<p>Underwater image enhancement is a crucial pre-processing task that supports various applications such as marine exploration, underwater robotics, environmental monitoring, archaeology, and biological studies. The underwater environment presents unique visual challenges due to light absorption, scattering, color distortion, and the presence of suspended particles, all of which degrade image quality. These issues result in reduced visibility, low contrast, and non-uniform illumination, complicating image analysis and interpretation. Traditional methods such as filtering, histogram equalization, dark channel prior, and image fusion have been widely applied to enhance underwater images. While effective under specific conditions, these approaches are limited by their dependence on hand-crafted features and assumptions. In recent years, the field has seen significant progress driven by deep learning techniques. Convolutional neural networks (CNNs), generative adversarial networks (GANs), transformer-based architectures, and state-space models have demonstrated superior performance in tasks such as restoration, dehazing, segmentation, and feature extraction under diverse underwater scenarios. This review presents a detailed overview of both traditional and deep learning-based underwater image enhancement techniques. Deep learning methods are categorized into different sections giving attention to their architectural design, performance, and application scope. Key underwater image datasets and widely used quality evaluation metrics, including UIQM and UCIQE, are also summarized. Persistent challenges such as noise, data scarcity, and non-uniform lighting remain. Future research directions are discussed, including physics-informed learning approaches and domain adaptation strategies. This review aims to provide a comprehensive understanding of the current landscape and promote continued advancements in underwater image enhancement.</p>