Underwater imaging is needed for uses like marine exploration, autonomous underwater vehicles (AUVs), environmental monitoring, and underwater robots. Image quality suffers, though, from physical effects such as light attenuation, scattering, and turbidity. The most significant of these is low-light conditions, particularly common in deep or turbid waters, which drastically decrease visibility, contrast, and color accuracy—impeding both visual interpretation and the ability of computer vision tasks like object detection and tracking. This work suggests a low-light underwater image enhancement framework based on quality-aware and degradation-specific enhancement. The framework is initiated with no-reference quality assessment and perceptual models for setting baseline image quality. It continues to the evaluation of enhancement performance in terms of benchmark measures including UIQM, UCIQE, PSNR, and SSIM. The framework features a degradation classification module to determine the distortion type with a light-weight classifier trained with labeled underwater images. For low-light images, a specific enhancement module is invoked, either using classical Retinex-based methods or deep learning frameworks such as attention-guided CNNs or GANs. Improved images are again subjected to a second quality evaluation to ensure improvement. The adaptive modular pipeline guarantees that only enhanced images are delivered to downstream applications. The iterative nature of the framework allows scalable enhancement of different types of degradations while permitting real-time testing and tuning. Experimental outcomes on standard test datasets demonstrate remarkable enhancements of visibility, contrast, and color balance, which validate the efficiency of the low-light enhancement module. The method presents a robust and adaptable solution for underwater image improvement, especially valuable for real-time AUV navigation and self-contained underwater perception in adverse marine conditions.