Usually, a head-eye system comprises of a pair of cameras mounted on a platform. Calibration of such a system can be divided into two parts. The first part is concerned with calibration of intrinsic parameters of the cameras where as the second part deals with calibration of extrinsic parameters of the cameras which is realized through kinematic calibration of the system. In this paper, we solve this kinematic calibration problem. First we formulate the problem for a 6-degree-of-freedom (DOF) head-eye system. It turns out that this problem is very similar to the hand-eye calibration problem, i.e., to solve an equation system of AX=XB, where X is the unknown transformation matrix which contains a rotation and a translation. In a special case, where the system has only rotational motion, the rotation and translation of X are decomposed into two independent equations which are solved separately. We propose a nonlinear optimization solution for the rotation. Algorithms from early work have also be implemented for the purpose of comparison. Experiments and tests are performed on both synthetic and real data. Results are compared and presented in this paper.
展开▼
