Multi-camera Multi-drone Detection, Tracking and Localization with Trajectory-based Re-identification

Abstract

This paper presents a real-time multiple camera system for detecting, tracking and localizing multiple moving drones simultaneously in a 3 dimension space. The distinct feature of the system is in its target re-identification process, which provides for information fusion among cameras based on the targets' trajectories and relative locations. Drones are detected by the multiple camera system based on motion-based blob detection, and the 2D locations of each drone in individual camera frames are tracked by A geometry- and camera-based model. From the paths of the tracked drones, their trajectories are examined using drone track feature variable. Cross-correlated among cameras for object re-identifications will allow the individual 2D position information to be integrated into overall global 3D positions of all the tracked drones from all cameras. Preliminary outdoor flight demonstrations with 2 drones flying in formation and using 3 cameras show optimal results. The system is able to detect, track, localize and re-identifying individual drone with average positional error of 8%.