Integrated hyperspectral imaging for enhanced landmine detection

Landmines remain major threats in post-conflict environments, threatening the lives of civilians and hindering recovery processes. This research examines the application of hyperspectral imaging (HSI) in the visible near-infrared (VNIR), shortwave infrared (SWIR), and longwave infrared (LWIR) spectral bands to improve landmine detection based on their distinct spectral and thermal properties. A hybrid detection framework is presented that combines pixel intensity analysis with anomaly detection methods to enhance accuracy while keeping computational efficiency. Controlled experiments were performed in laboratory and outdoor environments, including scenarios involving surface-laid, semi-buried, and buried landmines. Calibration using standard reference materials was used to ensure spectral fidelity in all imaging bands. The suggested methodology attained an average detection accuracy of 92.4%, and the precision and recall rates remained consistently high across terrain types, burial depths, and thermal conditions, as confirmed through stratified performance analysis. These results demonstrate the potential of HSI combined with machine learning algorithms, serving as a foundational baseline towards a scalable platform for landmine detection. However, transitioning this multi-modal system from highly characterized, semi-controlled testing environments to completely unmanaged operational environments remains an open challenge.