Mars Image Super-Resolution Based on Generative Adversarial Network
High-resolution (HR) Mars images have great significance for studying the land-form features of Mars and analyzing the climate on Mars. Nowadays, the mainstream image super-resolution methods are based on deep learning or CNNs, which are better than traditional methods. However, these deep learning...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2021-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9503382/ |
| _version_ | 1818455388055404544 |
|---|---|
| author | Cong Wang Yin Zhang Yongqiang Zhang Rui Tian Mingli Ding |
| author_facet | Cong Wang Yin Zhang Yongqiang Zhang Rui Tian Mingli Ding |
| author_sort | Cong Wang |
| collection | DOAJ |
| description | High-resolution (HR) Mars images have great significance for studying the land-form features of Mars and analyzing the climate on Mars. Nowadays, the mainstream image super-resolution methods are based on deep learning or CNNs, which are better than traditional methods. However, these deep learning based methods obtain low-resolution(LR) images usually by using an ideal down-sampling method (<italic>e.g.</italic> bicubic interpolation). There are two limitations in the existing SR methods: 1) The paired LR-HR data by using such methods can achieve a satisfactory results when tested on an ideal datasets. But, these methods always fail in real Mars image super-resolution, since real Mars images rarely obey an ideal down-sampling rule. 2) The LR images obtained by ideal down-sampling methods have no noise while real Mars images usually have noise, which leads to the super-resolved images are not realistic in texture details. To solve the above-mentioned problems, in this article, we propose a novel two-step framework for Mars image super-resolution. Specifically, to address limitation 1), we focus on designing a new degradation framework by estimating blur-kernels. To address limitation 2), a Generative Adversarial Network (GAN) is trained to generate noise distribution. Extensive experiments on the Mars32k dataset demonstrate the effectiveness of the proposed method, and we achieve better qualitative and quantitative results compared to other SOTA methods. |
| first_indexed | 2024-12-14T22:09:59Z |
| format | Article |
| id | doaj.art-757907e885404cd5b155c1fc4d203c03 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-14T22:09:59Z |
| publishDate | 2021-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-757907e885404cd5b155c1fc4d203c032022-12-21T22:45:46ZengIEEEIEEE Access2169-35362021-01-01910888910889810.1109/ACCESS.2021.31018589503382Mars Image Super-Resolution Based on Generative Adversarial NetworkCong Wang0Yin Zhang1Yongqiang Zhang2https://orcid.org/0000-0002-0437-7337Rui Tian3Mingli Ding4School of Instrument Science and Engineering, Harbin Institute of Technology (HIT), Harbin, ChinaSchool of Instrument Science and Engineering, Harbin Institute of Technology (HIT), Harbin, ChinaSchool of Instrument Science and Engineering, Harbin Institute of Technology (HIT), Harbin, ChinaSchool of Instrument Science and Engineering, Harbin Institute of Technology (HIT), Harbin, ChinaSchool of Instrument Science and Engineering, Harbin Institute of Technology (HIT), Harbin, ChinaHigh-resolution (HR) Mars images have great significance for studying the land-form features of Mars and analyzing the climate on Mars. Nowadays, the mainstream image super-resolution methods are based on deep learning or CNNs, which are better than traditional methods. However, these deep learning based methods obtain low-resolution(LR) images usually by using an ideal down-sampling method (<italic>e.g.</italic> bicubic interpolation). There are two limitations in the existing SR methods: 1) The paired LR-HR data by using such methods can achieve a satisfactory results when tested on an ideal datasets. But, these methods always fail in real Mars image super-resolution, since real Mars images rarely obey an ideal down-sampling rule. 2) The LR images obtained by ideal down-sampling methods have no noise while real Mars images usually have noise, which leads to the super-resolved images are not realistic in texture details. To solve the above-mentioned problems, in this article, we propose a novel two-step framework for Mars image super-resolution. Specifically, to address limitation 1), we focus on designing a new degradation framework by estimating blur-kernels. To address limitation 2), a Generative Adversarial Network (GAN) is trained to generate noise distribution. Extensive experiments on the Mars32k dataset demonstrate the effectiveness of the proposed method, and we achieve better qualitative and quantitative results compared to other SOTA methods.https://ieeexplore.ieee.org/document/9503382/Generative adversarial networkkernel estimationmars image super-resolutionnoise model |
| spellingShingle | Cong Wang Yin Zhang Yongqiang Zhang Rui Tian Mingli Ding Mars Image Super-Resolution Based on Generative Adversarial Network IEEE Access Generative adversarial network kernel estimation mars image super-resolution noise model |
| title | Mars Image Super-Resolution Based on Generative Adversarial Network |
| title_full | Mars Image Super-Resolution Based on Generative Adversarial Network |
| title_fullStr | Mars Image Super-Resolution Based on Generative Adversarial Network |
| title_full_unstemmed | Mars Image Super-Resolution Based on Generative Adversarial Network |
| title_short | Mars Image Super-Resolution Based on Generative Adversarial Network |
| title_sort | mars image super resolution based on generative adversarial network |
| topic | Generative adversarial network kernel estimation mars image super-resolution noise model |
| url | https://ieeexplore.ieee.org/document/9503382/ |
| work_keys_str_mv | AT congwang marsimagesuperresolutionbasedongenerativeadversarialnetwork AT yinzhang marsimagesuperresolutionbasedongenerativeadversarialnetwork AT yongqiangzhang marsimagesuperresolutionbasedongenerativeadversarialnetwork AT ruitian marsimagesuperresolutionbasedongenerativeadversarialnetwork AT mingliding marsimagesuperresolutionbasedongenerativeadversarialnetwork |