Feature Extraction and Classification of Radar Targets Based on High Resolution Range Profiles

High resolution range profile contains important structural features of the target. It is often used to classify radar targets. Hence, the feature extraction based on high resolution range profile is important. The neural network is often used as the classifier to recognize radar targets based on the features of high resolution range profile. In this paper, a set of features suitable for High Resolution Range Profile recognition are proposed, which have low dimension and good real-time performance without translation sensitivity and amplitude sensitivity. Combined with the signatures, a neural network for target recognition was generated based on the ground target simulation data. Then the adaptability of the network to signal-to-noise radio and elevation angles were discussed. Finally, the filed data were used to verify the universality of the proposed method.