A novel method for field complete parameters calibration of stereo vision sensor

作者:Wu Bin*; Gao Qingze; Zhu Hongyan
来源:4th International Seminar on Modern Cutting and Measurement Engineering, 2010-12-10 to 2010-12-12.
DOI:10.1117/12.885870

摘要

To meet the need of high precision vision measurement, a novel method for field complete parameters calibration of stereo vision sensor is presented based on the two-dimensional target with unknown motion. Method: Firstly, according to the perspective projection model of camera and its coordination's relationship, the measurement math model is set up for stereo vision sensor. Secondly, the precision parameters of camera are calibrated with Zhang's camera calibration method based on planar target and the initial transition matrix of stereo vision sensor can be got. Finally, with the 3D measurement math model of stereo vision sensor and Levenberg-Marquardt method, the optimized parameters of stereo vision sensor are achieved. Result: Based on the above calibration method and two-dimensional target with unknown motion, the complete parameters of stereo vision sensor (including the camera's parameters and the transition matrix between two cameras) are calibrated. Experiments show that the relative error of distance measurement is less than 1.6% using a calibrated stereo vision sensor. Conclusion: The method is simple and convenient, with better efficiency, lower labour intensity and higher precision.

全文