Scholarlink

<p>Infrared image super-resolution (SR) remains a challenging task due to inherent limitations in existing approaches: convolutional neural network (CNN)-based approaches struggle with long-range dependency modeling, whereas transformer-based approaches are computationally expensive and tend to overlook fine local details. To address these issues, we propose a novel hybrid perception enhancement network (HPEN). Its core component is a hybrid perception enhancement block (HPEB), which effectively combines a token aggregation block (TAB) for global context modeling, a multi-scale feature enhancement block (MFEB) for local detail extraction, and a convolutional layer for feature refinement. Extensive experimental results demonstrate that the proposed HPEN achieves leading performance among compared methods. For the challenging <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\times 4\)</EquationSource> </InlineEquation> SR task, it attains the best PSNR and SSIM values among the evaluated lightweight SR approaches, while demonstrating remarkable efficiency advantages. Specifically, compared to HiT-SR, HPEN reduces FLOPs by 42.9%, uses only 9.4% of the GPU memory usage, and delivers a <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(2.3\times\)</EquationSource> </InlineEquation> faster inference speed. The code is available at <a href="https://github.com/smilenorth1/HPEN-main">https://github.com/smilenorth1/HPEN-main</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

A lightweight hybrid perception enhancement network for infrared image super-resolution

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月29日

查看全文
  • Zepeng Liu,
  • Jiya Tian,
  • Chao Liu,
  • Guodong Zhang

摘要

Infrared image super-resolution (SR) remains a challenging task due to inherent limitations in existing approaches: convolutional neural network (CNN)-based approaches struggle with long-range dependency modeling, whereas transformer-based approaches are computationally expensive and tend to overlook fine local details. To address these issues, we propose a novel hybrid perception enhancement network (HPEN). Its core component is a hybrid perception enhancement block (HPEB), which effectively combines a token aggregation block (TAB) for global context modeling, a multi-scale feature enhancement block (MFEB) for local detail extraction, and a convolutional layer for feature refinement. Extensive experimental results demonstrate that the proposed HPEN achieves leading performance among compared methods. For the challenging \(\times 4\) SR task, it attains the best PSNR and SSIM values among the evaluated lightweight SR approaches, while demonstrating remarkable efficiency advantages. Specifically, compared to HiT-SR, HPEN reduces FLOPs by 42.9%, uses only 9.4% of the GPU memory usage, and delivers a \(2.3\times\) faster inference speed. The code is available at https://github.com/smilenorth1/HPEN-main.