Underwater Image Enhancement Algorithm Based on UWGAN and U-Net

Abstract

The underwater environment presents unique characteristics that often result in defects like low contrast and blurred edges. To address these issues and improve underwater target recognition for vehicles, a new method that combines UWMAN and U-Net algorithms has been developed. One key advantage of this algorithm is its ability to counteract the impact of underwater environmental factors. It achieves this by enhancing the color of the images and improving various details, leading to higher scores in evaluations of underwater color image quality and overall image quality. This means that the algorithm effectively enhances the clarity and visual appeal of underwater images. Additionally, the enhanced images obtained through this method demonstrate improved matching effects in feature point matching experiments. This indicates that the algorithm enhances the identification and alignment of crucial features within the images, resulting in more accurate target recognition. These capabilities are vital for underwater vehicles operating in challenging underwater environments. Overall, the combination of UWMAN and U-Net algorithms represents a significant advancement in enhancing image clarity and improving the accuracy of underwater target recognition. The algorithm’s ability to mitigate the impact of underwater environmental factors and produce visually pleasing images holds great potential for various applications related to underwater exploration, marine research, and inspection tasks beneath the water’s surface.