Efficient grouping technique for low mutual-coherence basis matrix in radar imaging

Sparse reconstruction has been recently proposed to overcome many challenges in radar imaging applications. In sparse reconstruction framework, the basis matrix of high-resolution radar image demonstrates high mutual coherence, which is inversely proportional to the quality of the reconstructed image. The high mutual coherence, also, weakens the noise immunity in the sparse reconstruction. A technique for building a basis matrix with low mutual coherence is proposed which preserves the high quality of the resultant radar image. The proposed technique groups the columns of the original basis matrix that share the same range of time delay into one column with the average time delay of the group. The resultant basis matrix is used in a sparse reconstruction algorithm to reconstruct the pixel values of a radar image. The technique was validated over realistic and simulated radar measurements and was shown to be computationally efficient with improved noise immunity.