A Novel Method for Sequential HRRPs Recognition based on Data Estimation

The High-Resolution Range Profile (HRRP) contains a lot of radar target feature information and is widely used in in the field of Radar Automatic Target Recognition (RATR). However, the information contained in a single image is limited, while the spatiotemporal correlation between adjacent HRRP has not been applied to recognition, which makes a poor performance of traditional recognition methods with low SNR. In this paper, we proposed a novel method based on Recurrent Temporal Restricted Boltzmann Machine (RTRBM) to replaces the single HRRP with HRRP sequence, and the recognition process can be divided into 3 steps: estimating the missing samples, training the RTRBM model and recognizing the new samples. Experiment results on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset show that our proposed model outperforms other traditional methods, indicating that our method is suitable for incomplete data and noise corrupted data.