基于磁光图像纹理特征的焊缝识别  

Seam recognition based on magneto-optical image texture

作　　者：莫玲[1] 高向东[1] 萧振林 陈晓辉[2] 

机构地区：[1]广东工业大学 机电工程学院,广州市510006 [2]广州番禺高勋染整设备制造有限公司,511400

出　　处：《焊接》2015年第8期13-17,73,共5页Welding & Joining

基　　金：国家自然科学基金项目(51175095);广东省学科建设科技创新项目(2013KJCX0063);广东省自然科学基金资助项目(10251009001000001);广州市科技计划项目(1563000554);佛山市科技创新专项资金项目(2014AG10015);广东省重大科技专项(2014B090921008)

摘　　要：研究一种基于磁光图像纹理特征的紧密对接焊缝(间隙小于0.1mm)的焊缝识别方法。与传统采用视觉传感器捕捉焊缝图像并利用焊缝灰度梯度特征进行焊缝边缘检测方法不同,通过磁光传感器采集焊接区域紧密对接焊缝图像,焊件在磁场激励器作用下形成反映焊缝位置特征的磁光图像,利用焊缝位置和母材的磁光成像纹理差异识别焊缝位置。对磁光图像的待焊区和母材区进行纹理分析,提取纹理特征,再利用纹理特征差别,将磁光图形划分为若干子区域,进行图像分割,确定焊缝区域。试验结果表明,该方法能够精确地识别微间隙紧密对接焊缝。A method to recognize the close butt weld(seam gap width was less than 0.1mm) was proposed based on magneto- optical image texture features.Unlike the traditional vision methods of weld seam detection,a magnetic optical sensor was adopted to collect the images of welding areas,magneto- optical images reflected the seam feature were formed under the action of the magnetic field exciter,the texture features of the magneto optic images were analyzed to distinguish the weld seam region and material region.Texture of the seam region and material region were analyzed,and the gray characteristics and gray level co- occurrence matrix texture features were calculated to achieve seam segmentation.Experimental results show that the proposed method can accurately identify the weld seam.

关 键 词：微间隙 磁光成像 纹理 灰度共生矩阵 

分 类 号：TG409[金属学及工艺—焊接]