Joint ISAR Imaging and Phase Error Correction Based on Sparse Bayesian Learning

The ISAR imaging algorithm has depend on the mathematical model of the observation process, and the inaccuracies in the observation model may cause the model errors. In this paper, ISAR imaging is regarded as a narrow-band version of the Computer Aided Tomography (CT), where the phase errors in ISAR data are treated as model errors. Based on the inherent sparsity of targets in the imaging area, the ISAR imaging joint with phase adjustment is represented by a sparse signal reconstruction problem, which is set up as an optimization problem in a Sparse Bayesian Learning (SBL) framework. Owing to the superiority of the SBL, we employ an expansioncompression variance-component based method (ExCoV) to reconstruct the target’s scattering coefficients and correct the phase error alternately via the maximum likelihood estimation. The numerical simulation results show the effectiveness of this novel method for various types of phase errors, which can produce a relatively well-focused image of the target and obtains the improvements over existing techniques for model error compensation in ISAR. 