用于管道焊接的双模板匹配视觉定位方法

现有管道焊接的需求量越来越大,对自动焊接提出了更高的要求。为高效实现管道焊缝自动跟踪,本文引入机器视觉方法,设计一个视觉系统采集焊缝坡口图像,提出基于双模板匹配的视觉定位方法对焊缝中心和焊枪摆动中心进行实时定位处理,并根据二者偏差值实现焊接过程中的自动纠偏操作。实验结果表明,即便在弧光干扰下,所提出的双模板匹配方法仍能很好的实现定位处理。在计算复杂度相当,不影响定位实时性的情况下,其匹配精度可由单模板条件下的70%左右提升到90%以上,有效地提升了管道自动焊接的精确度与可靠性。

A visual localization method of dual template matching for pipe welding

With the increasing demand for pipe welding,the higher requirements are put forward for automatic welding.For efficient implementation of pipe weld automatic tracking,machine vision method is introduced to design a visual system to collect images of weld groove.The visual localization method of dual template matching is proposed to locate the welding line and welding torch swing in real time,and the deviation between these two positions is used to realize the automatic correction operation.The experimental results show that the proposed dual template matching method can achieve the localization even under the arc interference.In the case that the computational complexity is almost equal,and the real-time positioning performance is not affected,the matching accuracy of the proposed method can be more than 90%,as the single template method is only about 70%.Thus,the accuracy and reliability of the automatic pipe welding is improved effectively.