Scholarlink

<p>Underwater image enhancement is crucial for marine research, yet challenges posed by haze and noise significantly degrade image quality. This study presents a lightweight deep hybrid convolutional neural network framework (AB-HCNN) incorporating an attention mechanism to enhance underwater images. The proposed model combines a lightweight dilated CNN with a convolutional block attention module and a modified U-Net architecture for upsampling. Through experiments on publicly available datasets, our framework achieves an average structural similarity index of 0.85 and a peak signal-to-noise ratio of 25.39&#xa0;dB, demonstrating its effectiveness in improving image clarity and color accuracy. The AB-HCNN contributes to restoring and enhancing underwater images, addressing color attenuation, low contrast, and blurring issues.</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

Lightweight deep hybrid CNN with attention mechanism for enhanced underwater image restoration

  • 期刊论文

  • 发表日期：

    2025年1月30日

查看全文
  • V. Karthikeyan,
  • S. Praveen,
  • S. Sudeep Nandan

摘要

Underwater image enhancement is crucial for marine research, yet challenges posed by haze and noise significantly degrade image quality. This study presents a lightweight deep hybrid convolutional neural network framework (AB-HCNN) incorporating an attention mechanism to enhance underwater images. The proposed model combines a lightweight dilated CNN with a convolutional block attention module and a modified U-Net architecture for upsampling. Through experiments on publicly available datasets, our framework achieves an average structural similarity index of 0.85 and a peak signal-to-noise ratio of 25.39 dB, demonstrating its effectiveness in improving image clarity and color accuracy. The AB-HCNN contributes to restoring and enhancing underwater images, addressing color attenuation, low contrast, and blurring issues.