基于区域相关哈希编码的面板螺钉装配质量机器视觉检测系统  

Machine Vision Inspection System for Panel Screw Assembly Quality Based on Region Related Hash Coding

作　　者：崔海荣[1] 梁晨 CUI Hairong;LIANG Chen(Xianyang normal universty,xianyang shaanxi 712000,China;Xi’an Meilechen Education Technology Co.,Ltd.,Xi’an 710000,China)

机构地区：[1]咸阳师范学院,陕西咸阳712000 [2]西安美乐宸教育科技有限公司,西安710000

出　　处：《自动化与仪器仪表》2024年第6期89-93,共5页Automation & Instrumentation

摘　　要：面板螺钉的装配质量对产品的性能和外观质量有着重要的影响。针对面板螺钉的装配质量检测问题,研究引入了改进的基于区域的快速卷积神经网络(Faster Region-based Convolutional Neural Network, Faster R-CNN)算法以及基于螺钉视觉检测的启发式哈希匹配算法。结果表明,研究所提算法的平均准确率均稳定在99.3%以上,且波动范围不超过0.3%。与Shape-Based matching算子相比,三种钢琴面板螺钉的平均检测ACC分别提升了6.04%、2.65%、4.25%。说明双特征融合的改进Faster R-CNN模型对钢琴面板螺钉的目标检测具有显著性能优势,并且基于螺钉视觉检测的启发式哈希匹配算法能够准确识别钢琴面板螺钉装配情况,极大地提高了检测效率及精度。其对于提高装配质量、生产效率和产品安全性,降低生产成本具有重要的现实意义。The assembly quality of panel screws has an important impact on the performance and appearance quality of products.In response to the issue of assembly quality detection for panel screws,an improved Fast Region based Convolutional Neural Network(Faster R-CNN)algorithm and a heuristic hash matching algorithm based on screw visual detection were introduced.The results show that the average accuracy of the proposed algorithms is stable at over 99.3%,and the fluctuation range does not exceed 0.3%.Compared with the Shape Based matching operator,the average detection ACC of the three piano panel screws increased by 6.04%,2.65%,and 4.25%,respectively.The improved Faster R-CNN model with dual feature fusion has significant performance advantages in target detection of piano panel screws,and the heuristic hash matching algorithm based on screw visual detection can accurately identify the assembly situation of piano panel screws,greatly improving detection efficiency and accuracy.It has important practical significance for improving assembly quality,production efficiency,product safety,and reducing production costs.

关 键 词：区域相关哈希编码 面板螺钉 装配质量 机器视觉检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]