Dense Group Target Number Estimation Method Based on Multiple Incoherent Observations

‍ ‍The monitoring of dense group targets， such as bird flocks and drone swarms， is a current research hotspot in radar target monitoring. Accurate estimation of the number of individual targets is a crucial prerequisite for achieving correct association， robust tracking， and situational awareness of group targets. Conventional target number estimation methods based on range profile peak detection are unsuitable when target distributions are dense and the radar range resolution is limited. The methods based on monopulse systems require iterative computation by integrating data from multiple channels， leading to high computational complexity. Although array-based methods reduce computational complexity， they require additional sensor resources. Therefore， to achieve accurate estimation of dense target numbers using a single sensor， this paper proposes a number estimation method for dense group targets based on multiple incoherent observations. First， a multiple observation signal model for a single sensor is constructed， followed by establishing the relationship between the eigenvalue distribution of the observation covariance matrix and target number. Based on the minimum description length criterion， major eigenvalues are segmented to accurately estimate the target number. The effectiveness of the method was verified and its performance was analyzed through simulation and experimental data.