Research on UAV Fusion Tracking and Identification Technology in Complex Environment

Abstract

Widespread application of unmanned aerial vehicles (UAVs) has brought new military threats. However, the stable tracking, classification and identification of UAV targets in complex environments restricts the overall improvement of the scale application and capabilities of anti-UAV systems. In response to actual application requirements, this paper first proposed the process of optoelectronic and radar coordinated detection. Then the probabilistic data association filtering (PDAF) algorithm was applied in radar and optoelectronic fusion tracking, which not only made the system suitable in environments where the false detection probability changed sharply in the surveillance area, but also improved the accuracy and data rate of optoelectronic and radar coordinated tracking, provided precise height information, and laid a good foundation for target classification and identification. Finally, the paper proposed multi-cascade feature classification algorithm with low miss rate based on deep forest, which not only had a low computational complexity, a small amount of training data, and fewer hyper-parameters, but also was suitable for decision-making in the parameter domain. Experimental results showed that the algorithm had high accurate, required a small amount of model parameters, and had some robustness in actual project deployment.