Radar Missile Micromotion Target Open-set Identification Method Based on RTF-Net

As using the radar to detect and identify the missile micromotion targets in the space, identification can be made using time-frequency images obtained from narrowband waveforms. It is important to note that the process of identifying missile micromotion targets in space often involves the presence of numerous unknown targets, which can potentially impact the accuracy of identifying known targets. To address these issues, this paper proposes a radar missile micromotion target open-set recognition method based on the reconstructed time-frequency image network(RTF-Net). In this paper, RTF-Net is designed based on the convolutional neural network (CNN) model idea for the time-frequency image recognition and input reconstruction of the narrowband echo. Firstly, the encoder constructed with the CNN model, is employed to train the known target and save the potential feature vectors during the training. Subsequently, the decoder uses these potential feature vectors to achieve the recognition of the unknown target. The experimental results demonstrate that the proposed method can achieve the superior recognition effect on the unknown class of targets while maintaining the recognition accuracy of the known class at a comparable level.