Method of Classification of Fixed Ground Objects by Radar Images with the Use of Artificial Neural Networks

The article considers the method of classification for ground stationary objects. As the source of data are used radar pictures of the land, which were received using the air of radio-electronic monitoring systems in the synthetic-aperture radar (SAR) mode. For the discovered ground objects, estimates of their characteristics are evaluated with the mutual orientation, geometric features of the location taken into account. The resulting data set allows creating a training sample, which is used to build an Artificial Neural Network. As a result, the Artificial Neural Network is able to classify the detected groups of objects with a given probability. In the process of modeling, the method used software module Image Processing Toolbox Matlab 2016, which allows evaluating the raster images of radar portraits. Software Neural Network Toolbox Matlab was used to form an artificial neural network. The obtained simulation results showed the possibility of application and using the technique in air radio electronic systems of monitoring the ground situation.