Contrast-Guided Adaptive Multiscale Retinex for Enhanced Satellite Image Visualization

Satellite images generally have low contrast, non- uniform illumination, and structural details visibility suffer from atmospheric interference, sensor limitations, and different acquisition conditions, which greatly hamper visual analysis and interpretation. In this paper, a Contrast-Guided Adaptive Multiscale Retinex (CG-AMSR) approach is proposed for satellite image visualization enhancement. Based on the theory of Multiscale Retinex with New Kernel (MSRNK), the authors of the method adaptive contrast guidance to control enhancement strength locally while retaining natural appearance. Multiscale processing is used for balancing global illumination correction and local detail enhancement, and an improved kernel formulation is implemented to stabilize illumination normalization without generating halo artifacts or saturation effects. Moreover, contrast-aware weighting and normalization techniques are used to preserve structural integrity and maintain the overall tonal balance. The authors assess the proposed model's effectiveness through visual examination and by using objective quality indices, such as entropy, PSNR, SSIM, and contrast enhancement index. The proposed approach outperforms the other three baseline methods quantitatively and qualitatively in terms of contrast enhancement as well as detail preservation. Also, the resultant images from the proposed method show more visual clarity and popularity. As such, the method can be considered as a viable candidate for handling real-world satellite image analysis and visualization problems.