焊接电弧图像的边缘检测及其批处理算法  

Edges detection and batch algorithm for welding arc images

作　　者：于英飞 朱志明[1,2] 孙博文 程世佳 

机构地区：[1]清华大学机械工程系,北京100084 [2]清华大学先进成形制造教育部重点实验室,北京100084

出　　处：《焊接学报》2018年第11期17-21,共5页Transactions of The China Welding Institution

基　　金：国家自然科学基金资助项目(51075231)

摘　　要：借助高速摄像捕捉不同时刻的焊接电弧图像,通过图像处理识别其边缘并提取其特征参数,可以实现对电弧燃烧和焊接过程稳定性的量化评价.在分析电弧图像灰度特征的基础上,基于传统图像处理和形态学图像处理,提出了电弧图像边缘辨识的2种简化算法,并从程序运行时间和边缘检测精度两方面,对比了传统图像处理的Sobel边缘检测算法与形态学边缘检测算法的优劣.选取传统图像处理简化算法构建了图像批处理程序,通过在原始电弧图像中添加随机高斯噪声和对TIG-MIG复合电弧图像边缘的准确识别来检验程序的性能.结果表明,所构建的图像批处理程序有良好抗干扰能力和适用性.By capturing the welding arc images at different time with the help of high-speed camera, and then recognizing their edges and extracting their characteristic parameters via images processing, the quantitative evaluation on the stability of arc burning and welding process could be achieved. On the basis of analyzing the grey characteristics of arc images, two simplified algorithms achieving the edge recognition of arc image were proposed based on the traditional and morphological image processing respectively. And the merits of Sobel edge detection algorithm in traditional image processing and the edge detection method in morphological image processing were compared from the two aspects of the program run time and the edge detection precision of arc image. According to the comparison result, the simplified algorithm of traditional image processing was chosen and applied into the construction of image batch processing algorithm. The performance of the program was tested by adding random gaussian noise to the original arc image andaccurately identifying the edge of TIG-MIG hybrid arc image.The results showed that the image batch processing algorithm had good anti-interference ability and applicability.

关 键 词：电弧形态 高速摄像 电弧图像 边缘检测 

分 类 号：TG403[金属学及工艺—焊接]