An Extraction Method Based on Kohonen Neural Network Used in Radar

The paper focuses on the research and application of neural network for feature extraction from the three dimensional radar echo data, proposed a method of combining the median filter and Kohonen neural network to resolve it. Simulation results show that the proposed method in the paper is feasible, the results is more stable and more accurate, apparently has an advantage over the ordinary self-organization neural network. To validate this method by applied to two groups of real radar data set. Result show that the proposed method is practical with ninety percent accuracy.