Global Saturation-Value Translation Approach for Haze Removal in Urban Aerial Imagery

This paper proposes Global Saturation-Value Translation (GSVT), a novel single-image dehazing technique that extends the principles of the Regional Saturation-Value Translation (RSVT) algorithm to achieve comprehensive haze removal across an entire image. While RSVT effectively mitigates haze in specific bright regions by leveraging localized saturation and value shifts in the HSV color space, its regional focus limits its applicability to heterogeneous scenes with varying haze densities. The proposed GSVT framework addresses this limitation by proposing a globally adaptive translation mechanism that dynamically adjusts saturation and value components across all image regions, guided by a spatially-aware translation map. By preserving hue consistency and employing a global coordinate shift in the saturation-value plane, GSVT ensures uniform dehazing without the need for soft segmentation, reducing computational complexity while enhancing color fidelity and detail recovery. Experimental results on synthetic and real-world data demonstrate that GSVT performs competitively when compared with state-of-the-art methods in terms of quantitative metrics and visual quality.