Deep Neural Networks for Hyperspectral Image Classification

Dear Colleagues,

Hyperspectral imaging captures detailed spectral signatures across hundreds of contiguous bands, providing rich information for precise material discrimination and fine-grained land-cover classification. This unique capability has made hyperspectral data a critical resource in numerous remote sensing applications, including agriculture, environmental monitoring, urban analysis, and resource management. However, the high dimensionality, spectral redundancy, strong spatial–spectral coupling, and scarcity of labeled samples present substantial challenges to traditional feature engineering and shallow learning approaches.

In recent years, deep learning—particularly deep neural networks (DNNs)—has emerged as a powerful paradigm for hyperspectral image classification. By automatically learning hierarchical and discriminative representations from raw data, DNN-based methods have demonstrated significant improvements in classification accuracy, robustness, and generalization. Advances in convolutional neural networks, attention mechanisms, transformer models, and graph-based learning have further expanded the modeling capacity for complex hyperspectral scenes. Despite these achievements, critical issues such as computational efficiency, model interpretability, scalability, and adaptability to real-world scenarios remain open. This Special Issue aims to address these challenges by showcasing cutting-edge DNN methodologies that push the boundaries of hyperspectral image classification research.

The aim of this Special Issue is to collect high-quality original research and comprehensive reviews that reflect recent methodological advances and emerging applications of deep neural networks for hyperspectral image classification. We welcome contributions that propose innovative network architectures, learning paradigms, and data processing strategies to effectively address core challenges such as limited labeled data, complex spatial–spectral relationships, multimodal data fusion, and model interpretability.

By providing a focused and interdisciplinary platform, this Special Issue seeks to promote the exchange of ideas between the remote sensing, deep learning, and computer vision communities. It aims to highlight emerging trends, benchmark state-of-the-art methods, and encourage the development of robust and efficient deep learning solutions that can be deployed in real-world remote sensing systems, advancing intelligent remote sensing interpretation. The topic of this Special Issue is fully aligned with the scope of Remote Sensing, emphasizing advanced sensing data analysis, intelligent information extraction, and the practical application of novel computational techniques in Earth observation.

This Special Issue welcomes high-quality submissions focusing on methodological research and practical applications of deep neural networks for hyperspectral image classification. Topics of interest include, but are not limited to, the following:

• Design and modeling of novel deep neural network architectures tailored to the characteristics of hyperspectral imagery;
• Hyperspectral image classification algorithms under complex data conditions, such as limited training samples and class imbalance;
• Classification methods based on the fusion of hyperspectral imagery with multi-source remote sensing data, including LiDAR and SAR;
• Performance optimization strategies for hyperspectral image classification based on attention mechanisms, transfer learning, and federated learning;
• Application-oriented studies of deep neural networks for hyperspectral classification in agriculture monitoring, ecological assessment, urban analysis, and other tasks related to intelligent remote sensing interpretation.

The Special Issue welcomes original research articles, comprehensive review papers, and short communications.

Prof. Dr. Zhen Ye
Prof. Dr. Xiangtao Zheng
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 250 words) can be sent to the Editorial Office for assessment.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-anonymized peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Remote Sensing is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2700 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• hyperspectral image classification
• deep neural networks
• hyperspectral image processing
• spectral–spatial feature learning
• band selection
• intelligent remote sensing interpretation
• attention mechanisms