DNTS: Experimental Evaluation of Underwater Image Quality Enhancement Scheme Based on Dense Neural Transformation Scheme

Underwater image quality faces challenges such as light absorption, scattering, and color distortion. This paper proposes a new underwater image enhancement scheme. It is called Dense Neural Transformation Scheme, abbreviated as DNTS, which has been cross-validated with classical deep learning approach called Convolutional Neural Network, CNN, to assess the efficacy of the proposed model. The proposed method uses a dense neural network architecture tailored to overcome the characteristics of underwater imaging. This architecture has the capability to learn complex transformations that could correct color aberration, enhance contrast, and visually restore the quality of the image. The DNTS architecture is based on a densely connected layer which allows the model to accept quite high levels of image degradation. This makes it possible to perform efficient feature extraction and refinement at any depth in the network. That's why our approach is validated by improved visible detail and more accurate color representation of the restored images. This work presents a comprehensive description of DNTS capabilities and its potential applications in marine studies and underwater exploration. The proposed approach yields a robust solution toward improving the quality of underwater images and enhancing more discriminative visual observations in underwater scenarios.