目的 随着计算机视觉技术的快速发展,缺陷检测在工业制造、质量控制尤其是卷烟生产领域变得越来越重要,为了解决传统的形态学缺陷检测方法存在检测效率低、泛化性差,检测结果容易受主观因素影响的问题,提出了一种双通道残差的表面缺陷检测方法。方法 首先,根据正样本降采样图像与缺陷样本降采样插值恢复图像在颜色、纹理分布上的相似性,设计图像预处理模块,有效增强缺陷区域的对比差异。其次,对双通道的残差网络结构进行设计,利用残差连接增强网络模型对缺陷区域的特征提取能力,并结合层间跳跃连接缓解深层特征梯度消失的问题。最后,采用LBP算法对两路模型的预测结果计算像素距离差异,经过粗粒度筛选与细粒度验证实现缺陷定位。结果 多种场景的实验结果表明,所提方法在像素级检测精度中可以获得98%的检测精度,表现出较高的检测准确性和泛化性,并在实际的生产流水线中表现出优越的性能。结论 所提方法有效减少了特征提取过程中对缺陷无关区域的依赖,强化了缺陷区域高频信息的提取,进一步提升实际产线中缺陷检测的效率。
Abstract
With the rapid development of computer vision technology, defect detection has become increasingly important in industrial manufacturing, quality control, and especially in cigarette production. The work aims to propose a dual-channel residual surface defect detection method to address the issues of low detection efficiency, poor generalization, and susceptibility to subjective factors in traditional morphological defect detection methods. Firstly, based on the similarity in color and texture distribution between downsampled images of positive samples and downsampled and interpolated images of defect samples, an image preprocessing module was designed to effectively enhance the contrast difference in defect regions. Secondly, a dual-channel residual network structure was designed, utilizing residual connections to enhance the model's feature extraction capability for defect regions, while incorporating skip connections to mitigate the issue of gradient vanishing in deep features. Finally, the LBP algorithm was employed to calculate pixel distance differences between the prediction results of the two models, and defect localization was achieved through coarse-grained screening and fine-grained verification. Experimental results across various scenarios demonstrated that the proposed method achieved a pixel-level detection accuracy of 98%, showcasing high detection accuracy and generalization capability, as well as superior performance in real-world production lines. The proposed method effectively reduces the reliance on non-defect regions during feature extraction, strengthens the extraction of high-frequency information in defect regions, and further improves the efficiency of defect detection in actual production lines.
关键词
先验图像 /
语义相似性 /
孪生网络 /
残差网络 /
缺陷检测
Key words
prior image /
semantic similarity /
twin network /
residual network /
defect detection
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 周茂军, 胡江涛, 王俊杰, 等. 改进SSD算法在工件表面缺陷检测中的应用[J]. 组合机床与自动化加工技术, 2024(8): 139-144.
ZHOU M J, HU J T, WANG J J, et al.Application of Improved SSD Algorithm in Surface Defect Detection of Workpieces[J]. Modular Machine Tool & Automatic Manufacturing Technique, 2024(8): 139-144.
[2] 任伟建, 陈明文, 康朝海, 等. 基于YOLOv3的金属表面缺陷检测研究[J]. 控制工程, 2024, 31(7): 1219-1228.
REN W J, CHEN M W, KANG C H, et al.Research on Metal Surface Defect Detection Based on YOLOv3[J]. Control Engineering of China, 2024, 31(7): 1219-1228.
[3] ZHANG Z, YANG B K, MA J.Multiple Constraints Flow for Weakly Observable Defect Detection Based on Defect-Free Samples[J]. IEEE Transactions on Instrumentation and Measurement, 2024, 73: 3541513.
[4] LIU T H, HE Z S, LIN Z J, et al.An Adaptive Image Segmentation Network for Surface Defect Detection[J]. IEEE Transactions on Neural Networks and Learning Systems, 2024, 35(6): 8510-8523.
[5] 乔辉, 曾勇, 刘德志, 等. 基于机器视觉的轮对磁粉探伤裂纹检测方法研究[J]. 现代制造工程, 2024(6): 128-134.
QIAO H, ZENG Y, LIU D Z, et al.Research on Crack Detection Method of Magnetic Particle Detection Based on Machine Vision[J]. Modern Manufacturing Engineering, 2024(6): 128-134.
[6] 张黎明, 贵宁, 李国维, 等. 支盘桩外轮廓跨孔超声波检测现场试验研究[J]. 公路交通科技, 2024, 41(7): 93-100.
ZHANG L M, GUI N, LI G W, et al.In-Situ Experimental Study on Squeezed Branch Pile Outer Profile Cross-Hole Ultrasonic Detection[J]. Journal of Highway and Transportation Research and Development, 2024, 41(7): 93-100.
[7] 郝玲, 潘一, 何建樑, 等. 刀具损伤视觉检测系统自动对准对焦研究[J]. 制造技术与机床, 2024(8): 72-77.
HAO L, PAN Y, HE J L, et al.Research on Automatic Alignment and Focusing of Tool Damage Visual Detection System[J]. Manufacturing Technology & Machine Tool, 2024(8): 72-77.
