Linear Catadioptric Camera Calibration from Sphere Images

A spherical object has been introduced into central catadioptric camera calibration through utilizing the properties of an image conic, which is the projection of the occluding contour of a sphere in the catadioptric image. However, previous calibration method using sphere images employs nonlinear optimization method and requires a good initial estimation to start the minimization. In this paper, we propose a novel linear approach to this problem. The main contribution of this paper is that we discovered that each sphere image is tangent to the modified image of the absolute conic (MIAC) at two double-contact image points, and a linear calibration method using sphere images is derived from this observation. The linear algorithm has been tested in extensive experiments with respect to noise sensitivity.