Single-Image Super-Resolution via Cascaded Non-Local Mean Network and Dual-Path Multi-Branch Fusion

Image super-resolution (SR) aims to reconstruct high-resolution (HR) images from low-resolution (LR) inputs. It plays a crucial role in applications such as medical imaging, surveillance, and remote sensing. However, due to the ill-posed nature of the task and the inherent limitations of imaging sen...

Full description

Saved in:
Bibliographic Details
Main Authors: Yu Xu, Yi Wang
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Sensors
Subjects:
Online Access:https://www.mdpi.com/1424-8220/25/13/4044
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849427484213248000
author Yu Xu
Yi Wang
author_facet Yu Xu
Yi Wang
author_sort Yu Xu
collection DOAJ
description Image super-resolution (SR) aims to reconstruct high-resolution (HR) images from low-resolution (LR) inputs. It plays a crucial role in applications such as medical imaging, surveillance, and remote sensing. However, due to the ill-posed nature of the task and the inherent limitations of imaging sensors, obtaining accurate HR images remains challenging. While numerous methods have been proposed, the traditional approaches suffer from oversmoothing and limited generalization; CNN-based models lack the ability to capture long-range dependencies; and Transformer-based solutions, although effective in modeling global context, are computationally intensive and prone to texture loss. To address these issues, we propose a hybrid CNN–Transformer architecture that cascades a pixel-wise self-attention non-local means module (PSNLM) and an adaptive dual-path multi-scale fusion block (ADMFB). The PSNLM is inspired by the non-local means (NLM) algorithm. We use weighted patches to estimate the similarity between pixels centered at each patch while limiting the search region and constructing a communication mechanism across ranges. The ADMFB enhances texture reconstruction by adaptively aggregating multi-scale features through dual attention paths. The experimental results demonstrate that our method achieves superior performance on multiple benchmarks. For instance, in challenging ×4 super-resolution, our method outperforms the second-best method by 0.0201 regarding the Structural Similarity Index (SSIM) on the BSD100 dataset. On the texture-rich Urban100 dataset, our method achieves a 26.56 dB Peak Signal-to-Noise Ratio (PSNR) and 0.8133 SSIM.
format Article
id doaj-art-304177a1fb744eac99103dc439cdd140
institution Kabale University
issn 1424-8220
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Sensors
spelling doaj-art-304177a1fb744eac99103dc439cdd1402025-08-20T03:29:02ZengMDPI AGSensors1424-82202025-06-012513404410.3390/s25134044Single-Image Super-Resolution via Cascaded Non-Local Mean Network and Dual-Path Multi-Branch FusionYu Xu0Yi Wang1School of Computer and Control Engineering, Yantai University, Yantai 264005, ChinaSchool of Computer and Control Engineering, Yantai University, Yantai 264005, ChinaImage super-resolution (SR) aims to reconstruct high-resolution (HR) images from low-resolution (LR) inputs. It plays a crucial role in applications such as medical imaging, surveillance, and remote sensing. However, due to the ill-posed nature of the task and the inherent limitations of imaging sensors, obtaining accurate HR images remains challenging. While numerous methods have been proposed, the traditional approaches suffer from oversmoothing and limited generalization; CNN-based models lack the ability to capture long-range dependencies; and Transformer-based solutions, although effective in modeling global context, are computationally intensive and prone to texture loss. To address these issues, we propose a hybrid CNN–Transformer architecture that cascades a pixel-wise self-attention non-local means module (PSNLM) and an adaptive dual-path multi-scale fusion block (ADMFB). The PSNLM is inspired by the non-local means (NLM) algorithm. We use weighted patches to estimate the similarity between pixels centered at each patch while limiting the search region and constructing a communication mechanism across ranges. The ADMFB enhances texture reconstruction by adaptively aggregating multi-scale features through dual attention paths. The experimental results demonstrate that our method achieves superior performance on multiple benchmarks. For instance, in challenging ×4 super-resolution, our method outperforms the second-best method by 0.0201 regarding the Structural Similarity Index (SSIM) on the BSD100 dataset. On the texture-rich Urban100 dataset, our method achieves a 26.56 dB Peak Signal-to-Noise Ratio (PSNR) and 0.8133 SSIM.https://www.mdpi.com/1424-8220/25/13/4044image super-resolutionself-attentionCNNnon-local meansmulti-branch fusion
spellingShingle Yu Xu
Yi Wang
Single-Image Super-Resolution via Cascaded Non-Local Mean Network and Dual-Path Multi-Branch Fusion
Sensors
image super-resolution
self-attention
CNN
non-local means
multi-branch fusion
title Single-Image Super-Resolution via Cascaded Non-Local Mean Network and Dual-Path Multi-Branch Fusion
title_full Single-Image Super-Resolution via Cascaded Non-Local Mean Network and Dual-Path Multi-Branch Fusion
title_fullStr Single-Image Super-Resolution via Cascaded Non-Local Mean Network and Dual-Path Multi-Branch Fusion
title_full_unstemmed Single-Image Super-Resolution via Cascaded Non-Local Mean Network and Dual-Path Multi-Branch Fusion
title_short Single-Image Super-Resolution via Cascaded Non-Local Mean Network and Dual-Path Multi-Branch Fusion
title_sort single image super resolution via cascaded non local mean network and dual path multi branch fusion
topic image super-resolution
self-attention
CNN
non-local means
multi-branch fusion
url https://www.mdpi.com/1424-8220/25/13/4044
work_keys_str_mv AT yuxu singleimagesuperresolutionviacascadednonlocalmeannetworkanddualpathmultibranchfusion
AT yiwang singleimagesuperresolutionviacascadednonlocalmeannetworkanddualpathmultibranchfusion