Geometric Calibration of CCD Camera Using Planar Object

Abstract

A method of geometric calibration of charge-coupled device (CCD) vision systems for metric measurement has been developed. The method does not require a special calibration facility or accurately surveyed three-dimensional (3D) control points. It only requires that one stereo pair of images of a planar object be acquired with the CCD camera that is to be calibrated. Using a planar wall constraint as a control, the study identified that a priori knowledge of seven parameters of interior orientation could be used to effectively model the interior geometry of known focal length CCD cameras equipped with zoom lenses. Brick walls provided an excellent calibration source of such a calibration facility, because they can provide a sufficiently large number of well-defined points throughout the stereo images at almost any focal setting. The corners of the bricks and mortar joints might be used as target points. Comparison with laboratory calibration using a 3D test field showed that the method of planar constraint was capable of providing results of comparable accuracy at 10–32 mm variable focal settings lens, and potentially even better results at larger focal settings. A root-mean square (RMS) error of better than ±0.2 pixels was achieved consistently for higher values of focal setting. Using about 40–50 stereo image points, the interior geometry of the CCD camera was effectively modeled. Increasing the number of image points increased the stability of the calibration parameters. The method has the potential to be advantageous in terms of practicality, availability and economy; time saving, and providing a large number of image points.