本文结合机械自动化和视觉激光器的优点, 对视觉激光器金属板材焊接系统及其应用进行了研究, 系统配置高分辨率相机、高性能CCS光源, 能够实时对高精度图像进行采集; 利用迭代重加权最小二乘法、累计概率Hough变换等图像处理算法, 对半导体激光器、非球面透镜边缘直线进行快速拟合, 实现自动对准。在PPHT算法对灰度随机分布低对比度俯视图失效时, 提出求导寻点法, 实现了非球面透镜和半导体激光器边缘直线特征点的提取, 并完成了直线拟合。系统采集的图像像素精度属于亚微米级别, 整个对准过程在0.5 s可完成, 大幅提高了产品的成品率和生产效率。

Based on the advantages of mechanical automation and visual laser, this paper studied the metal plate welding system of visual laser and its application. The system was equipped with a high-resolution camera and a high-performance CCS light source, which can collect high-precision images in real time. The image processing algorithms, such as iterative weighted least two multiplication and cumulative probability Hough transform, can be used for semiconductor laser and non sphere laser. The edge line of the face lens was fitted quickly to realize automatic alignment. When the PPHT algorithm fails to work on the low contrast top view with random distribution of gray scale, the method of seeking points was proposed to extract the linear feature points of the edge of aspheric lens and semiconductor laser, and the linear fitting is completed. The pixel precision of the image collected by the system belongs to the sub micron level, and the whole alignment process can be completed in 0.5 s, which greatly improved the product yield and production efficiency.