A Comparative Study of Deep Learning-Based Super-Resolution Techniques on Sentinel-2 and CAS500-1 Satellites

Jonggu Kang; Yang-Won Lee; Daesun Kim

doi:10.22905/kaopqj.2023.57.4.13

All Issue

2023 Vol.57, Issue 4 Preview Page Next Page

Research Article

A Comparative Study of Deep Learning-Based Super-Resolution Techniques on Sentinel-2 and CAS500-1 Satellites Sentinel-2 위성과 국토위성에서의 딥러닝 기반 초해상화 기법 비교 연구

31 December 2023. pp. 541-555

PDF XML

Abstract

Satellite images are used in a variety of fields, but the limited spatial resolution can make it difficult to identify fine details. Deep learning-based super-resolution models are attracting attention as a method for effectively converting low-resolution images to high-resolution images to obtain clearer images. In this study, we evaluated the performance of deep learning-based super-resolution models for Sentinel-2 and CAS500=1 satellite images. The results showed that deep learning-based models outperformed traditional interpolation methods, especially for CAS500-1 satellite images. However, the results showed that the super-resolution process can distort small objects. This can be improved in future studies by applying state-of-the-art (SOTA) models or developing super-resolution techniques that consider the size of objects. The results of this study confirm the effectiveness of deep learning-based super-resolution models for satellite images and are expected to be used in a variety of fields.

Keywords

Satellite image

Deep learning

Super-resolution

위성영상 자료는 다양한 분야에서 활용되고 있으나 광범위한 관측범위를 가지는 위성센서들의 공간해상도로는 정밀한 객체식별에 한계가 있어 고해상도화 방안이 필요하다. 딥러닝 기반 초해상화 모델은 저해상도 영상을 효과적으로 고해상도로 변환하여 보다 선명한 영상을 얻을 수 있는 방법으로 주목받고 있다. 이에 본 연구에서는 Sentinel-2 위성 영상과 국토위성 영상을 대상으로 딥러닝 기반 초해상화 모델의 성능을 평가하였다. 평가 결과, 딥러닝 기반 모델은 기존의 보간법에 비해 우수한 성능을 보였으며, 특히 국토위성 영상에서 더 효과적인 것으로 나타났다. 하지만, 객체가 공간해상도에 비해 작은 경우 초해상화 과정에서 왜곡되는 결과를 보였다. 이는 향후 연구에서 최신(State-of-the-Art, SOTA)모델의 적용하거나, 객체의 크기를 고려한 초해상화 기법을 개발하는 등의 방법으로 개선될 수 있을 것이다. 본 연구 결과는 위성영상에서 딥러닝 기반 초해상화 모델의 효과를 확인한 것으로, 다양한 분야에서 활용될 수 있을 것으로 기대된다.

키워드

위성영상

딥러닝

초해상화

References

구자용, 2011, "해상도별 위성영상을 이용한 체계적 토지피복 분류에 관한 연구," 국토지리학회지 45(3): 375-386.

국토위성센터, 2021, 국토위성센터 연구보고서, 국토지리정보원.

김진광･이덕규･고형호, 2022, "차세대중형위성 1호 운영 컨셉 및 운영 현황," 한국항공우주학회 학술발표회 초록집, 686-687.

최소연･박강현･강종구･김근아･윤유정･이양원, 2022, "U-Net과 HRNetV2 모델을 이용한 Sentinel-1 SAR 영상의 홍수탐지," 국토지리학회지 56(4): 409-419. 10.22905/kaopqj.2022.56.4.8

Benediktsson, J. A., J. Chanussot, and W. M. Moon, 2012, Very high-resolution remote sensing: Challenges and opportunities [point of view], Proceedings of the IEEE 100(6): 1907-1910. 10.1109/JPROC.2012.2190811

European Space Agency (ESA), n.d., Sentinel Online, SENTINEL-2 MISSION GUIDE, https://sentinel.esa.int/web/sentinel/home.

