Scholarlink

The problem of haze, incorrect color tones, and poor visibility in underwater images arises primarily due to backscattering and light absorption. To address this, the paper proposes a novel Convolutional Neural Network (CNN)-based approach designed specifically for real-time underwater image dehazing and color correction. Recognizing the inherent variability in underwater environments, the model is trained and evaluated using four diverse datasets—UIEB, EUVP, LSUI, and SQUID—enhancing its adaptability to different underwater conditions. The approach also employs four performance metrics: PSNR, SSIM, UCIQE, and UIQM, which collectively demonstrate its superiority over existing state-of-the-art methods in terms of visibility, color accuracy, and overall image quality. Notably, the proposed method achieves this while maintaining real-time processing capabilities, making it highly suitable for applications such as underwater exploration, monitoring, and robotics where both speed and accuracy are critical.

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

DCCUI: A Novel Technique for Dehazing and Colour Correction of Real-Time Underwater Images

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Sandeep Vishwakarma,
  • Anuradha Pillai,
  • Deepika Punj,
  • Atul Srivastava

摘要

The problem of haze, incorrect color tones, and poor visibility in underwater images arises primarily due to backscattering and light absorption. To address this, the paper proposes a novel Convolutional Neural Network (CNN)-based approach designed specifically for real-time underwater image dehazing and color correction. Recognizing the inherent variability in underwater environments, the model is trained and evaluated using four diverse datasets—UIEB, EUVP, LSUI, and SQUID—enhancing its adaptability to different underwater conditions. The approach also employs four performance metrics: PSNR, SSIM, UCIQE, and UIQM, which collectively demonstrate its superiority over existing state-of-the-art methods in terms of visibility, color accuracy, and overall image quality. Notably, the proposed method achieves this while maintaining real-time processing capabilities, making it highly suitable for applications such as underwater exploration, monitoring, and robotics where both speed and accuracy are critical.