基于MFA-UNet的铜制螺纹零件外表面缺陷检测*

doi:10.16731/j.cnki.1671-3133.2024.05.015

现代制造工程 ›› 2024, Vol. 524 ›› Issue (5): 113-120.doi: 10.16731/j.cnki.1671-3133.2024.05.015

• 仪器仪表/检测/监控 • 上一篇下一篇

基于MFA-UNet的铜制螺纹零件外表面缺陷检测^*

马涛, 李敬兆

安徽理工大学计算机科学与工程学院,淮南 232001

收稿日期:2023-07-12 出版日期:2024-05-18 发布日期:2024-05-30
通讯作者: 李敬兆,博士,教授,博士生导师,主要研究方向为嵌入式系统、物联网和工业互联网。
作者简介:马涛,硕士研究生,主要研究方向为缺陷检测、图像分割。E-mail:matao52725@163.com
基金资助:
^*国家自然科学基金资助项目(51874010)

Copper threaded part surface defect detection algorithm based on MFA-UNet

MA Tao, LI Jingzhao

School of Computer Science and Engineering,Anhui University of Science and Technology, Huainan 232001,China

Received:2023-07-12 Online:2024-05-18 Published:2024-05-30

摘要/Abstract

摘要： 针对工业现场铜制螺纹零件外表面缺陷检测效率低和精度差的问题,提出一种融合多尺度特征与注意力的U型网络(Multi-Scale Features and Attention Fused UNet,MFA-UNet)模型的铜制螺纹零件外表面缺陷检测算法。首先,设计一种双路下采样模块,并行使用普通卷积和空洞卷积提升模型的特征提取能力;其次,在跳跃连接部分加入复合空间注意力模块,增强分割模型对空间信息和边缘信息的提取能力;然后,在上采样过程中加入压缩激励模块,提高模型的表达能力和特征选择能力;最后,提出一种相似度对比算法,比较分割图像和掩码图像的相似度,得到缺陷检测结果。实验表明,所提分割模型在铜制螺纹零件缺陷检测数据集上PA指标达到94.81 %,MIoU指标达到93.78 %;所提算法缺陷检测准确率达到98.9 %,满足工业现场的使用需求。

关键词: 零件缺陷检测, 图像分割, 注意力机制, 相似度对比

Abstract: In industrial settings, detecting surface defects on copper threaded parts often faces challenges of low efficiency and poor accuracy. To address this, it proposes a copper threaded part surface defect detection algorithm based on MFA-UNet (Multi-Scale Features and Attention Fused UNet). Firstly, a dual down sampling module is designed, utilizing both ordinary convolution and dilated convolution to enhance the model′s feature extraction capabilities. Secondly, a compound spatial attention module is integrated into the skip-connection part to improve the model′s ability to extract spatial and edge information. Subsequently, a squeeze and excitation module is incorporated during the upsampling process to enhance the model′s expressive power and feature selection ability. Lastly, it proposes a similarity comparison algorithm that measures the similarity between segmented images and mask images to obtain the defect detection results. Experimental results demonstrate that the proposed segmentation model achieves a PA of 94.81 % and an MIoU of 93.78 % on the copper threaded part defect detection dataset. The defect detection accuracy of the proposed algorithm reaches 98.9 %, meeting the requirements for industrial field applications.

Key words: part defect detection, image segmentation, attention mechanism, similarity comparison

中图分类号:

TP391

马涛, 李敬兆. 基于MFA-UNet的铜制螺纹零件外表面缺陷检测^*[J]. 现代制造工程, 2024, 524(5): 113-120.

MA Tao, LI Jingzhao. Copper threaded part surface defect detection algorithm based on MFA-UNet[J]. Modern Manufacturing Engineering, 2024, 524(5): 113-120.

