为实现窄搭接焊的在线检测, 运用三维激光线阵相机获得焊后带钢的表面形貌, 通过OpenCV分离图像通道获得焊缝的灰度图和高度图。针对焊缝图像信息量大, 对比度低, 边缘模糊等特点, 通过卷积增强焊缝特征, 结合积分图提出改进的图像分割算法, 能快速准确提取焊缝边缘。结合形态学, 数字图像处理方法, 提出一种层次性的缺陷提取方案, 在保证运算时间短的同时有较高的准确性。针对焊缝缺陷特征分类缺陷, 并在图像中对每种缺陷进行标识。结果表明, 该方法能有效用于企业实际生产中。

For online inspection of narrow lap welding, the surface topography of the post-weld strip is obtained by using a three-dimensional laser line array camera, and the gray scale and height map of the weld are obtained by OpenCV separate image channel. Aiming at the large amount of weld image information, low contrast, blurred edges, etc., by converging enhanced weld features and combining the integral map, an improved image segmentation algorithm is proposed, the weld edges can be extracted quickly and accurately. Combining morphology and digital image processing, a hierarchical defect extraction scheme is proposed to ensure high accuracy while ensuring short computation time. The defects are classified according to the weld defect characteristics, and each defect is identified in the image. The results show that the method can be effectively used in the actual production of the enterprise.