Automatic steel surface defect classification is a well-known problem and is being considered for more than two decades. On the other hand, manual methods for defect classification in steel products are time consuming, labor intensive, and are prone to error. As a result, a real-time surface defect classification method using genetic algorithm is proposed to classify four kinds of surface defects in steel surfaces such as Roller Mark, Rust Spot, Emulsion Spot and Scrape. The proposed inspection system consists of visual dataset of 500 steel defect images for processing. The overview of the classification model has online and offline process. In offline, the input image is pre-processed and the following visual features such as geometric features, shape features and texture features are extracted from the defect image. In order to optimize the extracted visual features and to improve the accuracy of the classification model, genetic algorithm is used to find an optimal solution. For optimization, an initial population is generated randomly and the fitness function is calculated for every chromosome. Then the genetic operations such as selection, crossover and mutation are done to generate new offspring chromosome and evaluated to pick optimal chromosome or optimal feature. Finally, in online, the selected features are given as input to the Optimized SVM classifier, in order to predict the class of defect to which the input image belongs. Accuracy, sensitivity, specificity, f-measure and precision are used as the performance metrics to evaluate the system.
Introduction
Image Processing Overview
Image processing transforms images into digital form to enhance them or extract useful information. It involves operations on a 2D array of pixels, each having a value and location. Important types of image features include:
Geometric Features: Shapes defined by points, lines, and curves.
Shape Features: Attributes like rectangularity and density that describe object form.
Texture Features: Describe spatial arrangement of pixel intensities; extracted using techniques like LBP, GLCM, and Tamura.
2. Image Classification
Classifies pixels or regions into categories using image features. There are two main types:
Supervised Classification: Uses labeled training data to guide the classification.
Unsupervised Classification: Groups pixels based on inherent patterns without pre-labeled training data.
3. Genetic Algorithm (GA) Basics
A heuristic optimization technique inspired by natural evolution, useful in feature selection and classifier optimization.
Fitness Function: Measures the quality of a solution.
Encoding: Represents chromosomes using binary strings or other structures.
Operators:
Selection: Chooses parents for reproduction.
Crossover: Combines two parents to produce offspring.
Mutation: Randomly alters genes to introduce diversity.
Termination: Stops when a condition like max generations or optimal fitness is met.
4. Application Need
In steel manufacturing, surface defect detection is crucial for improving product quality and safety. Automating defect classification reduces human involvement in hazardous environments.
5. Literature Review
Various methods and classifiers (SVM, binary classifiers, fuzzy systems) have been explored:
SVM is widely used due to its accuracy and generalization ability.
Genetic Algorithms are applied to enhance classification by optimizing feature selection.
Hybrid approaches (e.g., combining wavelet/Gabor transforms with GA or SVM) have shown promising results for efficient and precise image classification.
6. Proposed Method
The paper proposes an Optimized SVM classification model enhanced by Genetic Algorithm for steel defect classification. It includes:
Architecture:
Offline Module:
Input image is pre-processed (noise removal, illumination correction, binarization).
Feature Extraction:
Geometric: Area, centroid.
Shape: Rectangularity, density.
Texture: LBP, GLCM, Tamura.
Feature Optimization: GA is used to select the best features for classification.
Optimized SVM: Trained with selected features.
Online Module:
Uses the optimized SVM to classify new defect images.
7. Methodology Steps
Image Pre-processing:
Convert to grayscale, remove noise, enhance contrast, binarize image.
Feature Extraction:
Compute geometric (area, centroid), shape (rectangularity, density), and texture features (LBP, GLCM, Tamura).
Feature Selection:
Genetic algorithm optimizes which features are most relevant.
Classification:
Multi-class SVM is trained and tested with the optimized features.
