| In the field of computer vision,acquiring the intrinsic parameters is also called camera calibration,it determines the accuracy of 3D information acquisition.Because the view angle of the ordinary camera is limited,it reduces the precision of camera calibration Compared with ordinary camera,the panoramic camera system has the advantage of the wide field of view.the imaging model is unit viewing sphere model:the first step,the space point projects to the unit viewing sphere;the second step,the projection of the point on the unit viewing sphere projects to the image plane by the virtual camera.Let a ball as a calibration object.it's shining,and the projection on the image plane forms a quadratic curve,it is easy to be extracted,the extraction precision is high.Based on the research of the properties of the paracatadioptric projection of the ball,This thesis presents the method of using the ball to solve paracatadioptric cameras within the parameters.The results show that our method is doable,the main content of this thesis can summarized as follows:(1)The ball is projected to the unit viewing sphere,forming a small circle.There is the great circle through the center of the unit viewing sphere and is parallel to the small circle.because of the relation between image of the small circle and image of the great circle on the image plane,projection of the great circle is calculated in the image plane by the virtual camera.(2)For the unit viewing sphere,on account of a point in the circle outside the great circle diameter is perpendicular to the ligature through the two endpoints of the circle diameter,a set of orthogonal vanishing points can be acquired on the image plane.A picture of a ball image provides a pair of orthogonal vanishing points.the result can be obtained by three pictures.(3)Using the MATLAB to do the experiment for this thesis and the algorithm proposed of this thesis,and the feasibility of the calibration algorithm in this thesis is verified. |
PDF Full Text Request