HRR radar imaging based on compressed samples using dynamic dictionaries

Compressed sensing (CS) theory suggests that accurate reconstruction of an signal can be achieved using its highly undersampled measurements, provided that the signal is sparse in an a priori known dictionary. According to the mathematical structure of the received echo, this dictionary is typically taken to be a DFT basis for the range imaging problem in wideband radar. However, since practical target scatterers do not lie exactly at spatial positions corresponding to frequency gridding points of the DFT basis, there is always a mismatch between the assumed DFT basis and the actual dictionary in which the radar echo is sparse. So the performance of classical CS reconstruction methods degrades considerably when applied to the radar imaging problem. We consider the dictionary as adjustable instead of fixed, and develop a compressive imaging approach using sparse reconstruction with dynamic dictionaries. The approach iteratively estimates sparse scattering coefficients given the dictionary and then updates the dictionary by EM algorithm to achieve better signal model fit. Through this approach, we achieve the high-quality range profiles of radar targets for both simulated data and measured data in an anechoic chamber.