Sustainable Environmental Monitoring: Multistage Fusion Algorithm for Remotely Sensed Underwater Super-Resolution Image Enhancement and Classification

Abstract

Oceans and seas cover more than 70% of the Earth's surface. If compared with the land mass there are a lot of unexplored locations, a wealth of natural resources, and diverse ocean creatures that are inaccessible to us humans. Underwater rovers and vehicles play a vital role in discovering these resources, yet limited visibility in deep waters and technological constraints impede underwater exploration. To address these issues, advanced image super-resolution and enhancement techniques are crucial for reliable resource identification, species recognition, and underwater ecosystem study. This will ultimately bridge the current gap in environmental monitoring by facilitating resource tracking and underwater waste assessment. This article proposes a novel multistage fusion algorithm for underwater image super-resolution, designed to enhance the quality and spatial resolution of low-resolution underwater images toward a more accurate object characterization. The effectiveness of the proposed super-resolution technique is demonstrated using multiple performance metrics including accuracy, f1-score, recall, and precision. By enhancing the spatial resolution of underwater images, our approach meets the increasing demand for detailed and accurate information in underwater earth observation applications.

Keywords

• Image classification
• image enhancement
• image processing
• image recognition
• remote sensing