基于Hough变换的焊接接头自动识别技术研究

Research on Auto-Recognition for the Welding Joint Positions Based on Hough Transforms

以作者所研制的焊接机械手为对象,针对其在焊接大型模具时,焊缝数量多的特点,采用提供信息量大、抗干扰能力强的视觉传感器CCD来获得焊接接头平面位置信息.在对CCD摄取的模架图像进行校正、灰度拉伸、灰度阈值变换后,采用Canny边缘检测算子检测出模架筋板上侧的边缘点,并针对传统的Hough变换方法不能检测出直线端点,容易重复检测直线以及提取直线精度不高的缺点加以改进,采用改进的Hough变换方法提取了筋板边缘线,求取这些边缘线交点即可获得焊接接头平面位置信息.实验结果表明,该方法可以有效解决Hough变换存在的上述问题,并能准确稳定地提取出模架筋板边缘线,达到识别焊接接头平面位置的目的.

This paper presents a technique for auto-recognizing the welding joint positions based on Hough transforms. Because there are a large number of welding joints, the CCD vision sensor , which can provide much information about welding joint positions and have a high capability of resisting interference,is used to get images of the welding joint positions. The edge detection algorithm Canny is applied to process the original image after we stretch grayscales and transform gray-scale threshold on these images. And then we have improved Hough transforms algorithm and applied it to extract lines from the die carrier image, for the two ends of the line can not be detected by using the conventional Hough transforms and the lines can only be detected repeatedly with low precision. The result from our experiments shows that the welding joint positions can be recognized effectively and correctly.