[8] 李洋, 吴晓群. 面向RGB-D图像的多层特征提取算法综述[J]. 中国图象图形学报, 2024, 29(5): 1346-1363.
LI Y, WU X Q.Survey of Multilevel Feature Extraction Methods for RGB-D Images[J]. Journal of Image and Graphics, 2024, 29(5): 1346-1363.
[9] 马志艳, 万海迪, 陈学海, 等. 基于RGB-D双目视觉的苗期玉米三维模型重构方法研究[J]. 中国农机化学报, 2024, 45(8): 148-153.
MA Z Y, WAN H D, CHEN X H, et al.Research on the Reconstruction Method of Maize Three-Dimensional Model in Seedling Stage Based on RGB-D Binocular Vision[J]. Journal of Chinese Agricultural Mechanization, 2024, 45(8): 148-153.
[10] 丁海洋, 董明利, 刘陈华, 等. 基于显著性自适应权重图的红外可见光图像融合[J]. 激光与光电子学进展, 2024, 61(10): 1037008.
DING H Y, DONG M L, LIU C H, et al.Infrared and Visible Image Fusion Based on Saliency Adaptive Weight Map[J]. Laser & Optoelectronics Progress, 2024, 61(10): 1037008.
[11] ZHAO Y H, LIU Q, SU H, et al.Attention-Based Multiscale Feature Fusion for Efficient Surface Defect Detection[J]. IEEE Transactions on Instrumentation and Measurement, 2024, 73: 5013310.
[12] DŽAKOVIĆ M, JOVANČEVIĆ I, DOŠLJAK V, et al. Defect Characterization on Complex Shape Aeronautical Parts via 3D Point Cloud Processing[C]// 2024 13th Mediterranean Conference on Embedded Computing (MECO). Budva, Montenegro. IEEE, 2024: 1-5.
[13] LU H T, FANG M Y, QIU Y, et al.An Anchor-Free Defect Detector for Complex Background Based on Pixelwise Adaptive Multiscale Feature Fusion[J]. IEEE Transactions on Instrumentation and Measurement, 2022, 72: 5002312.
[14] MARTINI M.A Simple Relationship between SSIM and PSNR for DCT-Based Compressed Images and Video: SSIM as Content-Aware PSNR[C]// 2023 IEEE 25th International Workshop on Multimedia Signal Processing (MMSP). Poitiers, France. IEEE, 2023: 1-5.
[15] 苗宇, 雷龙威, 刘泽. 基于CGAN的EMT钢轨伤损检测图像处理方法[J]. 北京理工大学学报, 2024, 44(12): 1287-1296.
MIAO Y, LEI L W, LIU Z.Image Optimization Algorithm Based on CGAN for Reconstructed Images of Rail Damage by EMT[J]. Transactions of Beijing Institute of Technology, 2024, 44(12): 1287-1296.
[16] LAN R, YU X Y, WANG B, et al.IFCM Clustering Algorithm Incorporating Fuzzy Robust Texture Feature for Complex Texture Image Segmentation[C]// 2024 6th International Conference on Natural Language Processing (ICNLP). Xi'an, China. IEEE, 2024: 245-252.
[17] BERGMANN P, JIN X, SATTLEGGER D, et al.The MVTec 3D-AD Dataset for Unsupervised 3D Anomaly Detection and Localization[C]// Proceedings of the 17th International Joint Conference on Computer Vision, Imaging and Computer Graphics Theory and Applications. Online Streaming, Select a Country. SCITEPRESS - Science and Technology Publications, 2022: 202-213.
[18] KIM J H, KWON G R.Unsupervised Visual Anomaly Detection Using Self-Supervised Pre-Trained Transformer[J]. IEEE Access, 2024, 12: 127604-127613.
[19] ZHONG Y M, LING Z G, LIU L, et al.Deep Gaussian Attention Network for Lumber Surface Defect Segmentation[J]. IEEE Transactions on Instrumentation and Measurement, 2024, 73: 5015512.
[20] LI K J, TANG Y S, PRISACARIU V A, et al.BNV-Fusion: Dense 3D Reconstruction Using Bi-Level Neural Volume Fusion[C]// IEEE/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, LA, 2022: 6156-6165.
[21] RUDOLPH M, WEHRBEIN T, ROSENHAHN B, et al.Asymmetric Student-Teacher Networks for Industrial Anomaly Detection[C]// 2023 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). Waikoloa, HI, USA. IEEE, 2023: 2591-2601.
[22] CHU Y M, LIU C, HSIEH T I, et al.Shape-Guided Dual-Memory Learning for 3D Anomaly Detection[C]// Proceedings of the 40th International Conference on Machine Learning (ICML 2023). 2023: 6185-6194.
[23] HORÉ A, ZIOU D.Image Quality Metrics: PSNR Vs. SSIM[C]// 2010 20th International Conference on Pattern Recognition. Istanbul, Turkey. IEEE, 2010: 2366-2369.
基金
四川中烟工业有限责任公司科研项目（JL/SCZYGSJ003-01）
PDF(4055 KB)
渝公网安备 50010702501716号 渝ICP备15012534号-2
