An Overview of Underwater Image Enhancement Techniques

As one of the important carriers for understanding the underwater environment, underwater images have important research significance in the fields of marine ecology, marine resources and marine environment monitoring. However, the degradation inherent in underwater images, such as color distortion due to light absorption and scattering, severely hampers their effectiveness in downstream applications, including object detection, recognition and tracking. Therefore, underwater image enhancement methods are crucial for restoring image quality and enhancing visual information. This paper aims to introduce several diverse enhancement approaches, including image-based, physics-model-based, and deep-learning-based. Additionally, we discuss relevant underwater image datasets and quality evaluation metrics. By doing so, we aim to provide insights into the landscape of underwater image enhancement research and its implications for advancing underwater vision systems.