Low false alarm radar detection for low altitude weak target based on multi-dimensional clustering extended features

Low-altitude flying objects, such as small rotary-wing drones and birds, pose significant safety risks to both national defense and civil aviation. Due to its all-weather and all-day capabilities, radar has become a crucial tool for detecting low-altitude flying targets, enabling critical tasks such as drone early warning and airport bird monitoring. However, the detection challenges posed by diverse and intense ground clutter often result in the loss of target location information, making it difficult for airports and military forces to take appropriate actions. Traditional algorithms are prone to high miss detection rates, making low-altitude radar detection a persistent challenge. Therefore, this paper first statistically analyzes the amplitude distribution of clutter using a high-resolution Ku-band bio-radar dataset, which includes various rotary-wing drones, aerial creatures, and multiple types of clutter, and identifies the most fitting probability distribution. Subsequently, an extended target detector based on amplitude priors and DBSCAN clustering is designed to achieve low false alarm radar detection of drones and aerial creatures under strong low-altitude clutter conditions. The results from processing real-world data demonstrate that the proposed algorithm not only successfully detects low-altitude targets but also reduces the false alarm rate to one percent of that produced by traditional methods.