Conclusion
The proposed system is to classify four kinds of surface defects in steel surfaces such as Roller Mark, Rust Spot, Emulsion Spot and Scrape. In offline, the input image is pre-processed and the following visual features such as geometric features, shape features and texture features are extracted from the defect image. In order to optimize the extracted visual features and to improve the accuracy of the classification model, genetic algorithm is used to find an optimal solution.For optimization, an initial population is generated randomly and the fitness function is calculated for every chromosome. Then the genetic operations such as selection, crossover and mutation are done to generate new offspring chromosome and evaluated to pick optimal chromosome or optimal feature. Finally, in online, the selected features are given as input to the Optimized SVM classifier, in order to predict the class of defect to which the input image belongs. Accuracy, sensitivity, specificity, f-measure and precision are used as the performance metrics to evaluate the system. The accuracy of the proposed system is reported by SVM as 78%. Thus, the proposed system provides an automatic surface inspection of steel surfaces and it can be employed in steel firms. As a future work, the research will focus on other shapes for more effective defect prediction and to improve accuracy.
References
[1] Yun J. P, Jeon Y. J Choi, D C Kim. (2012), ‘Real-time defect detection of steel wire rods using wavelet filters optimized by univariatedynamicencoding algorithm for searches’. JOSA A, Vol 29, No 5, pp 797-807.
[2] Alimohamadi H, Ahmadyfard A. (2009), ‘Detecting skin defect of fruits using optimal Gabor wavelet filter’. In Digital Image Processing, 2009 International Conference on IEEE, pp 62-64.
[3] Rahman, M. H, Chowdhury S, Bashar M. A. (2013), ‘An Automatic FaceDetectionandGenderClassificationfromColorImagesusingSupportVectorMachine,JournalofEmergingTrendsinComputingandInformationSciences,Vol4,No1,pp92-101.
[4] Moazenzadeh R, Mohammadi B, Shamshirband S, Kw Chau (2018) Coupling a firefly algorithm with support vector regression to predict evaporation in northern iran. EngApplComput Fluid Mech 12(1):584–597.
[5] Najafi B, FaizollahzadehArdabili S, Shamshirband S, Kw Chau, Rabczuk T (2018) Application of ANNs, ANFIS and RSM to estimating and optimizing the parameters that affect the yield and cost of biodiesel production. EngApplComput Fluid Mech 12(1):611–624.
[6] Houssein, E.H.; Gad, A.G.; Wazery, Y.M. Jaya algorithm and applications: A comprehensive review. MetaheuristicsOptim. Comput. Electr. Eng. 2021, 696, 3–24.
[7] Ogundokun, R.O.; Awotunde, J.B.; Sadiku, P.; Adeniyi, E.A.; Abiodun, M.; Dauda, O.I. An enhanced intrusion detection system using particle swarm optimization feature extraction technique. ProcediaComput. Sci. 2021, 193, 504–512.
[8] Zitar, R.A.; Al-Betar, M.A.; Awadallah, M.A.; Doush, I.A.; Assaleh, K. An intensive and comprehensive overview of JAYA algorithm, its versions and applications. Arch. Comput. Methods Eng. 2022, 29, 763–792.
[9] Mohammadi, M.; Rashid, T.A.; Karim, S.H.T.; Aldalwie, A.H.M.; Tho, Q.T.; Bidaki, M.; Rahmani, A.M.; Hosseinzadeh, M. A comprehensive survey and taxonomy of the SVM-based intrusion detection systems. J. Netw. Comput. Appl. 2021, 178, 102983.
[10] Chen, M. Y, Yang, Y. H, Ho, C. J, Wang, S. H, Liu, S. M, Chang, E, Ouhyoung, M. (2012), ‘Automatic chinese food identification and quantity estimation’, In Siggraph Asia, Technical Briefs pp. 29, ACM.
[11] Kaushik, A, Singh G, Bhatia A. (2013), ‘SVM Classification in Multiclass Letter Recognition System’, Global Journal of Computer Science and Technology, Vol 13, No 9, pp 39-48.
[12] Rahman, M. H, Chowdhury S, Bashar M. A. (2013), ‘An Automatic Face Detection and Gender Classification from Color Images using Support Vector Machine, Journal of Emerging Trends in Computing and Information Sciences, Vol 4, No 1, pp 92-101.
[13] Al-Al-Janabi, M.; Ismail, M.A. Improved intrusion detection algorithm based on TLBO and GA algorithms. Int. Arab J. Inf. Technol. 2021, 18, 170–179.
[14] Aljanabi, M.; Ismail, M.A.; Ali, A.H. Intrusion detection systems, issues, challenges, and needs. Int. J. Comput. Intell. Syst. 2021, 14, 560–571.
