Research On Camera Calibration Method For Micro-vision Positioning System

In modern manufacturing, automatic measuring and positioning technology has been widely used in CNC machining. In a variety of measuring techniques, vision measurement technology has attracted more attentions, because of its high precision, rapidity and the non-contact feature. The camera calibration is the basis of the visual measuring, which determines the mapping relationship between the spatial coordinates and the image coordinates of the observation point, so that the camera can be used to measure three-dimensional size and position of the observed object. Intending to achieve high-precision positioning with vision measurement in the CNC machining experiment platform, this paper uses the stereoscopic microscope as visual measuring tool. The stereomicroscope has a narrow field of view, small depth of field and complex non-linear factors. Because of those characteristics, the calibration of the stereoscopic microscope is difficult. Therefore, the study of more targeted camera calibration technology is necessary for implementing the vision measurement function in CNC. This paper mainly focuses on some key techniques as follows:To meet the needs of positioning and measuring in CNC, the binocular micro-vision measuring system has been built based on the stereoscopic microscope. This system is applied on two experimental platforms:one is a visual experimental platform and the other is a CNC machining experimental platform. According to the actual needs in the calibration process, some visual calibration experiments are performed in this paper to validate the method and theory.After considering the limitations of stereoscopic microscope, the binocular micro-vision model has been proposed, which contains the non-linear corrections on the linear imaging model, based on the analysis of existing imaging model. By verifying the feasibility of the existing calibration methods to stereoscopic microscopes, the camera calibration method of the main parameters has been proposed, including the scale factor calibration method based on Zhang’s method and the principal point calibration method based on the variable-magnification-method.A sufficient number of high-precision three-dimensional points have been collected by the stereoscopic microscope and the laser interferometer during the vertical movement of the CNC machining experiment platform. The points would be used as the input of the bundle-adjustment, which is the optimization method in this paper. Final calibration results can be obtained after optimization. In order to verify the calibration result, the three-dimensional points are reconstructed based on the calibrated camera model. Through the analysis of the difference between the reconstructed points and the real points, it is obvious that the calibration method in our paper is accurate enough.