camera.It is found that in the pencil of the images of a lines or a spheres and corrected absolute conic,the relationship between the line formed by degenerate conic and the point formed by dual degenerate conic satisfy the pole-polar relationship.Therefore,a linear method for obtaining the corrected intrinsic camera parameters was proposed when images of at least three lines or spheres are obtained.The images containing lines can be used to solve the camera intrinsic parameters by the property that the images of lines intersect at two points and the line decided by these points and the vanishing point in the image satisfy pole-polar relationship.At the same time,this conclusion is extended to the case of using spheres as the calibration object,and a method similar to calibration using lines as object can be used to calibrating.Therefore,the intrinsic parameters of the camera can be solved by obtaining two images of lines or spheres from three different directions and combining with the mirror contour.Finally,the mirror parameters can be gained via the corrected intrinsic parameters obtained in the first step.Through a lot of simulation and real experiments obtained by the images of spheres and lines,this thesis proves the feasibility of the proposed method.