文献类型：期刊文献

英文题名：A dual-path feature reuse multi-scale network for remote sensing image super-resolution

作者：Xiao, Huanling[1];Chen, Xintong[2];Luo, Liuhui[1];Lin, Cong[1,3]

机构：[1]Guangdong Ocean Univ, Sch Elect & Informat Engn, Zhanjiang 524000, Guangdong, Peoples R China;[2]Guangdong Ocean Univ, Sch Math & Comp Sci, Zhanjiang 524000, Guangdong, Peoples R China;[3]Hainan Univ, Coll Informat & Commun Engn, Haikou 570228, Hainan, Peoples R China

年份：2025

卷号：81

期号：1

外文期刊名：JOURNAL OF SUPERCOMPUTING

收录：SCI-EXPANDED(收录号：WOS:001337818400007)、、EI(收录号：20244317242419)、Scopus(收录号：2-s2.0-85206669759)、WOS

基金：This work was supported in part by the "Land, Sea, and Air" Aerospace Science and Technology Project: Innovation and Application of Hainan Vitality Index Based on Satellite Data (ATIC-202302001); the Stable Supporting Fund of the Acoustic Science and Technology Laboratory (JCK-YS2024604SSJS00301); the Undergraduate Innovation Team Project of Guangdong Ocean University (CXTD2024011); and the Open Fund of the Guangdong Provincial Key Laboratory of Intelligent Equipment for South China Sea Marine Ranching (2023B1212030003).

语种：英文

外文关键词：Remote sensing; Image super-resolution; Dual-path feature; Attention mechanism

外文摘要：Deep neural networks have achieved significant success in the super-resolution of remote sensing images. However, existing deep learning models still suffer from the issue of blurry pseudo-artifacts when restoring high-frequency details and textures. In this paper, a novel dual-path feature reuse multi-scale network (DFMNet) is proposed to more effectively utilize multi-scale features in remote sensing images, enhancing the detailed information in the restored images. Specifically, the designed dual-path feature reuse module adopts a symmetrical dual-path structure, with each path composed of convolutional layers of different sizes. This module enables deep feature reuse and multi-scale aggregation, improving the network's ability to handle and restore high-frequency details in the images. Furthermore, a cross-attention module is introduced to facilitate deep interactive fusion of multi-scale image features produced by the encoder output. Comparative experiments conducted on challenging UCMerced and AID remote sensing datasets demonstrate that the proposed DFMNet achieves superior performance in both objective and subjective evaluations.