An Improved Particle Filter Algorithm for Automated Aerial Refueling Based on Meanshift Clustering

Traditional particle filter algorithm requires a number of particle samples for the accuracy of tracking with high computational complexity, which does not satisfy the real-time requirements. In addition, the target is easily to lose under the occlusion condition when the particle tracing algorithm is applied. A target tracking algorithm of anti-blocking particle filter is proposed in this paper. On the basis of clustering, a target probability density estimation method is proposed to describe the target model, and the target can be tracked with less particles. Kalman filter is chosen to determine the target location when the target is blocked seriously. Experiments show that when the algorithm is employed the robustness and tracking accuracy are much better and the real-time requirement is met.