Optimizing Underwater Vision: A Rigorous Investigation into Cnn's Deep Image Enhancement for Subaquatic Scenes

Abstract

In this paper, Convolutional Neural Networks were used to enhance the visual fidelity of underwater images. The UWCNN is introduced in this article, which utilizes underwater scene priors and a CNN model to improve underwater photos. The UWCNN model proposes a method where the clear latent underwater image is immediately rebuilt using the underwater scene as training data, rather than relying on parameter guessing in an underwater imaging model. Our UWCNN model may be used for frame-by-frame augmentation in underwater videos because to its lightweight network structure and efficient training data. In this dataset for underwater image deterioration by integrating an underwater photography physical model with optical characteristics of underwater landscapes are provided which includes a wide range of water types and degradation levels. Alternatively, one might choose to go light. A Neural Network (CNN) model is constructed using the related training data in order to enhance the quality of underwater scenes. Ultimately, the UWCNN model is used to improve the quality of underwater videos. The efficacy of our technology is shown via the analysis of both authentic and synthetic underwater photos.