Surface Defect Detection via Entity Sparsity Pursuit With Intrinsic Priors
The DOI matches its Crossref record; 43 of the 43 references it cites were found in a registry or an index.
Abstract
Computer vision based methods have been widely used in surface defect inspection. However, most of these approaches are task specific, and it is hard to transfer them to similar detection scenarios. This paper proposes an entity sparsity pursuit (ESP) method to identify surface defects. Based on the observation that surface image textures usually form a low-rank structure and the structure can be violated by the presence of rare defects, we formulate the detection task as a low-rank and ESP problem. To alleviate the feature shortage issue existed in industrial gray-scale images, we customize a kind of intuitive features for surface defect inspection. Different from previous work utilizing complicated regularization terms, we resort to mine intrinsic priors of defect images, which can be neatly incorporated into the designed architecture. The proposed model is compact and able to detect surface defects in an unsupervised manner. To fully evaluate the presented method, we conduct a series of experiments using three real-world and one synthetic defect datasets. Experimental results demonstrate that ESP outperforms state-of-the-art methods.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Wang et al. [26] constructed a compact model to be regarded as a kind of guidance information by mining intrinsic image priors, it offers a good generalization ability for different detection tasks and is sufficiently robust to noise.”
From Automated Visual Defect Detection for Flat Steel Surface: A Survey (2020) · cites this paper as [26]
In the citing paper
“Essentially, feature extraction and feature selection are dedicated to learning the temporal, spatial [13-15] and spectral features [16, 17] in images and even the intrinsic priors in the production line [18] to narrow the intraclass variation and expand the interclass distance.”
From Automated Visual Defect Classification for Flat Steel Surface: A Survey (2020) · cites this paper as [18]
In the citing paper
“Be of rotation and gray invariance, fast calculation speed.”
From Automated Visual Defect Classification for Flat Steel Surface: A Survey (2020) · cites this paper as [18]
In the citing paper
“Sensitive to scale variation and noise interference.”
From Automated Visual Defect Classification for Flat Steel Surface: A Survey (2020) · cites this paper as [18]
Citations
Cites
A Threshold Selection Method from Gray-Level Histograms
Otsu, N. (1979) · IEEE Transactions on Systems, Man, and Cybernetics · doi:10.1109/tsmc.1979.4310076 · Cited by 43,398 · cited 1×
“With the refined defect map, OTSU algorithm [33] is then adopted to perform defect segmentation.”
MethodSLIC Superpixels Compared to State-of-the-Art Superpixel Methods
Achanta, R., Shaji, A., Smith, K., Lucchi, A., Fua, P., & Süsstrunk, S. (2012) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2012.120 · Cited by 9,201
Robust principal component analysis?
Candès, E. J., Li, X., Ma, Y., & Wright, J. (2011) · Journal of the ACM · doi:10.1145/1970392.1970395 · Cited by 6,921 · cited 2×
“Therefore, robust PCA (RPCA) [17] was proposed to tackle this problem.”
BackgroundMethodA comparative study of texture measures with classification based on featured distributions
Ojala, T., Pietikäinen, M., & Harwood, D. (1996) · Pattern Recognition · doi:10.1016/0031-3203(95)00067-4 · Cited by 6,882 · cited 1×
“To make it clear, we briefly review the concept of LBP [21].”
BackgroundA Singular Value Thresholding Algorithm for Matrix Completion
Cai, J.-F., Candès, E. J., & Shen, Z. (2010) · SIAM Journal on Optimization · doi:10.1137/080738970 · Cited by 5,809 · cited 1×
“Updating J: According to [29], the subproblem has the closed solution by matrix shrinkage operator.”
BackgroundHalko, N., Martinsson, P. G., & Tropp, J. A. (2011) · SIAM Review · doi:10.1137/090771806 · Cited by 4,002 · cited 1×
“In addition, the running speed can be further accelerated by introducing fast randomized SVD [39] or developing efficient implementations using high performance computing architectures [40], which also facilitate the proposed method for industrial applications.”
