基于条纹投射三维测量的铆钉自动化检测技术

An Automatic 3D Rivet Detection Technique Based on Fringe Projection

针对大型薄壁结构铆接点位自动化检测问题,提出了基于条纹投射三维测量的铆钉检测技术,实现了铆钉镦头尺寸特征高精度测量以及裂纹缺陷自动化识别。在传统条纹投射三维测量的基础上,引入高动态范围(HDR)纹理合成,提出二、三维结合的点云分割策略,实现铆钉尺寸特征高效提取。搭建深度学习神经网络,实现镦头表面缺陷识别。本文搭建了系统原理样机,以铆接桁条样品和标准器作为样件进行系统实验验证。结果表明:实验室条件下,该方法直径测量精度为0.040 mm,高度测量精度为0.013 mm,缺陷识别准确率可达99.30%。与传统方法相比,本文提出的方法更加易于集成,效率高,具有广泛应用价值。

Aiming at the problems of automatic detection for riveted large thin-wall structures,a rivet detection technique based on fringe projection is proposed,which enables high-accuracy dimension measurement and automatic crack detection.Based on the traditional three-dimensional(3D)fringe projection measurement,high dynamic range(HDR)texture merging is introduced,and a point cloud segmentation strategy combining two-dimensional(2D)and 3D information is proposed,with which the rivet size features can be efficiently extracted.A deep-learning neural network is built to achieve the intelligent detection of rivet cracks.Besides,a prototype of the system principle is built,and the riveted samples and standards are measured experimentally to verify the effectiveness and accuracy.The results show that under laboratory conditions,the accuracy of diameter measurement can achieve 0.040 mm,the accuracy of height measurement is 0.013 mm,and the accuracy of crack detection can reach 99.30%.Compared with traditional methods,the proposed method is easier to integrate,has high detection efficiency,and has extensive application value.