Paracatadioptric Camera Calibration By Using The Common Self-Polar Triangle From Sphere Image

Camera calibration is widely used in many fields of computer vision technology,including virtual reality,robot navigation,3D construction and visual surveillance.The central catadioptric camera provide a wider perspective than the ordinary camera,therefore,it has great theoretical significance and practical value to study the calibration of central catadioptric cameras.In this thesis,we discuss the case when the mirror parameter of the central catadioptric camera is 1.The ball is one of the common and completely symmetrical geometric figures,therefore,the use of space sphere to calibrate catadioptric cameras has become the focus of current research.In this thesis,we studied the central catadioptric camera model and discuss a method of using sphere calibration parabolic catadioptric camera,the specific methods are:first,projection of ball on a virtual camera are two small circles,we call them extension circles,one of them is visible and the other is invisible,the two small circles became two images of ball under the projection of the virtual optical centers;second,extract the great circle who parallel to the small circle and its center lies on the center of the unit ball,and calculate its image,it has one and only one common self-polar triangle with the antipodal sphere image,connect the vertices outside the image of the great circle,and the line is the vanishing line,the vanishing line and the image of sphere intersect at two images of circular points,take three pictures we can get six images of circular points,these six images of circular points form six constraint equations of camera intrinsic parameters,thus the camera intrinsic parameters can be obtained;finally,experiments with simulated and real image data are carried out by Matlab,the experimental results show the algorithm is feasible and effective.