基于纹理特征的焊缝识别方法

提出了一种基于纹理特征的焊缝自主识别视觉方法.不同于通常利用被焊工件的宏观结构特征的主动视觉和利用工件图像中明显的灰度梯度的被动视觉,该方法利用待焊区(焊缝位置)与母材区之间明显的纹理特征差异来实现;能够解决多层多道焊后几层、道次的焊接时通常视觉方法不易识别焊缝的问题.在该方法中,先对图像进行纹理分析.提取纹理特征,再利用待焊区和母材区的纹理特征差异,采用图像分割的方法,确定焊缝区域位置以及焊缝中心.结果表明,该方法能够适用于多层焊盖面焊焊缝识别.

Weld recognition based on texture feature

An automatic recognition and welding seam-tracking visual method was proposed.Different from the active vision method using the macrostructure feature of the workpiece, the passive vision method the gray gradient of the workpiece image is used, It uses the obvious difference of texture feature between welding seam region and base metal region.It has advantage to solve the weld recognition in the multi-layer welding, which is diffcult for the usual visual method.In the method, the textural feature of the image was firstly analyzed, and then the image was segmented by the difference of the image texture feature between the weld region and the base metal region;finally, the zone of the weld were recognized, and its position and the center could be found.Experiment showed the proposed method was effective for welding seam recognition in multi-layer welding.