图像处理在机器人焊缝识别中的应用

Application of Image Processing in Robot Weld Seam Recognition

采用Visual C++6.0开发了焊缝图像采集及处理系统,对电荷耦合器件(CCD)摄像头获取的焊缝激光条图像进行特征提取,分别采用两套不同的识别方案对激光条图像进行处理,前者采用中值滤波-拉普拉斯锐化-Sobel边缘检测-中线提取;后者采用平滑滤波(平均模板)-自适应阈值分割-Krisch边缘检测-中线提取。通过实际中试验结果的对比发现,后者能更好的对激光条图像进行特征识别。

This paper develops the weld seam image acquisition and processing system by VISUAL C＋＋6.0 and applies it to extract the feature of laser welding image, which is shot by the charge-coupled device （CCD） video camera. According to the laser image characteristics, the author adopt two different sets of identification programs for processing of the laser welding image.The former is the detection of median filter—Laplacian sharpening—Sobel edge detection—centerline extraction. The latter is the detection of smoothing filte1＂ （average template） --adaptive threshold segmentation - Krisch edge detection--centerline extraction. The comparison of the experiment results shows that the latter is a useful way to finish the feature recognition of laser welding image.