Scholarlink

<p>Underwater imaging poses significant challenges due to light absorption, scattering, and the presence of suspended particles, leading to degraded image quality. Traditional methods often fall short in addressing these complexities. This paper introduces DAW-Net, a comprehensive framework for underwater image enhancement that integrates multiple attention-based architectures, including Channel Attention, Spatial Attention, Color Correction, and Water-Type Classifier modules. The DAW-Net CNN module, with residual blocks, further enhances image quality. Experiments on benchmark and real-time datasets demonstrate DAW-Net’s superior performance over existing methods, achieving a PSNR of up to 33.86 dB and an SSIM of 0.9583 on certain datasets. Here, we show that DAW-Net effectively addresses color distortion, poor contrast, and blurriness, providing a robust solution for underwater image enhancement. This research contributes significantly to the fields of marine biology, environmental monitoring, and underwater inspections.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Underwater Imagery: A Deep Attention-Based deWater Network (DAW-Net) Approach

  • 期刊论文

  • 发表日期：

    2025年12月5日

查看全文
  • Vivek Kumar,
  • Kumaran P

摘要

Underwater imaging poses significant challenges due to light absorption, scattering, and the presence of suspended particles, leading to degraded image quality. Traditional methods often fall short in addressing these complexities. This paper introduces DAW-Net, a comprehensive framework for underwater image enhancement that integrates multiple attention-based architectures, including Channel Attention, Spatial Attention, Color Correction, and Water-Type Classifier modules. The DAW-Net CNN module, with residual blocks, further enhances image quality. Experiments on benchmark and real-time datasets demonstrate DAW-Net’s superior performance over existing methods, achieving a PSNR of up to 33.86 dB and an SSIM of 0.9583 on certain datasets. Here, we show that DAW-Net effectively addresses color distortion, poor contrast, and blurriness, providing a robust solution for underwater image enhancement. This research contributes significantly to the fields of marine biology, environmental monitoring, and underwater inspections.