摘要
针对弧焊中的电弧、飞溅等强光干扰,提出了一种结构光焊缝跟踪实时图像处理技术,研制出焊缝跟踪视觉传感器.首先对焊缝特征区域进行识别和捕获,提取焊缝特征模板,利用焊缝序列图像的相关性对焊缝位置变化进行观测,并对其后验概率进行评价;在此基础上,采用粒子滤波对焊缝实现实时跟踪.结果表明,其相关品配能够有效地消除弧光、飞溅的干扰,其焊缝特征模板的实时更新使得焊缝跟踪能够适应形状变化的焊缝,其特征区域约束和快速相关算法能够有效地提高跟踪的实时性.将所研究的系统在PC104(PIII 1.6 GB)处理平台上,于弧焊接距离传感器20mm且无遮挡情况下进行实验测试,结果发现其跟踪效果良好,每帧处理速度达10 ms,焊缝跟踪精度达0.5 mm.
In view of the strong disturbs of arc glare and splash during welding process,a real-time image processing technique for welding track was proposed. In this technique,a welding characteristic template is extracted in the pro-recognized welding region,and the correlation between this template and the following serial welding images is calculated to find the position variation of the welding joint.Based on this,the particle filtering method is employed to give a real-time welding track.With the feature of cor...
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
2008年第S1期61-64,共4页
Journal of Shanghai Jiaotong University
基金
先进成形制造教育部重点实验室开放基金资助项目(2006011)
国家高技术研究发展计划(863计划)资助项目(2007AA04Z242)
关键词
焊缝跟踪
结构光
粒子滤波
图像处理
实时
welding track
structure light
particle filtering
image process
real-time
