Scholarlink

<p>Visible and infrared image fusion aims to integrate multi-sensor information to produce high-quality images that enhance target visibility and texture under low-light conditions. However, existing methods often focus on single-domain feature fusion and overlook the interactions between spatial and frequency domains, resulting in poor texture details and visual effects in low-light conditions. To solve these difficulties, we propose using Spatial-Frequency feature for visible and infrared image fusion framework (SFVIF), which includes modules for Spatial feature extraction, Channel-Aware Fourier, and Local Fourier features, designed to extract spatial, global Channel-Aware Fourier, and local Fourier features, respectively. The features are subsequently combined using a feature fusion module, with a Cross-Gated Attention module further refining the fusion outcomes. Experimental results show that SFVIF generates images with detailed textures and strong contrast in challenging low-light conditions, achieving better fusion quality, visual effects and robustness than state-of-the-art methods on the MSRS, Roadscene, and <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="10586_2025_5137_Article_IEq1.gif" Format="GIF" Height="17" Rendition="HTML" Resolution="72" Type="Linedraw" Width="22" /> </InlineMediaObject> <EquationSource Format="TEX">\(\hbox {M}^3\)</EquationSource> <EquationSource Format="MATHML"><math> <msup> <mtext>M</mtext> <mn>3</mn> </msup> </math></EquationSource> </InlineEquation>FD datasets.</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

Using spatial-frequency features for visible and infrared image fusion

  • 期刊论文

  • 发表日期：

    2025年7月31日

查看全文
  • Fen Zhao,
  • Jun Guo,
  • Hongan Pan,
  • Yong Tang

摘要

Visible and infrared image fusion aims to integrate multi-sensor information to produce high-quality images that enhance target visibility and texture under low-light conditions. However, existing methods often focus on single-domain feature fusion and overlook the interactions between spatial and frequency domains, resulting in poor texture details and visual effects in low-light conditions. To solve these difficulties, we propose using Spatial-Frequency feature for visible and infrared image fusion framework (SFVIF), which includes modules for Spatial feature extraction, Channel-Aware Fourier, and Local Fourier features, designed to extract spatial, global Channel-Aware Fourier, and local Fourier features, respectively. The features are subsequently combined using a feature fusion module, with a Cross-Gated Attention module further refining the fusion outcomes. Experimental results show that SFVIF generates images with detailed textures and strong contrast in challenging low-light conditions, achieving better fusion quality, visual effects and robustness than state-of-the-art methods on the MSRS, Roadscene, and \(\hbox {M}^3\) M 3 FD datasets.