Scholarlink

<p>Image fusion technology, which integrates information from infrared and visible light sensors, is crucial for enhancing environmental perception in autonomous systems. However, most existing image fusion methods primarily emphasize intricate feature extraction and complex fusion strategies. As a result, they often fail to account for scene degradation, which severely impairs fusion quality in extreme environments. To address this problem, we propose SDCFuse, an infrared-visible image fusion network equipped with a scene degradation correction function. The network integrates a vision-language-based degradation discriminator for accurate identification of degradation types and a prior-guided correction mechanism to adaptively restore image quality. It effectively compensates for various forms of scene degradation, including haze, low light, and overexposure, thereby enhancing model performance under extreme conditions. Experimental results on four datasets demonstrate SDCFuse’s superior performance in correcting severe degradation, achieving state-of-the-art results with improvements of up to 25% in average gradient and 30% in entropy compared to existing methods. The source code and details are available at <a href="https://github.com/kayla0must/Image_Fusion_SDCFusion">https://github.com/kayla0must/Image_Fusion_SDCFusion</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

Scene degradation-aware fusion network for robust infrared and visible image synthesis in extreme conditions

  • 期刊论文

  • 发表日期：

    2025年12月12日

查看全文
  • Qingqing Hu,
  • Yiran Peng,
  • Zichun Shao,
  • Kintak U,
  • Junming Chen

摘要

Image fusion technology, which integrates information from infrared and visible light sensors, is crucial for enhancing environmental perception in autonomous systems. However, most existing image fusion methods primarily emphasize intricate feature extraction and complex fusion strategies. As a result, they often fail to account for scene degradation, which severely impairs fusion quality in extreme environments. To address this problem, we propose SDCFuse, an infrared-visible image fusion network equipped with a scene degradation correction function. The network integrates a vision-language-based degradation discriminator for accurate identification of degradation types and a prior-guided correction mechanism to adaptively restore image quality. It effectively compensates for various forms of scene degradation, including haze, low light, and overexposure, thereby enhancing model performance under extreme conditions. Experimental results on four datasets demonstrate SDCFuse’s superior performance in correcting severe degradation, achieving state-of-the-art results with improvements of up to 25% in average gradient and 30% in entropy compared to existing methods. The source code and details are available at https://github.com/kayla0must/Image_Fusion_SDCFusion.