Sparse Subband ISAR Imaging Based on Autoregressive Model and Smoothed $\ell^0$ Algorithm

In the inverse synthetic aperture radar imaging field, the increasing demand of characterizing small targets drives researchers to seek a higher resolution. One attractive way to improve the resolution is sparse subband imaging. In this paper, we propose a novel framework for sparse subband imaging. First, the fine imaging of each subband is performed, through which the high-order phase errors between subband data are compensated. Then, the Keystone transform is performed to unify the image scales of subbands, and sparse subband returns are coherently processed. Finally, the cross-range compressed subband data are interpolated based on the autoregressive model and then used to reconstruct the high-resolution range profile (HRRP) via the smoothed ℓ
⁰
algorithm. In the proposed framework, the cross-range compression, which is performed before HRRP reconstruction, enhances the sparsity of the returns and achieves a considerable SNR gain. Consequently, it paves the way for HRRP reconstruction. Furthermore, the proposed framework needs no filling of the band gap between subbands, and it is robust against the band gap. The real data experimental results demonstrate this framework's ability to achieve a high-resolution fusion image accurately, even under low SNR conditions.