Aerial Target Grouping based on Sparse Subspace Clustering with the Mixed constraints of Distance and Velocity

Aerial target grouping, partitioning targets into some clusters according to the features of attribute and behavior, contributes to tactical recognition and command decision-making. For the scenario of high-dimensional data in aerial target grouping, sparse subspace clustering is expected to be one of the effective methods, where high-dimensional data are represented by low dimensional subspaces using sparsity. In order to achieve aerial target grouping by sparse subspace clustering efficiently and accurately, an improved sparse subspace clustering with the mixed constraints of distance and velocity (SSC_MC) is proposed. First, k nearest neighbours algorithm is used to filter neighbours, so as to improve the self-represention by the whole data set into only by their nearest neighbours; next, the mixed constraints of distance and velocity are introduced in sparse subspace clustering, which emphasizes the importance of these features in aerial target grouping; last, the experiments for aerial target grouping are implemented, showing the proposed algorithm SSC_MC can be used to aerial target grouping with high efficiency and accuracy.