Camera calibration is the necessary procedure before utilizing digital cameras (or CCDs). Based on the assumption of pinhole camera model, conventional approach of camera calibration is to solve the extrinsic and intrinsic parameters simultaneously. In general, the whole process is very complicated and not guaranteed to obtain reasonable parameters. To simplify the calibration process and get reliable camera parameters, we use accurate position control to identify the extrinsic parameters while Levenberg-Marquardt method is implemented to optimize the intrinsic parameters. From experimental results, we discuss the reasons of parameters variance for real and imperfect images. To prove the accuracy of calibrated results, camera parameters are used to reconstruct 2D image directly from 3D object. Furthermore, the concept is also extended to stereo vision. It is shown that the calibration of binocular cameras system increases the accuracy of 3D metrology extremely.