Remote sensing image Super-resolution reconstruction by fusing multi-scale receptive fields and hybrid transformer
Abstract To enhance high-frequency perceptual information and texture details in remote sensing images and address the challenges of super-resolution reconstruction algorithms during training, particularly the issue of missing details, this paper proposes an improved remote sensing image super-resol...
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2025-01-01
|
| Series: | Scientific Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1038/s41598-025-86446-5 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832594755491463168 |
|---|---|
| author | Denghui Liu Lin Zhong Haiyang Wu Songyang Li Yida Li |
| author_facet | Denghui Liu Lin Zhong Haiyang Wu Songyang Li Yida Li |
| author_sort | Denghui Liu |
| collection | DOAJ |
| description | Abstract To enhance high-frequency perceptual information and texture details in remote sensing images and address the challenges of super-resolution reconstruction algorithms during training, particularly the issue of missing details, this paper proposes an improved remote sensing image super-resolution reconstruction model. The generator network of the model employs multi-scale convolutional kernels to extract image features and utilizes a multi-head self-attention mechanism to dynamically fuse these features, significantly improving the ability to capture both fine details and global information in remote sensing images. Additionally, the model introduces a multi-stage Hybrid Transformer structure, which processes features at different resolutions progressively, from low resolution to high resolution, substantially enhancing reconstruction quality and detail recovery. The discriminator combines multi-scale convolution, global Transformer, and hierarchical feature discriminators, providing a comprehensive and refined evaluation of image quality. Finally, the model incorporates a Charbonnier loss function and total variation (TV) loss function, which significantly improve training stability and accelerate convergence. Experimental results demonstrate that the proposed method, compared to the SRGAN algorithm, achieves average improvements of approximately 3.61 dB in Peak Signal-to-Noise Ratio (PSNR), 0.070 (8.2%) in Structural Similarity Index (SSIM), and 0.030 (3.1%) in Feature Similarity Index (FSIM) across multiple datasets, showing significant performance gains. |
| format | Article |
| id | doaj-art-45f377b851b54b59a569cc2e4bdcc2e7 |
| institution | Kabale University |
| issn | 2045-2322 |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Scientific Reports |
| spelling | doaj-art-45f377b851b54b59a569cc2e4bdcc2e72025-01-19T12:21:11ZengNature PortfolioScientific Reports2045-23222025-01-0115111410.1038/s41598-025-86446-5Remote sensing image Super-resolution reconstruction by fusing multi-scale receptive fields and hybrid transformerDenghui Liu0Lin Zhong1Haiyang Wu2Songyang Li3Yida Li4School of Electronics and Information, Xijing UniversitySchool of Electronics and Information, Xijing UniversitySchool of Electronics and Information, Xijing UniversitySchool of Electronics and Information, Xijing UniversitySchool of Electronics and Information, Xijing UniversityAbstract To enhance high-frequency perceptual information and texture details in remote sensing images and address the challenges of super-resolution reconstruction algorithms during training, particularly the issue of missing details, this paper proposes an improved remote sensing image super-resolution reconstruction model. The generator network of the model employs multi-scale convolutional kernels to extract image features and utilizes a multi-head self-attention mechanism to dynamically fuse these features, significantly improving the ability to capture both fine details and global information in remote sensing images. Additionally, the model introduces a multi-stage Hybrid Transformer structure, which processes features at different resolutions progressively, from low resolution to high resolution, substantially enhancing reconstruction quality and detail recovery. The discriminator combines multi-scale convolution, global Transformer, and hierarchical feature discriminators, providing a comprehensive and refined evaluation of image quality. Finally, the model incorporates a Charbonnier loss function and total variation (TV) loss function, which significantly improve training stability and accelerate convergence. Experimental results demonstrate that the proposed method, compared to the SRGAN algorithm, achieves average improvements of approximately 3.61 dB in Peak Signal-to-Noise Ratio (PSNR), 0.070 (8.2%) in Structural Similarity Index (SSIM), and 0.030 (3.1%) in Feature Similarity Index (FSIM) across multiple datasets, showing significant performance gains.https://doi.org/10.1038/s41598-025-86446-5Remote sensing imageImage Super-resolutionGANAttention mechanismHybrid transformerMulti-scale feature extraction |
| spellingShingle | Denghui Liu Lin Zhong Haiyang Wu Songyang Li Yida Li Remote sensing image Super-resolution reconstruction by fusing multi-scale receptive fields and hybrid transformer Scientific Reports Remote sensing image Image Super-resolution GAN Attention mechanism Hybrid transformer Multi-scale feature extraction |
| title | Remote sensing image Super-resolution reconstruction by fusing multi-scale receptive fields and hybrid transformer |
| title_full | Remote sensing image Super-resolution reconstruction by fusing multi-scale receptive fields and hybrid transformer |
| title_fullStr | Remote sensing image Super-resolution reconstruction by fusing multi-scale receptive fields and hybrid transformer |
| title_full_unstemmed | Remote sensing image Super-resolution reconstruction by fusing multi-scale receptive fields and hybrid transformer |
| title_short | Remote sensing image Super-resolution reconstruction by fusing multi-scale receptive fields and hybrid transformer |
| title_sort | remote sensing image super resolution reconstruction by fusing multi scale receptive fields and hybrid transformer |
| topic | Remote sensing image Image Super-resolution GAN Attention mechanism Hybrid transformer Multi-scale feature extraction |
| url | https://doi.org/10.1038/s41598-025-86446-5 |
| work_keys_str_mv | AT denghuiliu remotesensingimagesuperresolutionreconstructionbyfusingmultiscalereceptivefieldsandhybridtransformer AT linzhong remotesensingimagesuperresolutionreconstructionbyfusingmultiscalereceptivefieldsandhybridtransformer AT haiyangwu remotesensingimagesuperresolutionreconstructionbyfusingmultiscalereceptivefieldsandhybridtransformer AT songyangli remotesensingimagesuperresolutionreconstructionbyfusingmultiscalereceptivefieldsandhybridtransformer AT yidali remotesensingimagesuperresolutionreconstructionbyfusingmultiscalereceptivefieldsandhybridtransformer |