Residual LSTM Radar Target Detection Method Based on Global Channel Attention Mechanism

In radar target detection, large targets often mask small targets, and the traditional constant false alarm (CFAR) detection method is difficult to set a suitable threshold and cannot cope with the detection of this type of target well. Aiming at this kind of problem, a residual LSTM detection method based on global channel attention mechanism (Global Channel Attention Mechanism- Long Short Term Memory, GCAMLSTM) is proposed. This method processes the radar echo data, establishes the time-order relationship of the radar signals through the sliding window method, and proposes a new Global Channel Attention Mechanism module according to the characteristics of the time-order signals, so as to extract the target echo features more accurately. The experimental results show that the detection performance of the proposed method is $\mathbf{6. 8 \%}$ higher than that of the traditional CFAR algorithm. It has strong robustness in the above scenario and can achieve effective detection of radar targets.