SPARSE SELF-CALIBRATION FOR MICROWAVE STARING CORRELATED IMAGING WITH RANDOM PHASE ERRORS

Microwave Staring Correlated Imaging (MSCI) technology can obtain high-resolution images in staring imaging geometry by utilizing the temporal-spatial stochastic radiation field. In MSCI, sparse-driven approaches are commonly used to reconstruct the target images when the radiation fields are accurately calculated. However, it is challenging to compute radiation filed with high precision due to the existence of random phase errors in MSCI systems. Therefore, in this paper, a self-calibration method is proposed to handle the problem. Specifically, a two-step self-calibration framework is applied which alternately reconstructs the target image and estimates the random phase errors. In the target image reconstruction step, sparse-driven approaches are utilized, while in the random phase errors calibration step, an adaptive learning rate method is adopted. Moreover, the batch-learning strategy is utilized to reduce computation burden and obtain effective convergence performance. Numerical simulations verify the advantage of the proposed method to obtain good imaging results and improve random phase errors correction performance.