Neural network approach for multispectral image processing

The subject of the study in the paper is the neural network method of object recognition on multispectral Earth remote sensing. The analysis of the problem of methods and algorithms for processing Earth remote sensing multispectral data revealed the most promising to use agile algorithms that are accommodating to the object search monitoring conditions changes. One of the promising technologies for implementing such algorithms is using neural networks. The choice of convolutional neural networks for recognition is based on capacity of these networks (under the stipulation that the neural network trained correctly) to recognize objects in difficult monitoring conditions and when searching object is deformed. The neural network approach is offered for processing aerospace multispectral images. In the present paper will be described the network building algorithm, chosen practical field for proposed method application, and showed the results of its programming implementation. The obtained results made it possible to conclude that the proposed algorithm is working