BDSA: Remote Sensing Image Dehazing via Fusing Brightness Decomposition and Saturation Adjustment

Remote sensing imagery is often degraded by atmospheric scattering and suspended particles, which suppress brightness, distort colors, and obscure fine details. To mitigate these degradations, we introduce a dehazing approach termed BDSA that integrates brightness decomposition with saturation adjustment to restore visual clarity and fidelity. Specifically, our first performs a nonlinear transformation on the luminance channel of the YCbCr color space, wherein a logarithmic modulation scheme is applied to strengthen the visibility of dark regions while an exponential modulation strategy is simultaneously adopted to retain the fidelity of bright structures, thereby ensuring balanced enhancement across the entire dynamic range of the image. Meanwhile, we employ a multi-scale gamma correction strategy in the RGB space to generate image sequences with varying exposure levels, and calculate a saturation adjustment factor combining luminance and color difference information to restore color. Finally, we employ a multi-scale feature fusion strategy to extract structural features and texture details at different resolution levels, progressively guiding the reconstruction of high-quality remote sensing images. Extensive experimental results clearly demonstrate that our BDSA method consistently achieves superior dehazing performance across a wide range of remote sensing scenarios, empirically validating its advantage over state-of-the-art approaches in terms of contrast enhancement, detail restoration, and color fidelity.