A Hybrid Approach to Image Classification Using Kernel-Based Fuzzy Clustering and Support Vector Machine
JOIV : International Journal on Informatics Visualization · 31.03.2026 · 10 (2) · 760
DOI: 10.62527/joiv.10.2.4093 ↗Kurz gefasst
Fachartikel in der Zeitschrift JOIV : International Journal on Informatics Visualization, online erschienen am 31.03.2026. Verfasst von Dinh Toan Pham und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
This study proposes a novel image classification algorithm that integrates kernel-based fuzzy clustering with a support vector machine to improve classification accuracy. The primary contribution of this research is the development of an effective image processing and classification framework that enhances feature representation through advanced segmentation and structured feature extraction. First, a kernel-based fuzzy clustering technique is applied to image segmentation, where fuzzy memberships are constructed in a kernel-induced feature space to highlight differences among groups within each image. This segmentation process also enhances pixel values, enabling more accurate assessment of variations among image elements across different regions. Next, structural texture features are extracted from the segmented images by analyzing four orientation angles and three different distance levels. This process generates discriminative features that characterize image content and provide reliable information for classification. These features are then used as inputs to the support vector machine, which serves as the final classification mechanism and defines decision boundaries between image classes. Experimental results confirm that the proposed kernel-based fuzzy clustering significantly improves the quality of image features. Validation experiments with 48-, 60-, 64-, and 80-feature configurations show that the 60-feature setting achieves the best performance, with an accuracy of 91.24%. On the Lung image dataset, the proposed method achieves the same accuracy, outperforming ResNet50, classical CNNs, DenseNet, and Inception-ResNet-v2. Similar superior results are obtained on a second dataset, further validating the effectiveness of the proposed approach for medical image classification.
Zitieren
APA
Pham, D. T., Le, M. X. T., & Nguyen, H. D. T. (2026). A Hybrid Approach to Image Classification Using Kernel-Based Fuzzy Clustering and Support Vector Machine. JOIV : International Journal on Informatics Visualization, 10(2), 760. https://doi.org/10.62527/joiv.10.2.4093
Vancouver
Pham DT, Le MXT, Nguyen HDT. A Hybrid Approach to Image Classification Using Kernel-Based Fuzzy Clustering and Support Vector Machine. JOIV : International Journal on Informatics Visualization. 2026;10(2):760. doi:10.62527/joiv.10.2.4093
BibTeX
@article{Pham2026,
author = {Pham, Dinh Toan and Le, Mai Xuan Thi and Nguyen, Huong Diem Thi},
title = {A Hybrid Approach to Image Classification Using Kernel-Based Fuzzy Clustering and Support Vector Machine},
journal = {JOIV : International Journal on Informatics Visualization},
year = {2026},
volume = {10},
number = {2},
pages = {760},
doi = {10.62527/joiv.10.2.4093}
}
EUVIMED