Homotopy-based enhancement for robust visibility in low-light imaging
S. Shivam Kumar Jha, N. Mohana · The Imaging Science Journal · 2025
This paper introduces a novel Homotopy-Based Adaptive Image Enhancement (HBAIE) method for addressing the challenges of low-light image correction. The proposed approach leverages homotopy-based transformations in conjunction with algebraic operations in the min-plus and max-plus domains to adaptively enhance brightness and contrast while preserving fine structural details. A key contribution is the integration of brightness-awareness into the enhancement kernel, enabling context-sensitive adjustments that improve visual coherence. Unlike conventional methods, which often produce artifacts in severely underexposed conditions, the proposed framework dynamically adapts to illumination variations with minimal distortion. It is computationally efficient and suitable for real-time applications. Experimental results on the LOL and ExDark dataset demonstrate that the method consistently outperforms several state-of-the-art techniques in terms of PSNR, SSIM, and EME.