A geometric approach is proposed to calibrate the image center (also called principle point) of a camera, which is independent of other intrinsic parameters such as the focal length, distortion coefficients, and so on. The basic idea arises from intuitive experience that when the camera moves backward, the path of a static feature point will evolute as a straight line toward the principle point in the image space. The presented method does not require a precise calibration block, and it is insensitive to radical distortion and the variation of focal length. Both linear and nonlinear least square methods are exploited to obtain stable and accurate calibration results. Experiments are conducted to validate the proposed technique.
展开▼