Virtual Camera Based a Novel Calibration without Modeling

Usually, the camera calibration and distortion correction are both considered as procedures with two steps: modeling, and then solving the parameters. In this paper, we introduced a virtual camera into camera calibration and presented a novel method for calibration without modeling. Without the consideration of imaging model of cameras, we could build geometric relationship of imaging by constrains' mappings from 3D object space to 2D image plane through photoelectrical measuring, instead of the conventional calibration of intrinsic parameters. Based on the constraints' mapping, we designed a virtual and ideal pin-hole camera model for extrinsic parameters calibration, whose intrinsic parameters could be set arbitrarily and had no influence to final calibration results. Our method is especially appropriate for the camera models which cannot be simplified as pin-hole model or has much distortion, such as short focus wide-range imaging, fish-eye lens imaging and vision measurement system requiring high calibration accuracy.