计算机科学

计算机科学 ›› 2023, Vol. 50 ›› Issue (11A): 230100073-7.doi: 10.11896/jsjkx.230100073

• 图像处理&多媒体技术 • 上一篇    下一篇

基于模糊遮蔽与动态推理的生成式工业异常定位模型

吴天月1, 张辉2, 张邹铨1, 唐珺琨1   

  1. 1 长沙理工大学电气与信息工程学院 长沙 410000
    2 湖南大学机器人学院 长沙 410000
  • 发布日期:2023-11-09
  • 通讯作者: 张辉(zhanghuihby@126.com)
  • 作者简介:(yue__wuwuwu@163.com)
  • 基金资助:
    科技创新2030－“新一代人工智能”重大项目课题(2021ZD0114503);国家自然科学基金(61971071,62027810);湖南省杰出青年科学基金项目(2021JJ10025);湖南省研究生科研创新项目(CX20210797)

Generative Industrial Image Abnormal Location Model Based on Fuzzy Masking and Dynamic Inference

WU Tianyue1, ZHANG Hui2, ZHANG Zouquan1, TANG Junkun1   

  1. 1 School of Electrical & Information Engineering,Changsha University of Science and Technology,Changsha 410000,China
    2 School of Robotics,Hunan University,Changsha 410000,China
  • Published:2023-11-09
  • About author:WU Tianyue,born in 1999,postgraduate,is a member of China Computer Federation.Her main research interests include image processing,self-supervised learning,and anomaly detection.
    ZHANG Hui,born in 1983,Ph.D,professor.His main research interests are machine vision,sparse representation,and visual tracking.
  • Supported by:
    National Key R&D Program of China(2021ZD0114503),National Natural Science Foundation of China(61971071,62027810),National Science Found for Distinguished Young Scholars of Hunan Province,China(2021JJ10025) and Postgraduate Scientific Research Innovation Project of Hunan Province(CX20210797).

PDF (PC)

169

摘要： 工业生产机械化对工业产品质量检测环节提出了新的要求,需要一种具有高精度、易于移植的异常检测算法来适应生产方式的更新。针对工业生产中,异常样本出现概率低、无法完全预测的固有难题,提出了一种基于模糊遮蔽与动态推理的生成式工业异常定位模型。首先,设计了一个基于随机模糊遮蔽的对比样本生成模块,用于获取高质量的模拟异常图像。同时,利用浅层特征融合路径保留更多的边缘信息,使用损失函数加权使模型更加关注结构相似性,以及使用对比学习的方式使网络获得更好的表示能力。其次,为了缓解生成式模型输出图像模糊的问题,设计了多分支异常动态推理方法,使迭代生成和精准修复两分支相互配合,拉远背景噪声与真实异常间的距离。实验结果表明,所提方法在MVTec数据集上取得了91.42%的平均定位精度,其中有12类达到了前三的异常定位精度,能够较完整地获取异常地位置;对于纹理复杂和背景占比较大的图像,所提方法仍然保持着较高的指标敏感度,其异常定位性能在近年来提出的生成式检测模型中取得了最佳。

关键词: 工业图像检测, 异常定位, 深度学习, 生成式学习, 动态异常推理

Abstract: The mechanization of industrial production puts forward new requirements for the inspection of industrial product quality,and a high-precision,easy-to-transplantanomaly detection algorithm is required to adapt to the update of production methods.Aiming at the inherent problem of low probability of abnormal samples in industrial production and incomplete prediction,a generative industrial anomaly localization model based on fuzzy masking and dynamic reasoning is proposed.Firstly,a contrast sample generation module based on random fuzzy occlusion is designed to obtain high-quality simulated anomalous images.At the same time,the shallow feature fusion path is used to retain more edge information,the loss-loss function weighting is used to make the model pay more attention to structural similarity,and the contrast learning method is used to make the network obtain better representation ability.Secondly,in order to alleviate the problem of blurred output images of generative models,a multi-branch anomaly dynamic inference method is designed,and the two branches of iterative generation and accurate repair cooperate with each other to widen the distance between background noise and real anomalies.Experimental results show that the proposed method achieves an average localization accuracy of 91.42% on the MVTec dataset,and the top three anomalous localization accuracy are obtained in 12 classes.The location of anomalies can be obtained more completely.For images with complex textures and large backgrounds,it still maintains high index sensitivity,and the average anomaly localization performance has reached the best in published generative detection models published in recent years.

