Manuscript received February 25, 2025; revised March 6, 2025; published July 22, 2025
Abstract—With the rapid development of remote sensing technology, high-resolution satellite images play an essential role in fields such as environmental monitoring, urban planning, agricultural assessment, and disaster management. As one of the core tasks of satellite data processing, the accuracy and efficiency of satellite image classification directly affect the reliability of subsequent applications. This paper reviews the main methods of satellite image classification, including traditional machine learning methods (e.g., Support Vector Machine, Random Forest) and deep learning methods (e.g., Convolutional Neural Networks, Transformer). Firstly, this paper analyzes which classification tasks are the main focus of current research and then examines the advantages and disadvantages of different methods. In addition, this paper explores the application of techniques such as multi-source data fusion, few-shot learning and semantic segmentation in improving classification performance. The experimental results demonstrate that deep learning-based classification methods perform well in complex scenarios but still face challenges such as high sample labelling costs and insufficient model generalization capability. Finally, this paper suggests that future research could combine self-supervised learning, lightweight networks, and 3D satellite information mining to further enhance classification accuracy and practical applicability.
Keywords—classification method, classification types, classification role
Cite: Zhiyan Liu and Haining Zhang, "A Systematic Review of Satellite Image Classification," International Journal of Machine Learning vol. 15, no. 3, pp. 51-63, 2025.
Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).