Underwater Image Formation

Underwater image formation research focuses on mitigating the distortions caused by water's selective absorption and scattering of light, aiming to improve the quality and accuracy of underwater images and videos. Current research heavily utilizes deep learning, employing architectures like Vision Transformers and convolutional neural networks, often integrated with physically-based underwater image formation models to guide the learning process and improve realism. This work is crucial for advancing various marine applications, including autonomous underwater vehicle navigation, marine resource exploration, and ecological monitoring, by enabling more reliable image analysis and 3D scene reconstruction.