A Micro Vision-Based Method for Multi-Scale Geometric Parameters Measurement
Li-chao DING,Wei LIU,Peng GAO,Zhen-yuan JIA
DOI: https://doi.org/10.19356/j.cnki.1001-3997.2017.07.003
2017-01-01
Abstract:A method based on micro vision is proposed in this paper,and the vision inspection system combining CNC machine tools is established for measuring a variety of the size of one parts.First,in consideration of narrow view field,small depth of focus and too many nonlinear distortion factors,a binocular micro vision imaging model based on the non-linear corrections is established.Therefore,the calibration method for scale factor based on Zhang's method,the principal point calibration method based on the variable-magnification-method,and the optimization for the main parameters and distortion paraneters based on bundle adjustment are studied.Second,in view of some common features including straight line,round,free curve,etc,the linear feature extraction based on Robert operator and the Hough transform,the circular feature extraction method based on connected area,as well as the free curve extraction method based on least square method are explored respectively.At last,in order to verify the accuracy of the measuring system,a standard work-piece with multi-scale structures are measured.The results show that the average measurement error of the system is 3.15 μm andit satisfies the requirements of measurement accuracy.