膜式壁焊缝图像处理及其识别方法

Membrane Wall Weld Image Processing and Its Identification Method

提出一种基于新的单目视觉线性结构光的膜式壁焊缝识别方法,克服了焊接过程中的飞溅、烟雾等干扰,快速地提取了膜式壁角焊缝特征点。首先,根据激光条纹图像与背景的显著性差异,采用灰度级频率法提取了激光条纹图像;其次,基于骨架抽取提取了激光条纹中心线,并对中心线坐标应用Takagi Sugeno模糊算法进行滤波;最后,应用动态ROI搜寻法快速找到含有焊缝特征点的区域,结合快速排序算法找到焊缝特征点坐标。经实验证明,该算法能够快速准确地提取焊缝特征点,结合控制算法,能够准确跟踪焊缝。

A new method for membrane wall weld identification based on a new monocular visual linear structure light was proposed.It overcame the spatter,smoke and other disturbances during the welding process,and could quickly extract the characteristic points of the membrane wall fillet weld.Firstly,according to the significant difference between the laser fringe image and the background,the laser fringe image was extracted by the gray level frequency method.Secondly,the laser fringe center line was extracted based on the skeleton extraction,and the Takagi Sugeno fuzzy algorithm was applied to the center line coordinate for filtering.Finally,the dynamic ROI search method was used to quickly find the area containing the weld feature points,and the fast sort algorithm was used to find the feature point coordinates of the weld.The experimental results show that the algorithm can quickly and accurately extract the weld feature points and accurately track the weld seam combined with the control algorithm.