Robust Recovery of Subspace Structures by Low-Rank Representation
Liu, G., Lin, Z., Yan, S., Sun, J., Yu, Y., & Ma, Y. (2013) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2012.88 · Cited by 3,594 · cited 3×
“According to the analysis in [18], the overall computational cost is at most O ( d 2 N + d 3 ) provided that d ≤ N , which obviously holds in our experiment ( d = 15).”
BackgroundMethodInfluentialEfficient and Robust Feature Selection via Joint ℓ2,1-Norms Minimization
Nie, F., Huang, H., Cai, X., & Ding, C. H. Q. (2010) · neural information processing systems · Cited by 2,231
A Statistical Approach to Texture Classification from Single Images
Varma, M., & Zisserman, A. (2005) · International Journal of Computer Vision · doi:10.1007/s11263-005-4635-4 · Cited by 1,142 · cited 1×
“In addition, defects in surface images are with diverse shapes and patterns, and thus an eight-dimensional (8-D) texton feature [22] is employed for its advantage in texture classification [23].”
BackgroundShotton, J., Winn, J., Rother, C., & Criminisi, A. (2009) · International Journal of Computer Vision · doi:10.1007/s11263-007-0109-1 · Cited by 1,026 · cited 1×
“D) texton feature [22] is employed for its advantage in texture classification [23].”
Song, K., & Yan, Y. (2013) · Applied Surface Science · doi:10.1016/j.apsusc.2013.09.002 · Cited by 1,026 · cited 1×
“3) NEU Surface Defect Database [35]: It consists of 300 defective images per class of hot-rolled steel strip with bounding box annotations.”
BackgroundLinearized Alternating Direction Method with Adaptive Penalty for Low-Rank Representation
Lin, Z., Liu, R., & Su, Z. (2011) · arXiv (Cornell University) · doi:10.48550/arxiv.1109.0367 · Cited by 992 · cited 1×
“2) Optimization: To solve problem (13), we first introduce an auxiliary variable J and convert it to the following problem: Then, the alternating direction method (ADM) [28] is adopted to solve problem (16).”
MethodChen, F.-C., & Jahanshahi, M. R. (2018) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2017.2764844 · Cited by 930 · cited 1×
“As a result, several studies have attempted to use deep learning (DL) methods for defect inspection [14], [15], but these methods still face great challenges.”
BackgroundA unified approach to salient object detection via low rank matrix recovery
Shen, X., & Wu, Y. (2012) · 2012 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2012.6247758 · Cited by 691 · cited 1×
“In order to detect salient object more accurately, Shen and Wu [20] proposed a classical feature combination, in which features including color, steerable pyramids, and Gabor filters were extracted to form a 53-dimensional (53-D) feature vector.”
MethodAnomaly Detection in Hyperspectral Images Based on Low-Rank and Sparse Representation
Xu, Y., Wu, Z., Li, J., Plaza, A., & Wei, Z. (2016) · IEEE Transactions on Geoscience and Remote Sensing · doi:10.1109/tgrs.2015.2493201 · Cited by 602
A Fast Algorithm for Edge-Preserving Variational Multichannel Image Restoration
Yang, J., Yin, W., Zhang, Y., & Wang, Y. (2009) · SIAM Journal on Imaging Sciences · doi:10.1137/080730421 · Cited by 529 · cited 1×
“Based on the proofs in [31], we prove that the i th column of E can be deduced as Thus, we can update E in a columnwise manner.”
BackgroundA Generic Deep-Learning-Based Approach for Automated Surface Inspection
Ren, R., Hung, T., & Tan, K. C. (2018) · IEEE Transactions on Cybernetics · doi:10.1109/tcyb.2017.2668395 · Cited by 485 · cited 1×
“As mentioned in [16], training a DL model usually requires considerable samples while collecting and labeling a large number of defect images could be costly.”
BackgroundChen, J., Liu, Z., Wang, H., Nunez, A., & Han, Z. (2018) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2017.2775345 · Cited by 456 · cited 1×
“As a result, several studies have attempted to use deep learning (DL) methods for defect inspection [14], [15], but these methods still face great challenges.”
