Scholarlink

Abstract <p>A method for reconstructing fragments of optical satellite images is proposed on the basis of a deep neural network that uses a digital image acquired for the same area by a synthetic aperture radar as additional (conditional) information. The resulting solution is based on existing deep neural networks (DiffCR and Palette) used in image reconstruction tasks, with a modified spatial–channel attention module (convolutional block attention module–CBAM) and a modified loss function. Experimental studies have demonstrated improved quantitative characteristics and visual results compared to baseline models.</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

Reconstruction of Optical Remote Sensing Images Using Synthetic Aperture Radar Data and Diffusion Neural Networks

  • 期刊论文

  • 发表日期：

    2025年8月1日

查看全文
  • V. F. Konovalov,
  • V. V. Myasnikov,
  • V. V. Sergeev

摘要

Abstract

A method for reconstructing fragments of optical satellite images is proposed on the basis of a deep neural network that uses a digital image acquired for the same area by a synthetic aperture radar as additional (conditional) information. The resulting solution is based on existing deep neural networks (DiffCR and Palette) used in image reconstruction tasks, with a modified spatial–channel attention module (convolutional block attention module–CBAM) and a modified loss function. Experimental studies have demonstrated improved quantitative characteristics and visual results compared to baseline models.