管道焊接熔池图像处理与特征提取  

Image processing and features extraction of molten pool for pipe welding

作　　者：李原[1] 王庆林[1] 徐德[2] 谭民[2] 

机构地区：[1]北京理工大学信息科学技术学院,北京100030 [2]中国科学院自动化研究所复杂系统与智能科学重点实验室,北京100190

出　　处：《焊接学报》2008年第8期57-60,共4页Transactions of The China Welding Institution

基　　金：国家"863"高技术研究发展计划资助项目(2006AA04Z213);中国博士后科学基金项目资助(20070410463)

摘　　要：提出了一种管道焊接熔池成像系统,能够有效克服弧光干扰,获取焊接熔池在焊缝坡口内图像。采用基于统计的鲁棒图像处理方法,对熔池图像进行预处理,抑制飞溅等噪声。在图像自适应分割和边缘提取基础上,根据熔池图像和焊缝坡口图形特点,分别提取熔池图像在焊缝内的灰度分布特征以及焊缝坡口图形特征,得到熔池相对焊缝坡口的动态偏移量,以及熔池振动幅度和频率,为焊缝跟踪和焊接质量控制提供了视觉信息反馈。进行了焊接过程实时图像处理和特征提取试验,试验结果验证了熔池成像系统和图像处理方法的有效性和可靠性。A pool imaging system is designed for pipe welding, which can capture clear images of molten pool in grooves and reduce the disturbances of arc light. Robust statistics method was adopted for pool images preprocessing, and the splash in images was inhibited. After segmentation and edge detection, intensity distribution of pool images and welding groove figures are selected as image features and extracted. Then the deviation of the pool and groove is detected, and the vibration amplitude and frequency of molten pool are calculated. Experiments of welding with image processing and features extraction are conducted to verify the effectiveness of the imaging system and the proposed method.

关 键 词：管道焊接 熔池图像处理 特征提取 焊缝跟踪 

分 类 号：TG409[金属学及工艺—焊接]