Answers are grounded in this paper and cite their sources.
Comparative review evaluates classical and deep-learning image enhancement algorithms across aquatic datasets, highlighting performance trade-offs between physical models and generative networks.
Underwater image enhancement is an area of greater concern as underwater images present several hurdles due to the light propagation properties in aquatic environments. Critical application in areas of marine biology, offshore inspection, and underwater archaeology needs crystal clear images for fact-finding and inferring results. This review article comprises a comprehensive overview of methods for enhancing underwater images ranging from traditional techniques to advanced deep-learning approaches. Histogram-based approaches, filtering techniques, colour correction methods, and image fusion-based approaches are examined under conventional methods. In contrast, deep learning-based approaches including Convolutional Neural Networks (CNN) and Generative Adversarial Networks (GAN) produced potential results generating realistic underwater images. Firstly, we discuss the various pre-processing steps used in underwater image enhancement, succeeded by various noise reduction and image enhancement methods. A thorough and impartial analysis of existing literature is not currently available because of variations in the experimental setup and dataset used. Therefore, we analyse typical image enhancement algorithms on the HICRD dataset and evaluate them quantitatively and qualitatively based on the standard evaluation metrics. Utilizing the insights from classical and cutting-edge technologies, this review helps researchers select appropriate techniques for enhancing underwater imagery.
No takes yet. Share an insight, caveat, or question.
Kumar et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: