Unsupervised Anomaly Detection on Metal Surfaces Based on Frequency Domain Information Fusion

Metal products are widely used in industrial manufacturing, and the quality of metal products is becoming more and more demanding. At present, although there are many methods for detecting defects on metal surfaces, there are still various limitations. The limited number of defect samples, unpredict...

Full description

Saved in:
Bibliographic Details
Main Authors: Wenfei Wu, Tao Tao, Jinsheng Xiao, Yichu Yao, Jianfeng Yang
Format: Article
Language:English
Published: MDPI AG 2025-04-01
Series:Sensors
Subjects:
Online Access:https://www.mdpi.com/1424-8220/25/7/2250
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849730383790211072
author Wenfei Wu
Tao Tao
Jinsheng Xiao
Yichu Yao
Jianfeng Yang
author_facet Wenfei Wu
Tao Tao
Jinsheng Xiao
Yichu Yao
Jianfeng Yang
author_sort Wenfei Wu
collection DOAJ
description Metal products are widely used in industrial manufacturing, and the quality of metal products is becoming more and more demanding. At present, although there are many methods for detecting defects on metal surfaces, there are still various limitations. The limited number of defect samples, unpredictable defect characteristics, and the interference of metal grain bring great challenges to metal surface defect detection. For this reason, this paper proposes an unsupervised algorithm, FFnet, based on the fusion of frequency domain information, which introduces the frequency domain features into the unsupervised detection. A method of the adaptive enhancement of features in the frequency domain is proposed to make the features on the frequency domain more concerned with anomalies rather than textures. A scale-adaptive feature reconstruction module is used to effectively fuse the spatial and frequency domain features to fully utilize the information from different domains. In addition, a feature selection module is designed to improve the anomaly detection capability and reduce the computational redundancy by selecting the most representative subset of features. The proposed method outperforms other state-of-the-art methods on the connecting rod surface image dataset. In addition, in the generalization experiments of Kolektor Surface-Defect Dataset 2, our method also achieves optimal results and demonstrates strong generalization ability.
format Article
id doaj-art-5b7ec9f935e6447ebb3cdf870a1e47e0
institution DOAJ
issn 1424-8220
language English
publishDate 2025-04-01
publisher MDPI AG
record_format Article
series Sensors
spelling doaj-art-5b7ec9f935e6447ebb3cdf870a1e47e02025-08-20T03:08:54ZengMDPI AGSensors1424-82202025-04-01257225010.3390/s25072250Unsupervised Anomaly Detection on Metal Surfaces Based on Frequency Domain Information FusionWenfei Wu0Tao Tao1Jinsheng Xiao2Yichu Yao3Jianfeng Yang4School of Electronic Information, Wuhan University, Wuhan 430072, ChinaSchool of Electronic Information, Wuhan University, Wuhan 430072, ChinaSchool of Electronic Information, Wuhan University, Wuhan 430072, ChinaSchool of Computer Science, Northwestern Polytechnical University, Xi’an 710129, ChinaSchool of Electronic Information, Wuhan University, Wuhan 430072, ChinaMetal products are widely used in industrial manufacturing, and the quality of metal products is becoming more and more demanding. At present, although there are many methods for detecting defects on metal surfaces, there are still various limitations. The limited number of defect samples, unpredictable defect characteristics, and the interference of metal grain bring great challenges to metal surface defect detection. For this reason, this paper proposes an unsupervised algorithm, FFnet, based on the fusion of frequency domain information, which introduces the frequency domain features into the unsupervised detection. A method of the adaptive enhancement of features in the frequency domain is proposed to make the features on the frequency domain more concerned with anomalies rather than textures. A scale-adaptive feature reconstruction module is used to effectively fuse the spatial and frequency domain features to fully utilize the information from different domains. In addition, a feature selection module is designed to improve the anomaly detection capability and reduce the computational redundancy by selecting the most representative subset of features. The proposed method outperforms other state-of-the-art methods on the connecting rod surface image dataset. In addition, in the generalization experiments of Kolektor Surface-Defect Dataset 2, our method also achieves optimal results and demonstrates strong generalization ability.https://www.mdpi.com/1424-8220/25/7/2250defect detectionunsupervised learningfrequency domainfeature fusion
spellingShingle Wenfei Wu
Tao Tao
Jinsheng Xiao
Yichu Yao
Jianfeng Yang
Unsupervised Anomaly Detection on Metal Surfaces Based on Frequency Domain Information Fusion
Sensors
defect detection
unsupervised learning
frequency domain
feature fusion
title Unsupervised Anomaly Detection on Metal Surfaces Based on Frequency Domain Information Fusion
title_full Unsupervised Anomaly Detection on Metal Surfaces Based on Frequency Domain Information Fusion
title_fullStr Unsupervised Anomaly Detection on Metal Surfaces Based on Frequency Domain Information Fusion
title_full_unstemmed Unsupervised Anomaly Detection on Metal Surfaces Based on Frequency Domain Information Fusion
title_short Unsupervised Anomaly Detection on Metal Surfaces Based on Frequency Domain Information Fusion
title_sort unsupervised anomaly detection on metal surfaces based on frequency domain information fusion
topic defect detection
unsupervised learning
frequency domain
feature fusion
url https://www.mdpi.com/1424-8220/25/7/2250
work_keys_str_mv AT wenfeiwu unsupervisedanomalydetectiononmetalsurfacesbasedonfrequencydomaininformationfusion
AT taotao unsupervisedanomalydetectiononmetalsurfacesbasedonfrequencydomaininformationfusion
AT jinshengxiao unsupervisedanomalydetectiononmetalsurfacesbasedonfrequencydomaininformationfusion
AT yichuyao unsupervisedanomalydetectiononmetalsurfacesbasedonfrequencydomaininformationfusion
AT jianfengyang unsupervisedanomalydetectiononmetalsurfacesbasedonfrequencydomaininformationfusion