Neural technological approach to automated recognition of ground objects in images of space systems for remote sensing of the Earth

In this paper the ability to use neurocomputer technology to imaging remote sensing system are considered. The modeling of object classification is reviewed. The sequence and contents of the main stages of the construction of the neural network architecture are discussed. The spectral characteristics of different ground objects are used for classification and recognition of objects in satellite images. Comparative analysis of different types of neural networks in the classification of ground objects is given.