Asosiy kontentga oʻtish
Boshqa

Modern Approaches to Atmospheric Correction of Satellite Data: Review and Perspectives

Karimov Sardor Ilkhom UgliFergana State Technical University,Department of Computer Engineering and Artificial Intelligence,Fergana,UzbekistanFazlitdinov Muhammadali Xatamjon o’g’liFerghana Polytechnical InstituteAbdullayeva Mohigul Rahimjon QiziFergana State Technical University,Department of Computer Engineering and Artificial Intelligence,Fergana,Uzbekistan
2025

Ochiq kirish: yoʻq

ABI

Annotatsiya

in this article, methods and algorithms of atmospheric correction of satellite images, aimed at improving the accuracy of Earth remote sensing, are considered. Particular attention is given to the comparative analysis of traditional correction methods and modern approaches based on convolutional neural networks (encoder–decoder architecture). The proposed method automates the process of atmospheric correction of “raw” satellite data, ensuring high accuracy while minimizing the need for interactive processing of large volumes of imagery. Based on Sentinel-2 satellite data, a training dataset was formed, including images of levels L1C and L2A. Experimental results demonstrated that the developed neural network achieves high accuracy in atmospheric correction and can be applied not only to Sentinel-2 data but also to images from other satellites. In addition, the use of spectral indices (NDVI, SAVI, EVI, NDWI, NDSI) was examined for the analysis of agricultural lands in Uzbekistan. The results obtained confirm the effectiveness of the proposed approach for land resource monitoring and agricultural production management tasks.

Hali tarjima qilinmagan

Identifikatorlar

Iqtiboslar va manbalar