Coherent Processing and Superresolution Technique of Multi-Band Radar Data Based on Fast Sparse Bayesian Learning Algorithm

The coherent processing and superresolution of multi-band radar data from multiple spatially collocated radars is addressed by utilizing a sparse representation technique in this paper. Firstly a parametric model based on geometrical theory of diffraction (GTD) is adopted to construct a redundant dictionary and provide a good match to the scattering mechanism. Then weights of dictionary atoms are computed by using a fast sparse Bayesian learning algorithm. Meanwhile, a multilevel dynamic dictionary is proposed to improve the accuracy and avoid a vast number of computations due to a large dictionary matrix. The final weights and dictionary atoms are adopted to interpolate between and extrapolate outside of the measurement subbands. Furthermore, a procedure for coherent processing is presented to compensate for the lack of mutual coherence among multiband data from multiple spatially collocated radars. The capability and robustness of the proposed method are validated by applying it to the analytical, simulated and static-range data.