A TCN-BiLSTM-Based Recognition Algorithm for Low, Small, and Slow Targets

The complexity of the motion patterns and the high maneuverability of low, small, and slow targets pose significant challenges for traditional target recognition methods. Conventional approaches often have the limitation of in-depth exploitation of critical features in radar data, resulting in suboptimal recognition performance under low signal-to-noise ratios and in complex environments. To overcome these limitations, this study introduces four types of extended features during the data preprocessing stage. Based on this comprehensive feature construction, a TCN-BiLSTM-based method for recognizing low, small, and slow targets is proposed. The proposed method significantly improves target recognition accuracy and recall performance in cluttered and noisy operational environments. Experimental results demonstrate that the proposed TCN-BiLSTM method achieves outstanding performance in both target recognition accuracy and recall performance, attaining 97% in each metric, which significantly surpasses conventional algorithms. providing a reliable technical solution for the efficient recognition of low, small, and slow targets in radar detection scenarios.