MD-AGC: A Multi-Descriptor Adaptive Gamma Correction for Enhanced Urban Mapping and Smart City Analysis

Radiometric inconsistencies caused by highly heterogeneous illumination conditions remain a major challenge in the processing of satellite imagery for urban analysis and smart-city applications. Shadows, overexposed surfaces, and uneven luminance distributions degrade both visual interpretability and the reliability of downstream tasks such as segmentation, object extraction, and change detection. Existing gamma correction strategies either rely on fixed parameters or depend on a single global descriptor, limiting their ability to generalize across complex urban scenes. To address these limitations, we propose a novel Multi-Descriptor Adaptive Gamma Correction (MD-AGC) algorithm that jointly integrates luminance, histogram dispersion, and structural contrast into a unified illumination model. The fused descriptor guides the estimation of an adaptive gamma value specifically tailored to the radiometric characteristics of each image region. Extensive experiments conducted on diverse satellite datasets demonstrate that MD-AGC provides superior enhancement quality, improved radiometric normalization, and consistent gains in downstream smart-city pipelines. The proposed method offers a lightweight yet robust solution suitable for large-scale operational processing.