Optimization of classification of SAR radar ground targets by combining convolutional network and genetic algorithm used in command and control system

In command and control, it is necessary to provide clean data to the system for better decision-making in the battle scene. In this regard, artificial aperture radar is an imaging radar with high resolution, which needs to improve the quality and classification of these images to recognize the details of the scene. The presence of speckle noise is the most important factor in image quality degradation, and it is necessary to reduce the effect of speckle noise in the pre-processing stage. Also, one of the important methods in interpreting SAR radar images is the classification of images, which is very useful in examining the changes in ground targets, because the observation and monitoring of targets in command and control is considered a good tool. Determining a classification method with appropriate accuracy for SAR radar images with high spatial resolution is a goal in this article, as the old algorithms in this field such as Kmeans, Perceptron, SVM, and MLP have poor detection power and inappropriate speed. As a result, in this article, the aim is to present a powerful algorithm for optimizing the classification of SAR radar ground targets with the help of the proposed method of combining convolutional network and genetic algorithm, which can be obtained through airplanes, spacecraft or satellites to observe ground targets. has been In this article, in the pre-processing stage, after reducing the effect of speckle noise on SAR radar images with the help of the Lee filter, the optimization of the classification of SAR radar ground targets with the proposed method has been obtained, and acceptable results have been obtained. The proposed network has performed 99.33% better than other methods in terms of classification accuracy of denoised MSTAR images.