参考文献

[1] 李凌,黎明,鲁宇明.基于模糊灰度共生矩阵与隐马尔可夫模型的断口图像识别[J].中国图象图形学报,2010,15(9):1370-1375.
[2] 张学武,吕艳云,丁燕琼,等.小波统计法的表面缺陷检测方法[J].控制理论与应用,2010,27(10):1331-1336.
[3] 李维刚,叶欣,赵云涛,等.基于改进YOLOv3算法的带钢表面缺陷检测[J].电子学报,2020,48(7):1284-1292.
[4] 刘孝保,张嘉祥,阴艳超,等.主从特征融合驱动的铝型材表面缺陷检测[J].计算机集成制造系统,2023,29(1):192-199.
[5] RONNEBERGER O, FISCHER P, BROX T. U-net: Convolutional networks for biomedical image segmentation[C]//Medical Image Computing and Computer-Assisted Interven-tion-MICCAI 2015: 18th International Conference, Berlin: Springer, 2015: 234-241.
[6] LONG J, SHELHAMER E, DARRELL T. Fully convolu-tional networks for semantic segmentation[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.Piscataway,NJ:IEEE,2015: 3431-3440.
[7] BADRINARAYANAN V, KENDALL A, CIPOLLA R. Segnet: A deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017,39(12):2481-2495.
[8] ZHAO H, SHI J, QI X, et al. Pyramid scene parsing network[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.Piscataway,NJ:IEEE,2017:2881-2890.
[9] ZHOU Z, RAHMAN S M M, TAJBAKHSH N, et al. Unet++: A nested u-net architecture for medical image segmentation[C]//Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support: 4th International Workshop, DLMIA 2018, and 8th International Workshop, ML-CDS 2018, Held in Conjunction with MICCAI 2018, Berlin: Springer, 2018:3-11.
[10] CHEN L C, PAPANDREOU G, KOKKINOS I, et al. Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017,40(4):834-848.
[11] DAMACHARLA P, RAO A, RINGENBERG J, et al. TLU-net: a deep learning approach for automatic steel surface defect detection[C]//2021 International Conference on Applied Artificial Intelligence (ICAPAI).Piscataway,NJ:IEEE, 2021:1-6.
[12] ZHENG Z, YANG H, ZHOU L, et al. HLU 2-Net: a residual U-structure embedded U-Net with hybrid loss for tire defect inspection[J]. IEEE Transactions on Instrumentation and Measurement, 2021,70:1-11.
[13] SHARMA M,LIM J,LEE H. The amalgamation of the object detection and semantic segmentation for steel surface defect detection[J]. Applied Sciences,2022,12(12):6004-6021.
[14] HU J, SHEN L, SUN G. Squeeze-and-excitation networks[C]//Proceedings of the IEEE conference on computer vision and pattern recognition.Piscataway,NJ:IEEE,2018: 7132-7141.
[15] YANG Z, ZHU L, WU Y, et al. Gated channel transformation for visual recognition[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.Piscataway,NJ:IEEE,2020: 11794-11803.
[16] WANG Q, WU B, ZHU P, et al. ECA-Net: Efficient channel attention for deep convolutional neural networks[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.Piscataway,NJ:IEEE,2020: 11534-11542.
[17] WOO S, PARK J, LEE J Y, et al. Cbam: Convolutional block attention module[C]//Proceedings of the European conference on computer vision (ECCV).Berlin:Springer,2018: 3-19.
[18] FU J, LIU J, TIAN H, et al. Dual attention network for scene segmentation[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.Piscataway,NJ:IEEE,2019: 3146-3154.

基于MFA-UNet的铜制螺纹零件外表面缺陷检测^*

Copper threaded part surface defect detection algorithm based on MFA-UNet

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 12

编辑推荐

Metrics

[1]	莫坚;张泽. 轴承剩余使用寿命的注意力多尺度卷积神经网络预测[J]. 现代制造工程, 2023, 515(8): 148-154.
[2]	王瑞，李敬兆，石晴，许志，王龙，赵雪琦，陈强. 基于融合多维非对称策略的螺纹缺陷检测方法[J]. 现代制造工程, 2023, 513(6): 107-116.
[3]	阎馨，杨月川，屠乃威. 基于改进SSD的钢材表面缺陷检测[J]. 现代制造工程, 2023, 512(5): 112-120.
[4]	王凯，吉卫喜，卢璟钰，苏璇. 基于改进GADF-SeResNet滚动轴承故障诊断方法[J]. 现代制造工程, 2023, 512(5): 135-142.
[5]	董维振;陈燕. 基于改进麻雀搜索算法的带钢面标识别方法[J]. 现代制造工程, 2023, 509(2): 84-91.
[6]	薛阳;丁凯;李清;杨江天;李金星. 基于改进YOLOv5s的活塞杆表面缺陷检测[J]. 现代制造工程, 2023, 518(11): 104-112.
[7]	杨泽青，王春方，卢晨光，陈英姝，刘丽冰. 基于深度卷积神经网络的曲轴智能识别方法研究[J]. 现代制造工程, 2022, 499(4): 102-111.
[8]	仇娇慧，贝绍轶，尹明锋，卿宏军. 基于改进YOLOv5s的齿轮表面缺陷检测[J]. 现代制造工程, 2022, 498(3): 104-113.
[9]	黄昌顺，张金萍. 基于CBAM-CNN的滚动轴承故障诊断方法[J]. 现代制造工程, 2022, 506(11): 137-143.
[10]	伍萍辉，陈新，张馨，段丹丹，唐冰霞，张云鹤. 基于聚类-改进灰狼算法的设施番茄分割识别方法[J]. 现代制造工程, 2021, 489(6): 83-89.
[11]	党顺峰, 熊锐, 李继辉, 陈灿奇, 陈振威, 吴鑫. 基于改进Mask-RCNN算法的车位检测研究[J]. 现代制造工程, 2021, 484(1): 91-97.
[12]	卢满怀, 宇文旋. 多神经网络图像分割算法[J]. 现代制造工程, 2018, 459(12): 130-136.