Visual group target tracking algorithm based on MeanShift-PCA-PF

In recent years, target tracking has been widely studied and applied in intelligent video surveillance, human-machine interaction, radar early warning and tracking, and other fields. With the emergence of the new combat concept of unmanned aerial vehicle (UAV) swarm operations, and the maturity of large-area, high-resolution visible light COMS imaging device technology, the imaging detection and visualization of UAV swarm targets in large observation fields will greatly improve the efficiency of command systems. Therefore, the research on the group target detection, tracking, and recognition based on video image is very urgent and crucial. Group targets are collections of densely packed multiple objects with certain group characteristics that meet the given distance criterion. Due to the large number of targets, limited image resolution, and waste of communication resources, multiple target tracking techniques are challenging to implement effectively. In this paper, based on the characteristics of group targets, we propose a group target tracking framework based on optical measurement equipment. By dividing groups according to the density feature of group targets, we obtain the structural properties and spatial geometric features of the group and implement tracking using a particle filter. Finally, we validate the effectiveness of the group target tracking technique through simulation experiments in group target tracking.