Image super-resolution reconstruction based on multi-scale dual-attention
Image super-resolution reconstruction is one of the methods to improve resolution by learning the inherent features and attributes of images. However, the existing super-resolution models have some problems, such as missing details, distorted natural texture, blurred details and too smooth after ima...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Taylor & Francis Group
2023-03-01
|
| Series: | Connection Science |
| Subjects: | |
| Online Access: | http://dx.doi.org/10.1080/09540091.2023.2182487 |
| _version_ | 1797684031219826688 |
|---|---|
| author | Hong-an Li Diao Wang Jing Zhang Zhanli Li Tian Ma |
| author_facet | Hong-an Li Diao Wang Jing Zhang Zhanli Li Tian Ma |
| author_sort | Hong-an Li |
| collection | DOAJ |
| description | Image super-resolution reconstruction is one of the methods to improve resolution by learning the inherent features and attributes of images. However, the existing super-resolution models have some problems, such as missing details, distorted natural texture, blurred details and too smooth after image reconstruction. To solve the above problems, this paper proposes a Multi-scale Dual-Attention based Residual Dense Generative Adversarial Network (MARDGAN), which uses multi-branch paths to extract image features and obtain multi-scale feature information. This paper also designs the channel and spatial attention block (CSAB), which is combined with the enhanced residual dense block (ERDB) to extract multi-level depth feature information and enhance feature reuse. In addition, the multi-scale feature information extracted under the three-branch path is fused with global features, and sub-pixel convolution is used to restore the high-resolution image. The experimental results show that the objective evaluation index of MARDGAN on multiple benchmark datasets is higher than other methods, and the subjective visual effect is better. This model can effectively use the original image information to restore the super-resolution image with clearer details and stronger authenticity. |
| first_indexed | 2024-03-12T00:23:28Z |
| format | Article |
| id | doaj.art-d20997b33fdb441f95eeecb2575f8a71 |
| institution | Directory Open Access Journal |
| issn | 0954-0091 1360-0494 |
| language | English |
| last_indexed | 2024-03-12T00:23:28Z |
| publishDate | 2023-03-01 |
| publisher | Taylor & Francis Group |
| record_format | Article |
| series | Connection Science |
| spelling | doaj.art-d20997b33fdb441f95eeecb2575f8a712023-09-15T10:48:01ZengTaylor & Francis GroupConnection Science0954-00911360-04942023-03-010011910.1080/09540091.2023.21824872182487Image super-resolution reconstruction based on multi-scale dual-attentionHong-an Li0Diao Wang1Jing Zhang2Zhanli Li3Tian Ma4Xi'an University of Science and TechnologyXi'an University of Science and TechnologyXi'an University of Science and TechnologyXi'an University of Science and TechnologyXi'an University of Science and TechnologyImage super-resolution reconstruction is one of the methods to improve resolution by learning the inherent features and attributes of images. However, the existing super-resolution models have some problems, such as missing details, distorted natural texture, blurred details and too smooth after image reconstruction. To solve the above problems, this paper proposes a Multi-scale Dual-Attention based Residual Dense Generative Adversarial Network (MARDGAN), which uses multi-branch paths to extract image features and obtain multi-scale feature information. This paper also designs the channel and spatial attention block (CSAB), which is combined with the enhanced residual dense block (ERDB) to extract multi-level depth feature information and enhance feature reuse. In addition, the multi-scale feature information extracted under the three-branch path is fused with global features, and sub-pixel convolution is used to restore the high-resolution image. The experimental results show that the objective evaluation index of MARDGAN on multiple benchmark datasets is higher than other methods, and the subjective visual effect is better. This model can effectively use the original image information to restore the super-resolution image with clearer details and stronger authenticity.http://dx.doi.org/10.1080/09540091.2023.2182487super-resolution reconstructiongenerative adversarial networkattention mechanismloss function |
| spellingShingle | Hong-an Li Diao Wang Jing Zhang Zhanli Li Tian Ma Image super-resolution reconstruction based on multi-scale dual-attention Connection Science super-resolution reconstruction generative adversarial network attention mechanism loss function |
| title | Image super-resolution reconstruction based on multi-scale dual-attention |
| title_full | Image super-resolution reconstruction based on multi-scale dual-attention |
| title_fullStr | Image super-resolution reconstruction based on multi-scale dual-attention |
| title_full_unstemmed | Image super-resolution reconstruction based on multi-scale dual-attention |
| title_short | Image super-resolution reconstruction based on multi-scale dual-attention |
| title_sort | image super resolution reconstruction based on multi scale dual attention |
| topic | super-resolution reconstruction generative adversarial network attention mechanism loss function |
| url | http://dx.doi.org/10.1080/09540091.2023.2182487 |
| work_keys_str_mv | AT honganli imagesuperresolutionreconstructionbasedonmultiscaledualattention AT diaowang imagesuperresolutionreconstructionbasedonmultiscaledualattention AT jingzhang imagesuperresolutionreconstructionbasedonmultiscaledualattention AT zhanlili imagesuperresolutionreconstructionbasedonmultiscaledualattention AT tianma imagesuperresolutionreconstructionbasedonmultiscaledualattention |