Improving Steel Surface Defect Segmentation by Optimizing Gabor Filter Hyperparameters Using a Genetic Algorithm and Random Forest Classification

Document Type : Original Article

Authors

1 Master Student, Technische Faculty, FAU University, Erlangen, Germany

2 AI Engineer, Nikoo Rassam Vira Isatis Company, Iran

3 PhD student, Faculty of Computer Engineering, Yazd University, Yazd, Iran

10.22034/abmir.2025.23933.1182

Abstract

In the steel industry, it is crucial to control surface quality and to promptly detect defects such as cracks, pores, and impurities, as these flaws can compromise the mechanical performance, durability, and reliability of steel. In this study, a novel approach for image segmentation and intelligent detection of steel surface defects is presented using a combination of the Genetic Algorithm and Random Forest. In the proposed method, image features are extracted using Gabor filters, and then the pixels are classified into groups with similar characteristics through the Random Forest algorithm. The hyperparameters of the Gabor filters are considered as the genes of chromosomes in the Genetic Algorithm and are optimized to enhance the segmentation performance. The fitness function, based on the F1-score, measures the classification performance. The results indicate that optimizing the Gabor filter hyperparameters with the Genetic Algorithm enhances both the accuracy and efficiency of steel surface defect segmentation, achieving an F1-score of 0.9807 and an accuracy of 0.9810, and demonstrating that this method is an effective and reliable tool for automated quality control in steel production lines.

Keywords

Main Subjects

References

[1]   R. Neven and T. Goedemé, “A multi-branch U-Net for steel surface defect type and severity segmentation,” Metals, vol. 11, no. 6, 2021.
[2]   D. Sime, G. Wang, Z. Zeng and B. Peng, “Deep learning-based automated steel surface defect segmentation: a comparative experimental study,” Multimedia Tools Application, vol. 83, no. 1, pp. 2995–3018, 2024.
[3]   M. Sharma, L. Jongtae and L. Hansung, “The amalgamation of the object detection and semantic segmentation for steel surface defect detection,” Application Science, vol. 12, no. 12, 2022.
[4]   A. Ibrahim and T. Jules, “A survey of vision-based methods for surface defects’ detection and classification in steel products,” Informatics, vol. 11, no. 2, 2024.
[5]   N. Neogi, D. Mohanta,P. Dutta, “Defect detection of steel surfaces with global adaptive percentile thresholding of gradient image”, Journal of the Institution of Engineers, vol. 98, no. 6, pp. 557-565,2017.
[6]   A. Zeiler, A. Steinboeck, M. Vincze, M. Jochum, ,”Vision-based inspection and segmentation of trimmed steel edges”, IFAC-PapersOnLine, vol. 52, no.14, pp. 165-170, 2019.
[7]   R. Usamentiaga, D. Lema, O. Pedrayes and D. Garcia, “Automated surface defect detection in metals: a comparative review of object detection and semantic segmentation using deep learning,” IEEE Transaction Industial Application, vol. 58, no. 3, pp. 4203–4213, 2022.
 [8]  C. Zhang and X. Zhang, “Multi-target domain-based hierarchical dynamic instance segmentation method for steel defects detection,” Neural Computing Application, vol. 35, no. 10, pp. 7389–7406, 2023.
[9]   Q. Feng, F. Li, H. Li and X.Liu, “Feature reused network: a fast segmentation network model for strip steel surfaces defects based on feature reused,” Visual Computer, vol. 40, no. 5, pp. 3633–3648, 2024.
[10] S. Rawat, D. Banerjee, P.  Aggarwal and M. Singh, “CNN and random forest fusion for enhanced steel defect classification” , International Conference on Advances in Modern Age Technologies for Health and Engineering Science , pp. 1-6, 2024. 
[11] E. Güçlü, İ. Aydın and E. Akın, “Improved deeplabv3+ based steel surface defect segmentation,” in 2024 ASU International Conference in Emerging Technologies for Sustainability and Intelligent Systems ,pp. 609–613, 2024.
[12] H. Wang, Y. Xie, Y. Wang, C. Zhang, P.  Ni, Y.  Lu and Y. Wang, “ Surface defect segmentation of steel pipes based on superpixel prior knowledge”, Engineering Research Express, vol. 7, no. 3, 2025.
[13] Z. Shahidi Zandi and A. Latif, “Developing a modern method in circle detection in digital images by using genetic algorithm,” Machine Vision Image Processing, vol. 8, no. 1, pp. 35–44, 2021.
[14] S. Piryonesi and T. El-Diraby, “Data analytics in asset management: cost-effective prediction of the pavement condition index,” Infrastructre System, vol. 26, no. 1, p. 04019036, 2020.
[15] V. Rodriguez-Galiano, M. Sanchez-Castillo, M. Chica-Olmo and M. Chica-Rivas, “Machine learning predictive models for mineral prospectivity: An evaluation of neural networks, random forest, regression trees and support vector machines,” Ore Geology Reviews, vol. 71, pp. 804–818, 2015.
[16] T. Vijayan, M. Sangeetha and A. Kumaravel, “WITHDRAWN: Gabor filter and machine learning based diabetic retinopathy analysis and detection,” Microprocessors and Microsystems, 2020.
[17] “Kolektor Surface-Defect Dataset (KolektorSDD/KSDD).” Accessed: Nov. 04, 2025. [Online]. Available: https://www.vicos.si/resources/kolektorsdd/
[18] A. Lipowski and D. Lipowska, “Roulette-wheel selection via stochastic acceptance,” Physica A: Statistical Mechanics and its Applications, vol. 391, no. 6, pp. 2193–2196, 2012.
Applied and basic Machine intelligence research
Volume 3, Issue 2
December 2025
Pages 187-197

Files

Share

How to cite

Statistics