Scholarlink

<p>Synthetic aperture radar (SAR) images frequently suffer from speckle noise, which can greatly hinder subsequent processing and analysis tasks. As a&#xa0;result, speckle suppression is a&#xa0;challenge. Current deep learning-based approaches for speckle suppression often struggle to effectively combine local and global image features, frequently leading to the disruption of spatial continuity during the denoising process. To tackle these challenges, a&#xa0;multi-layer feature enhanced frequency correlation network (MFEFC-Net) was designed for speckle suppression. First, a&#xa0;multi-layer feature extraction block (MFEB) is developed to hierarchically extract deep representations from noise-affected images, enabling a&#xa0;more accurate characterization of the noise patterns. Second, a&#xa0;frequency correlator (FC) module is built to model the image over long distances and improve the gradient continuity of the denoised output. The FC module utilizes the discrete wavelet transform (DWT) to decompose the extracted deep features into high-frequency details and low-frequency structural components. It then leveraged a&#xa0;Transformer to establish frequency relationships by distinguishing diverse noise patterns across different sub-bands, ultimately removing noise while retaining more image details through the image feature enhancement (IFE) module. Finally, a&#xa0;feature decoder was employed to reconstruct the denoised image from the deep features, and a&#xa0;residual refine block (RRB) was utilized to restore more image details. Both quantitative evaluations and visual comparisons demonstrated that the proposed MFEFC-Net achieves outstanding performance compared to existing state-of-the-art denoising methods in both quantitative and qualitative assessments.</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

MFEFC-Net: A Multi-Layer Feature Enhanced Frequency Correlation Network for Speckle Suppression

  • 期刊论文

  • 发表日期：

    2026年5月7日

查看全文
  • Shuaiqi Liu,
  • Yonglu Zhou,
  • Zhiwei Zhang,
  • Yuhang Zhao,
  • Qi Hu,
  • Bing Li,
  • Yudong Zhang

摘要

Synthetic aperture radar (SAR) images frequently suffer from speckle noise, which can greatly hinder subsequent processing and analysis tasks. As a result, speckle suppression is a challenge. Current deep learning-based approaches for speckle suppression often struggle to effectively combine local and global image features, frequently leading to the disruption of spatial continuity during the denoising process. To tackle these challenges, a multi-layer feature enhanced frequency correlation network (MFEFC-Net) was designed for speckle suppression. First, a multi-layer feature extraction block (MFEB) is developed to hierarchically extract deep representations from noise-affected images, enabling a more accurate characterization of the noise patterns. Second, a frequency correlator (FC) module is built to model the image over long distances and improve the gradient continuity of the denoised output. The FC module utilizes the discrete wavelet transform (DWT) to decompose the extracted deep features into high-frequency details and low-frequency structural components. It then leveraged a Transformer to establish frequency relationships by distinguishing diverse noise patterns across different sub-bands, ultimately removing noise while retaining more image details through the image feature enhancement (IFE) module. Finally, a feature decoder was employed to reconstruct the denoised image from the deep features, and a residual refine block (RRB) was utilized to restore more image details. Both quantitative evaluations and visual comparisons demonstrated that the proposed MFEFC-Net achieves outstanding performance compared to existing state-of-the-art denoising methods in both quantitative and qualitative assessments.