Evaluation of the impact of super-resolution on GEOSAT-2 multispectral images


  • César Fernández GEOSAT Satélites SLU
  • Carolina de Castro GEOSAT Satélites SLU
  • Lucía García GEOSAT Satélites SLU
  • María Elena Calleja GEOSAT Satélites SLU
  • Rubén Niño GEOSAT Satélites SLU
  • Silvia Fraile GEOSAT Satélites SLU
  • Rafael Sousa GEOSAT Satélites SLU




super-resolution, GEOSAT-2, precision agriculture, artificial intelligence


The growing need to observe the Earth in greater detail means the appearance of new techniques to improve the geometric value of images, preserving their radiometric characteristics. Security and Defence sectors are strategic users of these advances, but not the only ones. By being able to preserve the radiometric characteristics of the data, precision agriculture is a key beneficiary of these improvements. In this way, more detailed data and information can be provided on the specific needs of each crop, which means a direct implication for the farmer, the agricultural companies, and the environment. In this work, the Random Forest and XGBoost methods were applied in order to improve the resolution of the GEOSAT-2 images while preserving their radiometric values. In addition, the quality of the enhanced images was evaluated. Also, the satisfactory evaluation of the improved images is presented, both in terms of resolution and the final quality obtained. This evaluation has been conducted by the Copernicus Coordinated data Quality Control (CQC) team, allowing the addition of a new product to the GEOSAT portfolio, ready to be integrated into the Copernicus Programme data offer.


Download data is not yet available.

Author Biographies

César Fernández, GEOSAT Satélites SLU


Carolina de Castro, GEOSAT Satélites SLU


Lucía García, GEOSAT Satélites SLU

Business Development

María Elena Calleja, GEOSAT Satélites SLU

Image Operations

Rubén Niño, GEOSAT Satélites SLU

Satellite Operations

Silvia Fraile, GEOSAT Satélites SLU


Rafael Sousa, GEOSAT Satélites SLU



Gironés, J., Casas, J., Minguillón, J., Caihuelas, R. 2017. Minería de datos - modelos y algoritmos. Barcelona: UOC.

Harrell, F. 2001. Regression Modeling Strategies. Springer Series in Statistics. New York: Springer. https://doi.org/10.1007/978-1-4757-3462-1

Huang, J., Siu, W. 2015. Practical application of random forests for super-resolution imaging. En: IEEE International Symposium on Circuits and Systems. Lisboa, Portugal, 24-27 Mayo. pp 2161-2164. https://doi.org/10.1109/ISCAS.2015.7169108

Lim, B., Son, S., Kim, H., Nah, S., Lee, K.M. 2017. Enhanced Deep Residual Networks for Single Image Super-Resolution. En: IEEE Conference on Computer Vision and Pattern Recognition Workshops. Honolulu, HI, EE.UU., 21-26 Julio. pp 1132-1140. https://doi.org/10.1109/CVPRW.2017.151

PRISM CQC Team. 2022. Copernicus Coordinated data Quality Control - GEOSAT-2 Test Data Set Analysis: Bundle Super Resolution. Copernicus Contributing Missions access support functions and platform. 8 Abril.

Raschka, S. 2016. Python Machine Learning. Birmingham: Packt Publishing.

Schulter, S., Leistner, C., Bischof, H. 2015. Fast and accurate image upscaling with super-resolution forests. En: IEEE Conference on Computer Vision and Pattern Recognition. Boston, MA, EE.UU., 7-12 Junio. pp 3791-3799. https://doi.org/10.1109/CVPR.2015.7299003

Zhang, Y., Li, K., Li, K., Wang, L., Zhong, B., Fu, Y.R., 2018a. Image Super-Resolution Using Very Deep Residual Channel Attention Networks. En: European Conference on Computer Vision. Munich, Alemania, 8-14 Septiembre. pp 294-310. https://doi.org/10.1007/978-3-030-01234-2_18

Zhang, Y., Tian, Y., Kong, Y., Zhong, B., Fu, Y.R., 2018b. Residual Dense Network for Image Super-Resolution. En: IEEE/CVF Conference on Computer Vision and Pattern Recognition. Salt Lake City, UT, EE.UU., 18-23 Junio. pp 2472-2481. https://doi.org/10.1109/CVPR.2018.00262





Practical cases