Directional Object Tracking Based on UAV Vision

Aiming at the problems of poor anti-interference ability and low tracking accuracy caused by excessive background information in the object bounding box of unmanned aerial vehicle(UAV) object tracking, this paper proposes a directional tracking system based on UAV vision to solve the issues of occlusion and adaptive tracking of directional objects. The article first constructs a tracking framework based on object motion characteristics to form a robust tracking system. Secondly, the article uses the moving horizon estimation(MHE) to model the object motion, extract the motion features, and establish the optimization function of the directional bounding box to perform adaptive tracking. Finally, a UAV platform is built for object tracking experiments, and the algorithm has excellent performance on the collected data sets.