BackgroundSurface defect saliency of magnetic tile
Huang, Y., Qiu, C., & Yuan, K. (2020) · The Visual Computer · doi:10.1007/s00371-018-1588-5 · Cited by 438 · cited 1×
“1) Magnetic Tile (MT) Blowhole Dataset [34]: The MT blow-hole dataset totally provides 115 surface defect images with pixelwise ground truth (GT).”
BackgroundSalient Object Detection via Structured Matrix Decomposition
Peng, H., Li, B., Ling, H., Hu, W., Xiong, W., & Maybank, S. J. (2017) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2016.2562626 · Cited by 378 · cited 1×
“Different from conventional low-rank-based models [26], [27] using complicated regularization terms, we present a compact architecture that can smoothly incorporate priors of defect images.”
MethodA Hierarchical Extractor-Based Visual Rail Surface Inspection System
Gan, J., Li, Q., Wang, J., & Yu, H. (2017) · IEEE Sensors Journal · doi:10.1109/jsen.2017.2761858 · Cited by 236 · cited 3×
“enable automatic and nondestructive detection, visual inspection systems (VISs) have found wide applications in surface inspection of steel [1], tire [2], rail [3], solar modules [4], printed circuit boards [5], and other planar products [6].”
BackgroundMethodSelf-citationSaliency-Based Defect Detection in Industrial Images by Using Phase Spectrum
Bai, X., Fang, Y., Lin, W., Wang, L., & Ju, B.-F. (2014) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2014.2359416 · Cited by 225 · cited 1×
“The experiments compare ESP with three state-of-the-art methods, including CAT [36], SBD [37], and SSD [38].”
MethodDefect Detection in Solar Modules Using ICA Basis Images
Tsai, D.-M., Wu, S.-C., & Chiu, W.-Y. (2013) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2012.2209663 · Cited by 187 · cited 1×
“enable automatic and nondestructive detection, visual inspection systems (VISs) have found wide applications in surface inspection of steel [1], tire [2], rail [3], solar modules [4], printed circuit boards [5], and other planar products [6].”
BackgroundHalfawy, M. R., & Hengmeechai, J. (2013) · Automation in Construction · doi:10.1016/j.autcon.2013.10.012 · Cited by 161
Steel Surface Defect Detection Using a New Haar–Weibull-Variance Model in Unsupervised Manner
Liu, K., Wang, H., Chen, H., Qu, E., Tian, Y., & Sun, H. (2017) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2017.2712838 · Cited by 161
TEXEMS: Texture Exemplars for Defect Detection on Random Textured Surfaces
Xie, X., & Mirmehdi, M. (2007) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2007.1038 · Cited by 159 · cited 1×
“To this end, Xie and Mirmehdi [11] presented texture exemplars for detecting and localizing defects in random color textures.”
BackgroundA Contrast Adjustment Thresholding Method for Surface Defect Detection Based on Mesoscopy
Win, M., Bushroa, A. R., Hassan, M. A., Hilman, N. M., & Ide-Ektessabi, A. (2015) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2015.2417676 · Cited by 142
Defect detection in textured materials using optimized filters
Kumar, A., & Pang, G. K. H. (2002) · IEEE Transactions on Systems, Man and Cybernetics, Part B (Cybernetics) · doi:10.1109/tsmcb.2002.1033176 · Cited by 124 · cited 1×
“Obviously, such methods are appropriate for surfaces appearing regular patterns and periodic textures such as fabrics [7], but the selection of filters can affect their performance to some extent.”
MethodSurface Defect Saliency of Magnetic Tile
Huang, Y., Qiu, C., Guo, Y., Wang, X., & Yuan, K. (2018) · 2018 IEEE 14th International Conference on Automation Science and Engineering (CASE) · doi:10.1109/coase.2018.8560423 · Cited by 117
Anomaly Detection in Images With Smooth Background via Smooth-Sparse Decomposition
Yan, H., Paynabar, K., & Shi, J. (2017) · Technometrics · doi:10.1080/00401706.2015.1102764 · Cited by 111 · cited 2×
“The experiments compare ESP with three state-of-the-art methods, including CAT [36], SBD [37], and SSD [38].”
