A Multibranch Deep Neural Network for the Superresolution of Solar Magnetograms
The existing superresolution (SR) models for solar magnetograms are mostly borrowed from the SR models for natural images. They are less effective for processing solar magnetograms with a very large dynamic range and very rich image features. In this paper, a multibranch superresolution (MBSR) model...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IOP Publishing
2024-01-01
|
| Series: | The Astrophysical Journal Supplement Series |
| Subjects: | |
| Online Access: | https://doi.org/10.3847/1538-4365/ad1760 |
| _version_ | 1797310964117274624 |
|---|---|
| author | Fengping Dou Long Xu Dong Zhao Zhixiang Ren |
| author_facet | Fengping Dou Long Xu Dong Zhao Zhixiang Ren |
| author_sort | Fengping Dou |
| collection | DOAJ |
| description | The existing superresolution (SR) models for solar magnetograms are mostly borrowed from the SR models for natural images. They are less effective for processing solar magnetograms with a very large dynamic range and very rich image features. In this paper, a multibranch superresolution (MBSR) model is specially designed for solar magnetograms. First, we split a low-resolution magnetogram into a group of overlapping image patches, and classify them into three categories according to magnetic flux intensity, namely simple, medium, and complex. Then, image patches of each category are fed into the corresponding branch of the MBSR network, the lightweight branch for simple image patches and the heavyweight one for complex image patches. The advantage of such a strategy is twofold. On the one hand, active regions are allocated more computational resources to train a heavyweight branch more fully, while quiet regions are allocated fewer computational resources to train a lightweight branch for saving computational resources. On the other hand, a lightweight network with a simple nonlinear function is preferable to simple regions, while a heavyweight one may be underfitting. Additionally, to verify the effectiveness of the proposed model, a magnetic field structure similarity metric is proposed to measure the artifacts of the generated high-resolution (HR) magnetograms. Experimental results show that the proposed MBSR model generates HR magnetograms highly consistent with the HMI ones, and achieves the best performance over five objective metrics, including peak signal-to-noise ratio and structure similarity, etc. |
| first_indexed | 2024-03-08T01:51:37Z |
| format | Article |
| id | doaj.art-22c299e55e2a4e09ae8adc9f44ed0ee7 |
| institution | Directory Open Access Journal |
| issn | 0067-0049 |
| language | English |
| last_indexed | 2024-03-08T01:51:37Z |
| publishDate | 2024-01-01 |
| publisher | IOP Publishing |
| record_format | Article |
| series | The Astrophysical Journal Supplement Series |
| spelling | doaj.art-22c299e55e2a4e09ae8adc9f44ed0ee72024-02-14T09:40:55ZengIOP PublishingThe Astrophysical Journal Supplement Series0067-00492024-01-012711910.3847/1538-4365/ad1760A Multibranch Deep Neural Network for the Superresolution of Solar MagnetogramsFengping Dou0Long Xu1https://orcid.org/0000-0002-9286-2876Dong Zhao2Zhixiang Ren3State Key Laboratory of Space Weather, National Space Science Center, Chinese Academy of Sciences , Beijing 100190, People's Republic of ChinaFaculty of Electrical Engineering and Computer Science, Ningbo University , People's Republic of China ; lxu@nao.cas.cn; Peng Cheng Laboratory , Shenzhen 518000, People's Republic of ChinaState Key Laboratory of Space Weather, National Space Science Center, Chinese Academy of Sciences , Beijing 100190, People's Republic of ChinaPeng Cheng Laboratory , Shenzhen 518000, People's Republic of ChinaThe existing superresolution (SR) models for solar magnetograms are mostly borrowed from the SR models for natural images. They are less effective for processing solar magnetograms with a very large dynamic range and very rich image features. In this paper, a multibranch superresolution (MBSR) model is specially designed for solar magnetograms. First, we split a low-resolution magnetogram into a group of overlapping image patches, and classify them into three categories according to magnetic flux intensity, namely simple, medium, and complex. Then, image patches of each category are fed into the corresponding branch of the MBSR network, the lightweight branch for simple image patches and the heavyweight one for complex image patches. The advantage of such a strategy is twofold. On the one hand, active regions are allocated more computational resources to train a heavyweight branch more fully, while quiet regions are allocated fewer computational resources to train a lightweight branch for saving computational resources. On the other hand, a lightweight network with a simple nonlinear function is preferable to simple regions, while a heavyweight one may be underfitting. Additionally, to verify the effectiveness of the proposed model, a magnetic field structure similarity metric is proposed to measure the artifacts of the generated high-resolution (HR) magnetograms. Experimental results show that the proposed MBSR model generates HR magnetograms highly consistent with the HMI ones, and achieves the best performance over five objective metrics, including peak signal-to-noise ratio and structure similarity, etc.https://doi.org/10.3847/1538-4365/ad1760Astronomy image processingAstronomy data modelingSolar physicsAstronomical techniques |
| spellingShingle | Fengping Dou Long Xu Dong Zhao Zhixiang Ren A Multibranch Deep Neural Network for the Superresolution of Solar Magnetograms The Astrophysical Journal Supplement Series Astronomy image processing Astronomy data modeling Solar physics Astronomical techniques |
| title | A Multibranch Deep Neural Network for the Superresolution of Solar Magnetograms |
| title_full | A Multibranch Deep Neural Network for the Superresolution of Solar Magnetograms |
| title_fullStr | A Multibranch Deep Neural Network for the Superresolution of Solar Magnetograms |
| title_full_unstemmed | A Multibranch Deep Neural Network for the Superresolution of Solar Magnetograms |
| title_short | A Multibranch Deep Neural Network for the Superresolution of Solar Magnetograms |
| title_sort | multibranch deep neural network for the superresolution of solar magnetograms |
| topic | Astronomy image processing Astronomy data modeling Solar physics Astronomical techniques |
| url | https://doi.org/10.3847/1538-4365/ad1760 |
| work_keys_str_mv | AT fengpingdou amultibranchdeepneuralnetworkforthesuperresolutionofsolarmagnetograms AT longxu amultibranchdeepneuralnetworkforthesuperresolutionofsolarmagnetograms AT dongzhao amultibranchdeepneuralnetworkforthesuperresolutionofsolarmagnetograms AT zhixiangren amultibranchdeepneuralnetworkforthesuperresolutionofsolarmagnetograms AT fengpingdou multibranchdeepneuralnetworkforthesuperresolutionofsolarmagnetograms AT longxu multibranchdeepneuralnetworkforthesuperresolutionofsolarmagnetograms AT dongzhao multibranchdeepneuralnetworkforthesuperresolutionofsolarmagnetograms AT zhixiangren multibranchdeepneuralnetworkforthesuperresolutionofsolarmagnetograms |