A Low-Altitude Small Target Radar Recognition Algorithm Based on Micro-Doppler Features and Vision Transformer Network

In low-altitude airspace surveillance scenarios, accurately distinguishing between birds and unmanned aerial vehicles (UAVs) is of vital importance. However, the high similarity in flight altitude and speed between birds and UAVs makes it extremely difficult to distinguish between these two types of targets. To enhance radar recognition performance for both UAVs and birds, this paper proposes a low-altitude target recognition algorithm that combines micro-Doppler features with a Vision Transformer network (ViT) under a staring radar system. Verified with real-measured data, the recognition accuracy for three types of UAVs and birds reaches 94%.