Scholarlink

<p>In recent years, various deep learning-based methods have been proposed for image dehazing. However, these methods often struggle with adaptability to diverse atmospheric conditions due to the overlook of polarization information. To address this issue, we present a polarization-aware image dehazing method that incorporates the Segment Anything Model (SAM) for enhanced performance. Our approach leverages a polarization-inspired structural design to improve feature representation capabilities and integrates SAM to support the network’s parameter learning process. By accurately estimating the degree of polarization (DoP) features of both transmitted light and airlight, as well as the atmospheric light features, our method achieves more accurate dehazing. Furthermore, the SAM-guided fusion network refines dehazing features, resulting in high-quality dehazing images. Extensive experiments on both synthetic and real-world datasets demonstrate the superiority of our method over state-of-the-art approaches, highlighting its improved adaptability and effectiveness in various scenarios. Code is available at <a href="https://github.com/chenhuiphd/IPPIS">https://github.com/chenhuiphd/IPPIS</a>.</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 image dehazing with polarization awareness and SAM-guided fusion

  • 期刊论文

  • 发表日期：

    2025年7月14日

查看全文
  • Hui Chen,
  • Nannan Li,
  • Ming An,
  • Chengxi Xia,
  • Kekun Zhu

摘要

In recent years, various deep learning-based methods have been proposed for image dehazing. However, these methods often struggle with adaptability to diverse atmospheric conditions due to the overlook of polarization information. To address this issue, we present a polarization-aware image dehazing method that incorporates the Segment Anything Model (SAM) for enhanced performance. Our approach leverages a polarization-inspired structural design to improve feature representation capabilities and integrates SAM to support the network’s parameter learning process. By accurately estimating the degree of polarization (DoP) features of both transmitted light and airlight, as well as the atmospheric light features, our method achieves more accurate dehazing. Furthermore, the SAM-guided fusion network refines dehazing features, resulting in high-quality dehazing images. Extensive experiments on both synthetic and real-world datasets demonstrate the superiority of our method over state-of-the-art approaches, highlighting its improved adaptability and effectiveness in various scenarios. Code is available at https://github.com/chenhuiphd/IPPIS.