The Key Technology Research Of Large-sized Workpiece Measuring Based On Multi-camera Vision

With the large-sized workpiece measuring in the industrial field increasingly showing the following characteristics: the measuring process is automated, the measuring environment is complicated and the measuring object is diverse, the traditional measuring means hasn’t met the needs of large-sized workpiece measuring. The visual measuring technology has the folllowing advantages: non-contract, high speed, high automation and good stability, it is used widely in the industrial measuring field. Multi-purpose vision measuring technology based on binocular stereo vision is used to measure geometric parameters of truck side-member assembly holes according to the actual measuring needs of car factory. The main research contents and achievements are as follows:(1) The vision measurement scheme is determined: The measurement area is divided into several sub-regions, and the measurement data of each sub-region is obtained based on the binocular vision measurement system. The camera coordinate system of a sub-region is regarded as the world coordinate system, and the measurement data of other sub-regions is integrated into the world coordinate system to achieve wider measuring range.(2) The feature points are extracted based on Harris operator and SUSAN operator. SUSAN operator is better to extract feature points of holes’ edge. Propose to seek feature points by pasting mark points. Every two mark points can be connected into a linear, and the intersections of the linear and assembly holes’ edge are seen as match points, this will avoid match difficulties effectively.(3) The workpiece’s edge is detected based on different detection operators. Canny operator balances edge points’ precise positioning and noise suppression legitimately, so the edge pixels are extracted based on canny operator, and the traditional Canny algorithm is improved. The elliptic equation is fitted based on the least square method.(4) Each camera is calibrated based on Halcon, and the three-dimensional coordinates of holes’ edge points are obtained according to calibration results. The center coordinates and radius of holes is fitted based on its geometrical properties. Finally, the measurement error is analyzed.(5) The three-dimensional data registration method based on planar target is proposed. A calibration board that has larger size and higher accuracy is placed between adjacent sub-regions. Three non-collinear markers are captured by each sub-region to achieve data registration between each sub-region. Finally the method’s accuracy is verified based on calculating the distance between two markers of the target.