论文部分内容阅读
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 realize the real-time detection of large size rolling geometry, shape error and surface defects, a multi-line structured optical vision sensor measuring system was built by using 4 lasers and 4 CCD cameras. Aimed at the key problem of global calibration of multi-vision sensor, a special shaped target is designed and a new global calibration method is proposed. Utilizing the principle of cross ratio invariance, multiple calibration points on the light plane are acquired, and the local calibration and global calibration of the vision sensor are realized simultaneously. The basic principles of calibration and the process of system construction are expounded in detail. The errors existing in calibration are analyzed and the methods to reduce the errors are proposed. Experimental verification system repeated measurement accuracy within 0.04mm, the relative error is less than 0.9%.