中文摘要：

焊缝跟踪精度对焊缝成形质量有很重要的影响，V型坡口是一种最常见的焊接接头。提出了一种提高激光视觉机器人跟踪V型坡口精度的软件措施，介绍了提取V型坡口中心亚像素级坐标的方法。采用多尺度的LOG（Laplacian of Gaussian）算子检测焊接坡口特征边缘，在分析图像灰度直方图的基础上选取适当的阈值进行二值化。用曲率极值的角点检测法从细化后的图像中提取焊接坡口中心点像素级图像坐标，然后对此点邻域的梯度值进行拉格朗日多项式插值求得亚像素级坐标。实验结果表明，该方法简单可行，能满足机器人焊缝跟踪的实时性要求。

英文摘要：

Welding seam tracking precision has great effect on welding seam formation quality, and the V-groove is a type of common joint in welding. We develop the software for enhancing the V-groove＇s accuracy to be tracked by a robot with laser vision, and present a method for extracting the subpixel coordinates of the V-groove center. We use the multi-scale LOG（Laplacian of Gaussian） operators to detect the feature edge of the welding groove. After analyzing the gray level histogram of its image, we select an appropriate threshold to binarize the image. Using the extremum curvature＇s comer detection method, we extract the pixel coordinates of the V-groove＇s central point from the pruned and thinned binary image and then carry out the Lagrange polynomial interpolation of the neighboring gradient values of the central point, thus obtaining its subpixel coordinates. Experimental results show that the method is simple and feasible, and satisfies the real-time requirements for a robot to do welding seam tracking.