A Practical Calibration Method for Multiple Cameras

Abstract

In this paper, we propose a practical factorization- and-position based method for multiple cameras calibration. The method yields a simple calibration means for an arbitrary number of linear projective cameras while maintaining the handiness and flexibility of the original method. A freely moving planar pattern as a calibration object at a few different orientations is only required. All the cameras do not have to see this pattern at all orientations, and only reasonable overlap between camera subgroups is necessary. We divide these cameras into groups according to their positions and orientations first, and then calibrate each camera in the world coordinate system of its own group via a factorization-based method. Common view fields of planar pattern are used to estimate the Euclidean transformation between these world coordinate systems and represent all cameras in a same world coordinate system. Both the intrinsic and extrinsic parameters of cameras can be obtained in a uniform world coordinate system of accuracy to within a pixel.