一种基于结构光视觉的焊缝跟踪特征提取方法  

A method of feature extraction for joint tracking with structured-light vision sensing

作　　者：龚烨飞[1] 戴先中[1] 李新德[1] 朱伟 廖建雄 

机构地区：[1]东南大学复杂工程系统测量与控制教育部重点实验室,江苏南京210096 [2]昆山工研院工业机器人研究所,江苏昆山215300 [3]江苏省焊接自动化工程技术研究中心,江苏昆山215301

出　　处：《电焊机》2009年第4期51-55,共5页Electric Welding Machine

基　　金：江苏省重大科技成果转化资金资助项目(BA2007058)

摘　　要：对基于结构光视觉的焊缝跟踪特征提取方法进行了研究。首先通过逐列分割结构光条纹灰度波峰并计算其重心位置实现对结构光条纹中心线的提取,其中采用自适应阈值分割和波峰特征距离计算相结合的方法以保证提取的可靠性;其次利用逐点递增式的分段直线段最小二乘拟合提取了接头轮廓的形状特征,进一步利用预先定义的基元符号实现了接头轮廓的形状描述;最后在此基础上实现了对焊缝接头的识别和跟踪点的提取。分别进行了对接、搭接以及角接焊缝实验,验证了本方法的有效性。The method of weld feature extraction in the structured-light vision sensing seam tracking system has been studied. Centerline points of structured-light stripe were extracted sequentially by calculating the mass position of gray peaks,which were segmented reliably by a combination of adaptive threshold method and peak＇s feature distance calculation,and a incremental piecewise line segments fitting using the least squares method was proposed to extract the shape of joint profile,which was further described qualitatively by the predefined primitive labels,and lastly weld joint was recognized by siring matching and tracking point was acquired by rules.Experiments for butt, lap, and fillet weld joints are conducted to verify the effectiveness.

关 键 词：焊缝接头 结构光视觉 图像处理 模式识别 

分 类 号：TG409[金属学及工艺—焊接] TP242.6[自动化与计算机技术—检测技术与自动化装置]