Multicircular SAR 3-D Imaging: A Multisubaperture Sparse Processing Approach via Structured Matrix Recovery
Abstract
Multicircular synthetic aperture radar (MCSAR) is an effective measurement tool for 3-D mapping of urban scenes and can acquire full azimuth-angle observations with different perspectives at multiple elevation positions. In the conventional noncoherent MCSAR 3-D imaging framework, subapertures are processed separately to mitigate anisotropic scattering effects, leaving the inherent correlations among adjacent observations underutilized. The accuracy of 3-D imaging can be further improved by exploiting these correlations. To this end, we propose a multisubaperture sparse processing-based structured matrix recovery (MSSP-SMR) algorithm, which exploits intersubaperture correlations through a sparsity prior along the elevation direction. In particular, a Hankel–Toeplitz structured matrix model is constructed to characterize the structure of topographic sparsity. Under the low-rank assumption, this structure enables the nonconvex maximum likelihood estimation (MLE) of the elevation spatial spectrum to be transformed into a rank-constrained structured matrix recovery (SMR) problem. Furthermore, the SMR formulation is extended to the multiple measurement vector (MMV) case to exploit the joint sparsity of adjacent subapertures. The proposed MSSP-SMR algorithm enables accurate estimation of continuous spatial frequencies, thereby supporting high-precision superresolution 3-D imaging. Experiments on both simulated and measured datasets validate the effectiveness of the proposed algorithm.