Aerial Targets Clustering with Unknown Cluster Number

摘要

Aiming at the uncertainty of enemy target cluster number in aerial targets clustering process, an improved affinity propagation (AP) algorithm is adopted to determine the target cluster number and clustering result. Firstly, Between-Within Propagation (BWP) index is used as the measure of cluster effectiveness, and Bisection method is used to determine the optimal number of clusters;. Secondly, scanning of preferences subspace corresponding to the optimal cluster number, and determine the clustering result. Finally, the effectiveness and stability of the improved affinity propagation algorithm are verified by experiments.