Modeling and sphere constrained calibration of a vision coordinate measurement machine

Abstract

A new vision coordinate measurement machine (v-CMM) is proposed by integrating a binocular stereo vision tracking device and a 1D distance sensor to measure the surface. Using a standard sphere, the calibration process is carried out in two steps. Fewer points on the sphere are measured and the linear calibration model is used to calculate the approximated transformation. More points are collected and the radius-difference based optimization model is used to calculate the accurate transformation. Using the Monte Carlo method, the calibration uncertainty is analyzed by deriving the relationship between calibration uncertainty and tracking random errors. The effectiveness of the calibration algorithm is verified by analyzing the situation where the center error and tracking error mix. Two experiments are carried out: One is the calibration experiment, which will obtain the structure parameters of v-CMM with uncertainty. In the experiment, the measured runout error of the sphere is 0.024 mm, which is able to evaluate the accuracy of the calibration process. The other experiment is that a part with three cylindrical bars is measured. Key sizes of the part are calculated, containing bars radiuses and the distances between them. The experimental results show that after calibrating, the measuring precision of v-CMM is highly improved. The v-CMM is suitable in specific scenarios, like transparent media surface measuring.