[15] Kavak, H.; Padilla, J.J.; Vernon-Bido, D.; Diallo, S.Y.; Gore, R.; Shetty, S. Simulation for cyber security: State of the art and future directions. J. Cyber Secur. 2021, 7, tyab005.
[16] Al-Qatf, M.; Lasheng, Y.; Al-Habib, M.; Al-Sabahi, K. Deep learning approach combining sparse autoencoder with SVM for network intrusion detection. IEEE Access 2018, 6, 52843–52856.
[17] A.R.B. Abad, H. Ghorbani, N. Mohamadian, S. Davoodi, M. Mehrad, S.K.ye Aghdam, H.R. Nasriani, Robust hybrid machine learning algorithms for gas flow rates prediction through wellhead chokes in gas condensate fields, Fuel 308 (2022).
[18] Al-Mahasneh, A. J., Anavatti, S. G., & Garratt, M. A. (2018a). Review of Applications of Generalized Regression Neural Networks in Identification and Control of Dynamic Systems.
[19] Azam, A., Bardhan, A., Kaloop, M. R., Samui, P., Alanazi, F., Alzara, M., &Yosri, A. M. (2022). Modeling resilient modulus of subgrade soils using LSSVM optimized with swarm intelligence algorithms. Scientific Reports 2022 12:1, 12(1), 1–20.
[20] A.R. Behesht Abad, S. Mousavi, N. Mohamadian, D.A. Wood, H. Ghorbani, S. Davoodi, …., K. Shahbazi, Hybrid machine learning algorithms to predict condensate viscosity in the near wellbore regions of gas condensate reservoirs, Journal of Natural Gas Science and Engineering 95 (2021).
[21] Q. Fan, T. Fan, A Hybrid Model of Extreme Learning Machine Based on Bat and Cuckoo Search Algorithm for Regression and Multiclass Classification, Journal of Mathematics 2021 (2021).
[22] J.C. Huang, K.M. Ko, M.H. Shu, B.M. Hsu, Application and comparison of several machine learning algorithms and their integration models in regression problems, Neural Computing and Applications 32 (10) (2020) 5461–5469.
[23] I. Izonin, R. Tkachenko, M. Gregus ml, K. Zub, P. Tkachenko, A GRNN-based approach towards prediction from small datasets in medical application, Procedia Computer Science 184 (2021) 242–249.
[24] N. Kardani, A. Bardhan, B. Roy, P. Samui, M. Nazem, D.J. Armaghani, A. Zhou, A novel improved Harris Hawks optimization algorithm coupled with ELM for predicting permeability of tight carbonates, Engineering with Computers 1 (2021) 1–24.
[25] H. Khodakhah, P. Aghelpour, Z. Hamedi, Comparing linear and non-linear data-driven approaches in monthly river flow prediction, based on the models SARIMA, LSSVM, ANFIS, and GMDH, Environmental Science and Pollution Research 29 (15) (2022) 21935–21954.
[26] Kui, L., Ke, Y., Xiaojun, X., Jianjun, Z., Yun, L., &Jinghong, W. (2022). Information Price Intelligent Prediction Model of Major Equipment in Power Grid Projects Based on the SSA-LSSVM Model. 2022 IEEE Asia-Pacific Conference on Image Processing, Electronics and Computers, IPEC 2022, 549–553.
[27] W. Lu, R. Du, P. Niu, G. Xing, H. Luo, Y. Deng, L. Shu, Soybean Yield Preharvest Prediction Based on Bean Pods and Leaves Image Recognition Using Deep Learning Neural Network Combined With GRNN, Frontiers in Plant Science 12 (2022) 3160.
[28] L. Mancipe-Castro, R.E. Gutiérrez-Carvajal, Prediction of environment variables in precision agriculture using a sparse model as data fusion strategy, Information Processing in Agriculture 9 (2) (2022) 171–183.
[29] C.S.W. Ng, A. JahanbaniGhahfarokhi, M. Nait Amar, Application of nature-inspired algorithms and artificial neural network in waterflooding well control optimization, Journal of Petroleum Exploration and Production 11 (7) (2021) 3103–3127.