Local Contrast Enhancement in LDR Images via Adaptive Distribution of Clipped-histogram Excess

We propose a multiscale histogram excess redistribution framework addressing key limitations of contrastlimited adaptive histogram equalization (CLAHE) and Retinex methods-poor dark-region recovery, block artifacts, and color distortion. The method exploits coarse scales for global contrast, fine scales for local details, and medium scales for balanced fusion. At each scale, the redistribution ratio is adaptively determined by tile brightness: coarse scales redirect excess to originally occupied bins while finer scales regulate enhancement for smooth transitions. Scale-specific lookup tables (LUTs) are interpolated and fused with background restoration, chroma compensation, and asymmetric Gaussian offset ensuring natural tone. Results demonstrate superior detail preservation and color fidelity in low-light and high-dynamic-range (HDR) scenarios compared with LLE-NET, Retinex-CLAHE, and Kwon et al.