Radar HRRP Target Recognition with Target Aware Two-Dimensional Recurrent Neural Network

In this paper, a Target Aware Two-Dimensional Recurrent Neural Network(TATDRNN) is proposed for High-Resolution Range Profile(HRRP) based Radar Automatic Target Recognition(RATR), which characterizes the temporal dependence across the range cells within the single HRRP. Specifically, in our proposed TATDRNN, a Two-Dimensional RNN is introduced to model target region and noise region of HRRP data separately and a beyesian generative model is defined to separate target region and noise region according to their distribution character. Experimental results based on measured data demonstrate that our proposed TATDRNN achieves competitive recognition performance compared with traditional methods.