The welding technology is an indispensable process technology of industrial production. Traditional welding quality control method, which consists of making process before welding and detecting the weld joint after welding, can not meet the demand of high-quality and precision welding. By using optical sensors, the image of weld pool, weldment and weld seam can be captured, the characters of the targets can be extracted by image processing algorithm, finally the real size and shape of the targets can be extracted by three-dimensional reconstruction algorithm, which is beneficial to extraction of welding quality information, on-line adjustment of welding process and development of remote robot welding. In this paper, the common three-dimensional methods, which is the monocular vision, binocular vision and multi-view stereo vision t methods were firstly analyzed, and then the overview of three-dimensional researches in welding, especially in gas metal arc welding (GMAW) process, was summarized. The three-dimensional reconstruction status of weld groove, weld seam and weld pool in GMAW was summarized in which the Structured-Light method, the Shape-From-Shading method and the Binocular-Stereo method were utilized. At last the shortcomings of three-dimensional reconstruction methods in welding mentioned above were analyzed. The complement of the reconstruction methods and the improvement of efficiency and accuracy of the reconstruction results in welding should be the focus of follow-up study.%焊接技术是工业生产不可或缺的工艺技术手段,传统焊前制定工艺、焊后品质检测的品质控制方法已经不能满足高效高质精密焊接的需求.利用光学传感器对熔池、工件、接缝等目标特征进行传感,通过三维重建方法获得目标的三维尺寸和形状,有利于提取焊接过程的品质信息并实现在线调整和机器人遥控焊接.文章分析了单目视觉、双目视觉和多目视觉三维重建方法,归纳了国内外焊接领域三维重建研究应用现状,重点针对弧焊特别是气体保护焊,综述了结构光、阴影恢复法和双目立体法三种焊接接缝坡口和熔池三维重建的研究现状,并分析了焊接领域中三维重建研究存在的问题,提出了完善重建方法、提高重建准确度和效率是今后研究的重点.
展开▼