MethodConstructing a Nonnegative Low-Rank and Sparse Graph With Data-Adaptive Features
Zhuang, L., Gao, S., Tang, J., Wang, J., Lin, Z., Ma, Y., & Yu, N. (2015) · IEEE Transactions on Image Processing · doi:10.1109/tip.2015.2441632 · Cited by 110 · cited 1×
“It explicitly assumes that the corruptions in data are “sample specific,” i.e., some data vectors are corrupted and the others are clean [19].”
BackgroundSelf-citationSegmentation Driven Low-rank Matrix Recovery for Saliency Detection
Zou, W., Kpalma, K., Liu, Z., & Ronsin, J. (2013) · British Machine Vision Conference (BMVC) · doi:10.5244/c.27.78 · Cited by 103 · cited 1×
“To enhance consistency of defect maps of such images, one can deploy the strategy proposed in [32] to modify the defect level of each superpixel by its adjacent superpixels, i.e., where α is a weight to control the influence of neighbors on the current superpixel, which is set to 0.5 in our…”
MethodInfluentialDetection of surface crack defects on ferrite magnetic tile
Li, X., Jiang, H., & Yin, G. (2013) · NDT & E International · doi:10.1016/j.ndteint.2013.10.006 · Cited by 89 · cited 1×
“Statistical approaches usually employ co-occurrence matrix [8], histogram [9], and other texture statistics to detect surface defects.”
MethodFabric defect inspection using prior knowledge guided least squares regression
Cao, J., Zhang, J., Wen, Z., Wang, N., & Liu, X. (2015) · Multimedia Tools and Applications · doi:10.1007/s11042-015-3041-3 · Cited by 85 · cited 4×
“Typically, model-based approaches tend to require a high computational complexity [13].”
BackgroundMethodResultInfluentialAutomatic Detection of Defects in Tire Radiographic Images
Zhang, Y., Lefebvre, D., & Li, Q. (2015) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2015.2469594 · Cited by 70 · cited 1×
“enable automatic and nondestructive detection, visual inspection systems (VISs) have found wide applications in surface inspection of steel [1], tire [2], rail [3], solar modules [4], printed circuit boards [5], and other planar products [6].”
BackgroundSelf-citationAccurate and Efficient Inspection of Speckle and Scratch Defects on Surfaces of Planar Products
Kong, H., Yang, J., & Chen, Z. (2017) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2017.2668438 · Cited by 64 · cited 1×
“enable automatic and nondestructive detection, visual inspection systems (VISs) have found wide applications in surface inspection of steel [1], tire [2], rail [3], solar modules [4], printed circuit boards [5], and other planar products [6].”
BackgroundAutomatic Visual Defect Detection Using Texture Prior and Low-Rank Representation
Huangpeng, Q., Zhang, H., Zeng, X., & Huang, W. (2018) · IEEE Access · doi:10.1109/access.2018.2852663 · Cited by 53 · cited 4×
“Specifically, images in both [13] and [25] are divided into uniform-sized patches rather than varisized superpixels.”
BackgroundMethodResultInfluentialA Monitoring and Diagnostic Approach for Stochastic Textured Surfaces
Bui, A. T., & Apley, D. W. (2017) · Technometrics · doi:10.1080/00401706.2017.1302362 · Cited by 45 · cited 1×
“Based on the Markov random field model, a supervised method [12] was developed to monitor and diagnose stochastic textured surfaces.”
MethodDouble Low-Rank and Sparse Decomposition for Surface Defect Segmentation of Steel Sheet
Zhou, S., Wu, S., Liu, H., Lu, Y., & Hu, N. (2018) · Applied Sciences · doi:10.3390/app8091628 · Cited by 28 · cited 2×
“For the convenience of evaluation, we select the surface defect images of type patch, whose pixelwise GT is provided in [27].”
MethodCorner Point-Based Coarse–Fine Method for Surface-Mount Component Positioning
Bai, L., Yang, X., & Gao, H. (2017) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2017.2705078 · Cited by 27 · cited 1×
“enable automatic and nondestructive detection, visual inspection systems (VISs) have found wide applications in surface inspection of steel [1], tire [2], rail [3], solar modules [4], printed circuit boards [5], and other planar products [6].”
