Andalib et al. (2026) An intelligent dual-stage fusion framework of optical and radar data for land cover classification

Identification

• Journal: Remote Sensing Applications Society and Environment
• Year: 2026
• Date: 2026-03-28
• Authors: Hooriyeh Andalib, Hamid Ebadi, Rana Naanjam
• DOI: 10.1016/j.rsase.2026.101987

Research Groups

• Faculty of Geodesy and Geomatics Engineering, K. N. Toosi University of Technology, Tehran, Iran

Short Summary

This study introduces a novel dual-stage fusion framework that integrates optical and radar remote sensing data at both feature and knowledge levels to improve land cover classification accuracy. The proposed method achieves an overall accuracy of 94.7% and a Kappa coefficient of 0.93 in urban environments.

Objective

• To develop and evaluate a novel dual-level fusion approach that combines optical and radar imagery at both the feature and knowledge levels for robust and accurate land cover classification, reducing reliance on manual labeling.

Study Configuration

• Spatial Scale: Urban areas (Tehran and Tabriz, Iran)
• Temporal Scale: Not explicitly stated for data acquisition; study published in 2026.

Methodology and Data

• Models used: Dual-level fusion framework (feature-level fusion, feature–knowledge fusion), machine learning-based classification model with automatic training sample selection.
• Data sources: Optical satellite imagery, Synthetic Aperture Radar (SAR) satellite imagery.

Main Results

• The proposed dual-level fusion approach achieved an overall classification accuracy of 94.7%.
• The Kappa coefficient for the classification was 0.93.
• The method demonstrated enhanced classification performance and improved robustness by reducing reliance on manual labeling.

Contributions

• Introduction of a novel dual-level fusion framework that integrates optical and radar data at both the feature and knowledge levels.
• Development of a method that incorporates feature–knowledge fusion during training-sample selection, thereby reducing the need for manual labeling and enhancing classification robustness.
• Provision of a practical and reliable solution for land cover mapping in complex urban environments with significantly improved accuracy compared to using single data sources.

Funding

• Not specified in the provided text.

Citation

@article{Andalib2026intelligent,
  author = {Andalib, Hooriyeh and Ebadi, Hamid and Naanjam, Rana},
  title = {An intelligent dual-stage fusion framework of optical and radar data for land cover classification},
  journal = {Remote Sensing Applications Society and Environment},
  year = {2026},
  doi = {10.1016/j.rsase.2026.101987},
  url = {https://doi.org/10.1016/j.rsase.2026.101987}
}