Radar Target Recognition Based on Dictionary of Time-Frequency Feature and Nonnegative Sparse Decomposition

A new approach for radar HRRP target recognition is presented in this paper which combines the empirical mode decomposition (EMD) method with the nonnegative gradient projection for sparse reconstruction (NGPSR) method. EMD is used to extract the feature vector for dictionary learning and NGPSR is used to reconstruct the HRRP signal. The radar HRRPs are classified according to the sum of the sparse representation coefficients. The experimental results based on simulated radar HRRP targets recognition show that the proposed method can achieve a higher correct recognition rate compared with classical classification methods.