Ledig, C., L. Theis, F. Huszár, J. Caballero, A. Cunningham, A. Acosta, and W. Shi, 2017, Photo-realistic single image super-resolution using a generative adversarial network, In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition 4681-469. 10.1109/CVPR.2017.19

Lim, B., S. Son, H. Kim, S. Nah, and K. Mu Lee, 2017, Enhanced deep residual networks for single image super-resolution, In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops 136-144. 10.1109/CVPRW.2017.151

Merino, M. T. and J. Nunez, 2007, Super-resolution of remotely sensed images with variable-pixel linear reconstruction, IEEE Transactions on Geoscience and Remote Sensing 45(5): 1446-1457. 10.1109/TGRS.2007.893271

Pan, Z., J. Yu, H. Huang, S. Hu, A. Zhang, H. Ma, and W. Sun, 2013, Super-resolution based on compressive sensing and structural self-similarity for remote sensing images, IEEE Transactions on Geoscience and Remote Sensing 51(9): 4864-4876. 10.1109/TGRS.2012.2230270

Park, S. C., M. K. Park, and M. G. Kang, 2003, Superresolution image reconstruction: A technical overview, IEEE Signal Processing Magazine 20(3): 21-36. 10.1109/MSP.2003.1203207

Shen, H., M. K. Ng, P. Li, and L. Zhang, 2009, Super-resolution reconstruction algorithm to MODIS remote sensing images, The Computer Journal 52(1): 90-100. 10.1093/comjnl/bxm028

Wang, X., K. Yu, S. Wu, J. Gu, Y. Liu, C. Dong, and C. Change Loy, 2018, Esrgan: Enhanced super-resolution generative adversarial networks, In Proceedings of the European conference on computer vision (ECCV) workshops. 10.1007/978-3-030-11021-5_5

Wang, Z., A. C. Bovik, H. R. Sheikh, and E. P. Simoncelli, 2004, Image quality assessment: from error visibility to structural similarity, IEEE Transactions on Image Processing 13(4): 600-612. 10.1109/TIP.2003.819861

Xu, W., X. U. Guangluan, Y. Wang, X. Sun, D. Lin, and W. U. Yirong, 2018, High quality remote sensing image super-resolution using deep memory connected network, In IGARSS 2018-2018 IEEE International Geoscience and Remote Sensing Symposium 8889-8892. 10.1109/IGARSS.2018.8518855

Yang, D., Z. Li, Y. Xia, and Z. Chen, 2015, Remote sensing image super-resolution: Challenges and approaches, In 2015 IEEE International Conference on Digital Signal Processing (DSP) 196-200. 10.1109/ICDSP.2015.7251858

Yang, J., J. Wright, T. S. Huang, and Y. Ma, 2010, Image super-resolution via sparse representation, IEEE Transactions on image Processing 19(11): 2861-2873. 10.1109/TIP.2010.205062520483687

Yu, J., Y. Fan, J. Yang, N. Xu, Z. Wang, X. Wang, and T. Huang, 2018, Wide activation for efficient and accurate image super-resolution, arXiv preprint arXiv:1808.08718.

Zhang, H., Z. Yang, L. Zhang, and H. Shen, 2014, Super-resolution reconstruction for multi-angle remote sensing images considering resolution differences, Remote Sensing 6(1): 637-657. 10.3390/rs6010637

Information

Publisher :The Korean Association of Professional Geographers
Publisher(Ko) :국토지리학회
Journal Title :국토지리학회지
Journal Title(Ko) :THE GEOGRAPHICAL JOURNAL OF KOREA
Volume : 57
No :4
Pages :541-555
Received Date : 2023-12-19
Revised Date : 2023-12-27
Accepted Date : 2023-12-27
DOI :https://doi.org/10.22905/kaopqj.2023.57.4.13

국토지리학회지 ISSN:1225-3766(Print) THE GEOGRAPHICAL JOURNAL OF KOREA

All Issue