基于双目视觉的钣金件边缘检测技术研究与系统开发  

Technique Research and System Development for Sheet Metal Edge Inspection Based on Binocular Vision

作　　者：刘之远 张丽艳[1] LIU Zhiyuan;ZHANG Liyan(College of Mechanical and Electrical Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)

机构地区：[1]南京航空航天大学机电学院,南京210016

出　　处：《航空制造技术》2020年第7期30-39,共10页Aeronautical Manufacturing Technology

摘　　要：针对钣金类零件边缘的快速精确检测需求,设计并开发了一套基于双目视觉的钣金件边缘检测原型系统(SMEIS)。手持式测量装置围绕表面贴有圆形标记点的被测钣金件边缘连续移动,同时测量装置中的线激光发射器向钣金件边缘投射激光条纹,双目相机实时获取同步图像并传输至计算机;系统软件通过并行处理模式实时对输入图像中的激光条纹中心点进行增量式三维重建,实现对钣金类零件边缘的高效检测。阐述了系统的工作流程和软、硬件结构,并对其中的硬件结构布局、激光条纹中心点实时提取以及三维测点实时拼接等关键技术作了详细讨论。对实际钣金零件边缘的现场检测试验结果表明,SMEIS系统的检测速度大于30fps,且获得的点云数据质量良好。对1mm厚度的平面侧壁检测试验表明,SMEIS系统的平均检测误差约为0.04mm,标准差约为0.03mm。Aiming at the real-time and precise inspection of sheet metal part edges,a binocular vision-based inspection system that mainly consists of two CMOS cameras and multiple line laser transmitters is developed.The handheld inspection device is continuously moved around the edge of the measured sheet metal part with the circular marks on the surface,while the line laser transmitters project laser stripes to the edge of the sheet metal part.The cameras acquire the synchronized image pair in real time and transmit them to the computer.The system software performs incremental 3D reconstruction of the laser stripe centers on the sheet metal edge in real time via parallel processing.The workflow,software and hardware structure of the system are introduced.The key techniques such as real-time extraction of laser stripe center points and real-time registration of the 3D measurement points are discussed in detail.The inspection speed of the developed system is verified to be more than 30fps,and the measurement points on the sheet metal edges output by the system is of high quality.Moreover,experiments on a planar side wall with 1mm thickness shows that the mean inspection error of SMEIS system is about 0.04mm,and the standard deviation is about 0.03mm.

关 键 词：钣金件边缘 双目视觉 线激光 实时检测 三维重建 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TG115.28[自动化与计算机技术—计算机科学与技术] V262.3[金属学及工艺—物理冶金]