A novel Bayesian compressive sensing method for complex variables in SAR imaging

The complex multitask Bayesian compressive sensing (CMT-BCS) is well adaptive for sparse complex signal in low signal-to-noise ratio (SNR). However, the dimensions of measurement vector and dictionary matrix are double its original one with more iteration time. In this paper, we extend the Bayesian compressive sensing (BCS) reconstruction method to real and imaginary components with the phase estimation of scene in advance, and then combine the results of both real and imaginary components to optimize the final SAR image. Both the simulations and real experiments show that the proposed method can deal with complex sampling model effectively with the improvement of reconstruction precision. Comparing to the CMT-BCS, the dimensions of measurement vector and dictionary matrix will be reduced by half and the proposed method requires less running time and iterations.