A method of small-field-of-view and binocular camera calibration by using a wafer

The premise condition of the water jet guide laser cutting wafer based on machine vision is to obtain a precise wafer alignment. However, the wafer alignment accuracy is directly affected by the calibration accuracy of the visual system. In this paper, a method of the small-field-of-view and binocular camera calibration is presented for improving the wafer alignment precision and efficiency. The method only requires each camera to observe a wafer with LED substrate at a few (at least three) different unknown orientations, then the five intrinsic parameters can be determined linearly by assuming that the center of the frame as image origin and the two image axes are very perpendicular to each other. The method consists of initializing camera parameters, refining nonlinear parameters by the Levenberg-Marquardt algorithm and the calculating relative position between binocular cameras. In particular, the sub-pixel feature point extraction with more accuracy is obtained by using imaging processing technology. In the end, the simulation and experiment show that the proposed method is accurate.