Key words: Industrial image detection, Anomaly location, Deep learning, Generative learning, Dynamic anomaly inference

中图分类号: 

  • TP3
[1]TAO X,GONG X,ZHANG X,et al.Deep Learning for Unsupervised Anomaly Localization in Industrial Images:A Survey[J].IEEE Transactions on Instrumentation and Measurement,2022,71:1-21.
[2]LV C,SHEN F,ZHANG F.Review of Image Anomaly Detection[J].Acta Automatica Sinica,2022,48(6):1402-1428.
[3]CHOI J,KIM C.Unsupervised detection of surface defects:A two-step approach[C]//2012 19th IEEE International Conference on Image Processing.2012:1037-1040.
[4]BERGMANN P,FAUSER M,SATTLEGGER D,et al.MVTec AD--A comprehensive real-world dataset for unsupervised anomaly detection[C]//Proceedings of the IEEE/CVF Confe-rence on CVPR.2019:9592-9600.
[5]CHEN Y,TIAN Y,PANG G,et al.Deep One-Class Classification via Interpolated Gaussian Descriptor[J/OL].(2022-05-24)[2022-07-31].https://arxiv.org/pdf/2101.10043v5.pdf.
[6]HAN J,CHENG J F,LI Y,et al.Self-supervised Deep Clustering Algorithm Based on Self-attention[J].Computer Science,2022,49(3):134-143.
[7]GONG D,LIU L,LE V,et al.Memorizing normality to detectanomaly:Memory-augmented deep autoencoder for unsupervised anomaly detection[C]//Proceedings of the IEEE International Conference on Computer Vision.2019:1705-1714.
[8]PARK H,NOH J,HAM B.Learning memory-guided normality for anomaly detection[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2020:14372-14381.
[9]ZHOU K,LI J,XIAO Y,et al.Memorizing Structure-Texture Correspondence for Image Anomaly Detection[J].IEEE Transactions on Neural Networks and Learning Systems,2021,33(6):2335-2349.
[10]YANG L,JIANG A L,QIANG Y.Structure Preserving Unsupervised Feature Selection Based on Autoencoder and Manifold Regularization[J].Computer Science,2021,48(8):53-59.
[11]VZ A,MK A,DS A.Reconstruction by inpainting for visualanomaly detection[J].Pattern Recognition,2021,112(2):107706.
[12]LI Z,LI N,JIANG K,et al.Superpixel masking and inpainting for self-supervised anomaly detection[C]//British Machine Vision Conference.2020:7-10.
[13]YAN X,ZHANG H,XU X,et al.Learning semantic contextfrom normal samples for unsupervised anomaly detection[C]//Proceedings of the AAAI Conference on Artificial Intelligence.2021:3110-3118.
[14]WANG S,WU L,CUI L,et al.Glancing at the Patch:Anomaly Localization with Global and Local Feature Comparison[C]//Computer Vision and Pattern Recognition.2021:254-263.
[15]SALEHI M,SADJADI N,BASELIZADEH S,et al.Multiresolution Knowledge Distillation for Anomaly Detection[C]//Computer Vision and Pattern Recognition.2021:14897-14907.
[16]XING P,JIANG X,TANG J H,et al.Feature Consistent Restricted Distillation Learning for Visual Anomaly Detection[J/OL].Journal of Software.(2021-10-11)[2022-07-31].http://jos.org.cn/jos/article/abstract/Lf051.
[17]LI C L,SOHN K,YOON J,et al.Cutpaste:Self-SupervisedLearning for Anomaly Detection and Localization[C]//Proceedings of the IEEE International Conference on Computer Vision.Nashville:IEEE Press,2021:9664-9674.
[18]AHORÉ,ZIOU D.Image quality metrics:PSNR vs.SSIM[C]//20th International Conference on Pattern Recognition(ICPR 2010).IEEE Computer Society,2010:2366-2369.
[19]PERERA P,NALLAPATI R,BING X.OCGAN:One-ClassNovelty Detection Using GANs With Constrained Latent Representations[C]//2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition.2019:2898-2906.
[20]DEHAENE D,FRIGO O,COMBREXELLE S,et al.Iterative energy-based projection on a normal data manifold for anomaly localization[J/OL].(2020-12-10)[2022-10-25].https://arxiv.org/abs/2002.03734v1.
[21]HUANG C Q,XU Q W,WANG Y F,et al.Self-Supervised Masking for Unsupervised Anomaly Detection and Localization[J].IEEE Transactions on Multimedia.doi:10.1109/TMM.2022.3175611.
Viewed
Full text


Abstract

Cited
  Shared   
  Discussed   
No Suggested Reading articles found!
渝ICP备16013121号-4
地址：重庆市渝北区洪湖西路18号    邮编：401121  
电话：023-63500828（编辑部） 023-67039612（会议/专题） 023-67039623（财务/发票）
E-mail：jsjkx12@163.com
本系统由北京玛格泰克科技发展有限公司设计开发