Volume 22, Issue 1 (5-2025)
JSDP 2025, 22(1): 71-82 |
Back to browse issues page
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Firuz Mahjanabadi Z, Jafari P, Rezaei M. Automated Classification of Steel Phases in Scanning Electron Microscope Images. JSDP 2025; 22 (1) :71-82
URL: http://jsdp.rcisp.ac.ir/article-1-1400-en.html
URL: http://jsdp.rcisp.ac.ir/article-1-1400-en.html
Abstract: (78 Views)
The properties of steels are determined by their microstructural components known as phases. These phases can be observed in images obtained from the steel surface using a scanning electron microscope. The identification and classification of these phases can contribute to a better understanding of steel properties. This article presents an automated classification algorithm for identifying different steel phase ferrite shapes in scanning electron microscope (SEM) images. The algorithm utilizes texture feature extraction methods and machine learning models for classification. The dataset used in this study consists of SEM images with dimensions of 768x1024. The images are divided into 128x128 blocks, and classification is performed on each block individually. The number of inputs is increased using techniques such as rotation and scaling. Various image processing methods are employed for feature extraction to obtain texture features. The algorithm successfully recognizes different classes of ferrite phases using two classification models: K-means algorithm and a softmax neural network. The proposed method combines different texture features of an image with a machine learning model and is based on 1522 image blocks for acicular and granular ferrite, as well as 1558 SEM image blocks for acicular, granular, and Widmanstatten ferrites. Additionally, the PCA method is used to eliminate redundant and non-useful features. Despite the complex shapes of the ferrites and the absence of specific patterns in the SEM images, the proposed algorithm achieves high accuracy in automatically identifying different ferrite classes. It achieves an accuracy of 99% for two classes and 86 percent for three classes.
Keywords: Scanning Electron Microscope, Automated Classification, Steel Phases, Neural Networks, K-Means Algorithm
Type of Study: Research |
Subject:
Paper
Received: 2023/09/26 | Accepted: 2025/03/8 | Published: 2025/06/21 | ePublished: 2025/06/21
Received: 2023/09/26 | Accepted: 2025/03/8 | Published: 2025/06/21 | ePublished: 2025/06/21
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
