Enhanced Deep Super-Resolution Model Parameter Optimization to Improve the Quality of Geostationary Meteorological Satellite Images

Harimino Andriamalala RAJAONARISOA

¹Dynamic Atmosphere, Climate, and Ocean Laboratory, Physics and Applications, Sciences and Technologies, University of Antananarivo, Madagascar

Email: [email protected]

²Dynamic Atmosphere, Climate, and Ocean Laboratory, Physics and Applications, Sciences and Technologies, University of Antananarivo, Madagascar

Email: [email protected]

*Corresponding author

The aim of this study was to determine the parameters of the EDSR (Enhanced Deep Super-Resolution) model that best improve the quality of images from geostationary meteorological satellites. The datasets used were composed of one hundred pairs of images from geostationary meteorological satellites. Each pair was made up of a low resolution image and a corresponding high resolution image all coming from the same meteorological satellite. The training (respectively the test) of the EDSR model was done with 80% (respectively 20%) of all the datasets. Only the “batch size” and “epoch” parameters of the EDSR model were considered. The other EDSR model parameters were the default parameters used by the EDSR model designer. After training and testing the model using specifically selected datasets, the image generated by the trained EDSR model was compared with the corresponding high-resolution image using the PSNR (Peak Signal-to-Noise Ratio) metric. A high PSNR value indicates a strong resemblance between two images. The “batch size” and “epoch” values corresponding to the best image from the PSNR evolution curve were selected.

Artificial intelligence, Batch size, Epoch, EDSR, PSNR, Super-Resolution

Harimino Andriamalala RAJAONARISOA, Adolphe Andriamanga RATIARISON (2024). Enhanced Deep Super-Resolution Model Parameter Optimization to Improve the Quality of Geostationary Meteorological Satellite Images. Journal of Artificial Intelligence and Systems, 6, 1–10. https://doi.org/10.33969/AIS.2024060101.

[1] M. S. Nadler and J. M. Mendel. (1981). Image sequence enhancement by digital signal processing, Proceedings of the IEEE.

[2] Dong, C., Loy, C. C., He, K., & Tang, X. (2015). Image super-resolution using deep convolutional networks. IEEE transactions on pattern analysis and machine intelligence, 38(2), 295-307.

[3] Shi, W., Caballero, J., Huszár, F., Totz, J., Aitken, A. P., Bishop, R., & Wang, Z. (2016). 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).

[4] Kim, J., Lee, J. K., & Lee, K. M. (2016). Accurate image super-resolution using very deep convolutional networks. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 1646-1654).

[5] Ledig, C., Theis, L., Huszár, F., Caballero, J., Cunningham, A., Acosta, A., & Shi, W. (2017). Photo-realistic single image super-resolution using a generative adversarial network. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 4681-4690).

[6] Lim, B., Son, S., Kim, H., Nah, S., & Mu Lee, K. (2017). Enhanced deep residual networks for single image super-resolution. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops (pp. 136-144).

[7] EUMETSAT, Meteosat 0 degree Natural Colour Full disc low resolution. https://eumetview.eumetsat.int/static-images/MSG/RGB/NATURALCOLOR/FULLDISC/ 

[8] EUMETSAT, Meteosat 0 degree Natural Colour Full disc high resolution https://eumetview.eumetsat.int/static-images/MSG/RGB/NATURALCOLOR/FULLRESOLUTION/  

[9]Harimino Andriamalala RAJAONARISOA, S.RAKOTONASY, F. RANDRIAMAHALEO, A. A.RATIARISON (2023). Super Resolution Generative Adversarial Network Model parameter optimization to improve the quality of geostationary meteorological satellite images. Journal of Artificial Intelligence and Systems, 5, 115–124. https://doi.org/10.33969/AIS.2023050108.

[10] Alubinskas, G. (2017). Image similarity/distance measures: what is really behind MSE and SSIM?. International Journal of Image and Data Fusion, 8(1), 32-53.

[11] Hore, A., & Ziou, D. (2010, August). Image quality metrics: PSNR vs. SSIM. In 2010 20th international conference on pattern recognition (pp. 2366-2369). IEEE.