Journal of the Korea Institute of Military Science and Technology (한국군사과학기술학회지)

The Korea Institute of Military Science and Technology (한국군사과학기술학회)
권호 표지
DOI QR코드

DOI QR Code

Validation of Map Production through Deep Learning-based Satellite Image Super-Resolution

딥러닝 기반 위성영상 초해상화를 통한 지도 제작 유효성 검증

  • 박재일 (한화시스템(주) SW팀(미래기술)) ;
  • 신상헌 (한화시스템(주) SW팀(미래기술)) ;
  • 곽송연 (한화시스템(주) SW팀(미래기술)) ;
  • 장영찬 (한화시스템(주) SW팀(미래기술)) ;
  • 박진혁 (로딕스)
  • Received : 2024.10.15
  • Accepted : 2025.01.06
  • Published : 2025.04.05
Download PDF
⟨ Previous Next ⟩

Abstract

In recent years, significant efforts have been made to address the cost and time challenges associated with satellite-based remote sensing in both civilian and defense sectors by using deep learning-based super-resolution models to enhance satellite imagery. This paper designs and trains a deep learning-based super-resolution model for satellite images. Utilizing high-resolution satellite images generated from the super-resolution model, we produced maps for RPC error correction, point cloud generation, and DSM creation. We validated the effectiveness of these maps by comparing them with maps produced from original satellite images for both outcome and accuracy. Significant results were achieved in RPC error correction, point cloud generation, and DSM creation. Despite increasing the resolution with the super-resolution model, accuracy was either improved or maintained, confirming its validity for map production.

Keywords

Acknowledgement

이 논문은 2024년 정부(방위사업청)의 재원으로 국방기술진흥연구소의 지원을 받아 수행된 연구임(KRIT-CT-21-040).

References

  1. Park, S. C., Park, M. K., & Kang, M. G., "Superresolution image reconstruction: A technical overview," IEEE Signal Processing Magazine, 20(3), pp. 21-36, 2003. https://doi.org/10.1109/MSP.2003.1203207
  2. Shermeyer, J., & Van Etten, A., "The effects of super-resolution on object detection performance in satellite imagery," In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, pp. 1862-1867, 2019.
  3. Rabbi, J., et al., "Small-object detection in remote sensing images with end-to-end edge-enhanced GAN and object detector network," Remote Sensing, 12(9), p. 1432, https://doi.org/10.3390/rs12091432
  4. Kang, J., Lee, Y.-W., & Kim, D., "A comparative study of deep learning-based super-resolution techniques on Sentinel-2 and CAS500-1 satellites," Journal of the Korean Geographical Society, 57(4), pp. 541-555, 2023. https://doi.org/10.22905/kaopqj.2023.57.4.13
  5. Kim, J., Lee, J. K., & Lee, K. M., "Accurate image super-resolution using very deep convolutional networks," In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 1646-1654, 2016.
  6. Shi, W., et al., "Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network," In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 1874-1883, 2016.
  7. Wang, Z., Chen, J., & Hoi, S. C. H., "Deep learning for image super-resolution: A survey," IEEE Transactions on Pattern Analysis and Machine Intelligence, 43(10), pp. 3365-3387, https://doi.org/10.1109/TPAMI.2020.2982166
  8. Zhao, H., et al., "Loss functions for image restoration with neural networks," IEEE Transactions on Computational Imaging, 3(1), pp. 47-57, 2016. https://doi.org/10.1109/TCI.2016.2644865
  9. Wang, Z., Simoncelli, E. P., & Bovik, A. C., "Multiscale structural similarity for image quality assessment," In The Thirty-Seventh Asilomar Conference on Signals, Systems & Computers IEEE, pp. 1398-1402, 2003.
  10. Deng, L.-J., et al., "Detail injection-based deep convolutional neural networks for pansharpening," IEEE Transactions on Geoscience and Remote Sensing, 59(8), pp. 6995-7010, https://doi.org/10.1109/TGRS.2020.3031366
  11. Masi, G., et al., "Pansharpening by convolutional neural networks,” Remote Sensing, 8(7), p. 594, 2016. https://doi.org/10.3390/rs8070594
  12. Sheikh, H. R., & Bovik, A. C., "Image information and visual quality," IEEE Transactions on Image Processing, 15(2), pp. 430-444, 2006. https://doi.org/10.1109/TIP.2005.859378
  13. Hirschmuller, H., & Scharstein, D., "Evaluation of stereo matching costs on images with radiometric differences," IEEE Transactions on Pattern Analysis and Machine Intelligence, 31(9), pp. 1582-1599, 2009. https://doi.org/10.1109/TPAMI.2008.221
  14. Grodecki, J., & Dial, G., "Block adjustment of high-resolution satellite images described by rational polynomials," Photogrammetric Engineering & Remote Sensing, 69(1), pp. 59-68, 2003. https://doi.org/10.14358/PERS.69.1.59
  15. Oh, J. H., Seo, D. C., & Lee, C. N., "A study on DEM generation from Kompsat-3 stereo images," Journal of the Korean Society of Surveying, Geodesy,  Photogrammetry and Cartography, 32(1), pp. 19-27, 2014. https://doi.org/10.7848/KSGPC.2014.32.1.19
  16. Oh, K.-Y., et al., "Comparison and analysis of matching DEM using KOMPSAT-3 in/cross-track stereo pair," Korean Journal of Remote Sensing, 34(6_3), pp. 1445-1456, 2018. https://doi.org/10.7780/KJRS.2018.34.6.3.10
  17. Kim, S., et al., "Estimation of flooded area using satellite imagery and DSM Terrain data," Journal of the Korean Society of Hazard Mitigation, 19(7), pp. 471-483, 2019. https://doi.org/10.9798/KOSHAM.2019.19.7.471
  18. Müller, M. U., Ekhtiari, N., Almeida, R. M., & Rieke, C., "Super-resolution of multispectral satellite images using convolutional neural networks," ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 1, pp. 33-40), 2020. https://doi.org/10.5194/isprs-annals-V-1-2020-33-2020