<p>Enhancing images and videos in low-light conditions is a complex task that involves more than just adjusting brightness. Without fixing issues like artifacts, distortions, and noise in dark areas, simply increasing brightness can worsen overall quality. This paper presents an adaptive fusion of Dual Super-Resolution Generative Adversarial Networks (DSRGAN) and Top-Hat Gradient-Domain Filtering (THGDF) for low-light image and video enhancement on mobile devices. A soft thresholding mechanism combines the Memory Residual Super-Resolution Generative Adversarial Network (MRSRGAN) and the Weighted Perception Super-Resolution Generative Adversarial Network (WPSRGAN). MRSRGAN aims to improve fine details for better objective performance metrics, while WPSRGAN enhances overall details to improve subjective quality. Top-Hat Gradient-Domain Filtering is then used to reduce artifacts, distortions, and noise in both images and videos, resulting in significantly higher perceptual scores. The proposed method is validated through quality assessment metrics, including Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), and Information Fidelity Criterion (IFC). Extensive experiments on methods such as BGAN, RGSR, BDRVSR, RESRGAN, DBVSR, and BlindSR, using datasets like Sun-Hays, Urban100, Set5, Set11, and a mobile dataset, show that the proposed approach outperforms current state-of-the-art techniques.</p>