BackgroundSelf-citationA Monitoring and Diagnostic Approach for Stochastic Textured Surfaces
Bui, A. T., & Apley, D. W. (2017) · Figshare · doi:10.6084/m9.figshare.4736344.v2 · Cited by 17
A Monitoring and Diagnostic Approach for Stochastic Textured Surfaces
Bui, A. T., & Apley, D. W. (2017) · Figshare · doi:10.6084/m9.figshare.4736344 · Cited by 17
Detection of defects in fabrics using subimage-based singular value decomposition
Chandra, J. K., & Datta, A. K. (2012) · Journal of the Textile Institute · doi:10.1080/00405000.2012.721206 · Cited by 11 · cited 1×
“Unfortunately, both spectral and statistical methods can hardly deal with the situation where stochastic variations appeared in randomly textured surfaces [10].”
Background
Cited by
Dong, H., Song, K., He, Y., Xu, J., Yan, Y., & Meng, Q. (2020) · IEEE Transactions on Industrial Informatics · doi:10.1109/tii.2019.2958826 · Cited by 559
“Other approaches based on hand crafted feature are used for defect detection in industrial applications (such as solar modules [3], metal [4], and steel [5]) and have achieved good result in recent years.”
BackgroundAutomated Visual Defect Detection for Flat Steel Surface: A Survey
Luo, Q., Fang, X., Liu, L., Yang, C., & Sun, Y. (2020) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2019.2963555 · Cited by 559
“[26] proposed a LBP-inspired feature extractor by estimating the variations of four directions”
BackgroundMethodInfluentialVisual-Based Defect Detection and Classification Approaches for Industrial Applications—A SURVEY
Czimmermann, T., Ciuti, G., Milazzo, M., Chiurazzi, M., Roccella, S., Oddo, C. M., & Dario, P. (2020) · Sensors · doi:10.3390/s20051459 · Cited by 441
Steel Surface Defect Recognition: A Survey
Wen, X., Shan, J., He, Y., & Song, K. (2022) · Coatings · doi:10.3390/coatings13010017 · Cited by 210
MSC-DNet: An efficient detector with multi-scale context for defect detection on strip steel surface
Liu, R., Huang, M., Gao, Z., Cao, Z., & Cao, P. (2023) · Measurement · doi:10.1016/j.measurement.2023.112467 · Cited by 159
Metadata
- Authors
- Wang, J., Li, Q., Gan, J., Yu, H., & Yang, X.
- Affiliations
- Beijing Jiaotong University (CN)
- Global Energy Interconnection Research Institute (China) (CN)
- Beijing Wuzi University (CN)
- Venue
- IEEE Transactions on Industrial Informatics
- Published
- 2020-01
- Volume
- 16
- Issue
- 1
- Pages
- 141-150
- ISSN
- 1551-3203 (ISSN-L)
- 1941-0050
- OpenAlex
- W2945270739
- Semantic Scholar
- c944f84bbfaf1e63617b4fabdfd41322f497cfc7
- Cited by
- 141
- References
- 42
- Citations by year
- 2026: 13 · 2025: 21 · 2024: 26 · 2023: 22 · 2022: 31 · 2021: 14 · 2020: 13
- FWCI
- 10.89
- Type
- journal article
- Language
- en
- Open access
- closed
- Licence
- Locations
- IEEE Transactions on Industrial Informatics (publishedVersion)
- Funders
- Fundamental Research Funds for the Central Universities (2018YJS042, 2016JBZ006)
- Open Fund of Beijing Intelligent Logistics System Collaborative Innovation Center (2018KF-07)
- Shanghai Research Program (17511102900)
- Topics
- Physical Sciences › Engineering › Industrial and Manufacturing Engineering › Industrial Vision Systems and Defect Detection
- Physical Sciences › Engineering › Civil and Structural Engineering › Infrastructure Maintenance and Monitoring
- Physical Sciences › Engineering › Mechanical Engineering › Non-Destructive Testing Techniques
- Keywords
- surface defect detection, low-rank structure, intrinsic priors, unsupervised defect detection, grayscale images, texture structure, industrial inspection
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.