为实现大尺寸轧钢几何尺寸、形状误差、表面缺陷的实时检测, 采用4个激光器和4个CCD摄像机搭建了多线结构光视觉传感器测量系统。针对多视觉传感器全局校准的关键问题, 设计了特殊形状的靶标, 提出了新的全局标定方法。利用交比不变原理获取光平面上多个标定点, 同步实现视觉传感器的局部标定和全局校准。详细阐述了标定的基本原理和系统构建过程, 分析了标定中存在的误差, 并提出了减小误差的方法。实验验证系统重复测量精度在0.04mm以内, 测量相对误差小于0.9%。

In order to measure the large forgings’ size, shape error and surface defects online, an experimental system for 3D multi-vision measurement is set up with four sets of vision sensors. To solve the problem of the global calibration, a new calibration method based on a special shape target is proposed. On the basis of the invariance of cross ratio, more calibration points are generated. Moreover, local calibration and global calibration are unified by the proposed method. The measuring principle and the building process of the system are described. To improve the calibration accuracy, the error factors in calibration are analyzed and the method of reducing error is put forward. The experiments demonstrate that the repeatable accuracy is 0.04mm, and the relative error is less than 0.9%.