Deep convolutional neural networks in hyperspectral remote sensing data processing

During the last decade the deep convolutional neural networks (DСNN) were successfully applied in the fields related to processing of large satellite images of high resolution that are used in various inverse problems on retrieval of the earth atmosphere characteristics and the earth boundary reflectance via remote sensing data analysis.
The presented paper contains the information on the research state related to application of neural network methods to satellite hyper-spectral image processing, including brief information on the main features of convolutional neural networks (CNN), deep learning (DL) and autoencoders (AE) that are used for information compression.
Up to present time a considerable number of DСNN models created is located for open access in the Internet. These verified models with well performance allow to develop new advanced models of DСNN. A brief information on some Internet models of open access is contained in the present paper. A more detailed information on neural network models located in open Internet access, and also on large data sets that are necessary for DСNN tuning, will be contained in the second part of the present paper, that is planned to be published.