基于机器视觉的汽车漆面缺陷检测技术  

Vehicle Paint Defect Detection Technology Based on Machine Vision

作　　者：赵健 付琴 Zhao Jian;Fu Qin(Shanghai Nio Automobile Co.,Ltd.,Shanghai,201805;Hi-P International Co.,Ltd.,Shanghai 201313)

机构地区：[1]上海蔚来汽车有限公司,上海201805 [2]赫比国际有限公司,上海201313

出　　处：《汽车工艺与材料》2022年第7期16-19,共4页Automobile Technology & Material

摘　　要：对基于机器视觉的汽车漆面缺陷检测技术进行了深入研究,分析了隧道式和机器人式缺陷检测系统的优劣势,比较了传统图像算法和深度学习算法的异同点,并通过应用案例来展示漆面缺陷检测系统的运行状态和经济效益,为有意向应用该项技术的企业提供指导和帮助。在实际应用中,缺陷识别准确率可以达到98.5%,一套系统可节约8名操作人员,完全满足汽车漆面缺陷检测的要求,投资回报率高。This paper made in-depth research on the detection technology of automobile paint surface defects,analyzed the advantages and disadvantages of frame-based and robot-based defect detection systems,compared the similarities and differences between traditional image algorithms and deep learning algorithms,and demonstrated the operating status and economic benefits of the paint defect detection system with specific application cases,providing guidance and assistance for companies intending to apply this technology.In practical applications,the defect recognition accuracy rate can reach 98.5%,and a set of system can save 8 labors,fully meet the requirements of automobile paint surface defect detection,and the return on investment is high.

关 键 词：汽车漆面 缺陷检测 机器视觉 深度学习 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] U466[自动化与计算机技术—计算机科学与技术]