Camera Calibration Simulation using a Randomly Generated Spherical Point Distribution

In this paper, we present a method that utilizes computer vision, specifically projective geometry, to map a known distribution of points on a sphere - with known diameter - along with an arbitrary image of these points on an image plane to identify the configuration of the camera. In other words, knowing the sets of 2D-3D corresponding points, one can extract the camera matrix and dissect it into parameters of interest: intrinsics and extrinsics. The method that is validated by code shows in detail how to setup a theoretical world and camera coordinate frame, and then through the knowledge of the correspondence, displays the solution to the optimization problem. The results are then analyzed noting the relative error between the retrieved and actual camera matrices.