APPLICATION OF HSV COLOR ANALYSIS AND GREEN MASKING FOR UXO DETECTION IN DENSE VEGETATIONS

Abstract

This study examined the applicability of HSV color-space analysis and green-mask segmentation for the preliminary detection of unexploded ordnance (UXO) in densely vegetated terrain using UAV-acquired imagery. The research addressed the problem of limited visibility in natural environments where vegetation frequently obscures surface objects and reduces the effectiveness of conventional optical detection. High-resolution RGB images collected by a low-altitude UAV flight were converted into the HSV domain to isolate vegetation based on hue, saturation, and value distributions. Segmentation thresholds were optimized through histogram-based calibration and iterative refinement. The analysis revealed that vegetation covered approximately 70–74% of the test area, resulting in complex contour patterns and underscoring the need for color-space preprocessing in cluttered environments. Importantly, UXO models were consistently excluded from the green mask, indicating that the segmentation method successfully differentiated non-vegetative objects from natural background elements. These findings demonstrated that HSV-based vegetation extraction enhanced object visibility and provided a reliable foundation for subsequent UXO recognition. The study concluded that integrating HSV segmentation with UAV imagery improves early-stage UXO detection and should be further expanded through multimodal sensing and machine-learning approaches.