Moving Target Imaging for Synthetic Aperture Radar Via RPCA

Synthetic aperture radar (SAR) imaging of moving targets is a challenging task, as standard techniques have been developed for stationary scenes. Motivated by success of robust principal component analysis (RPCA) in change detection for video processing, we establish a rank-1 and sparse decomposition framework for the SAR problem in the image domain. We construct the phase-space reflectivity matrix for single-channel SAR systems reconstructing images at various hypothesized velocities and show that it is the superposition of a rank-1 matrix and a disjoint sparse matrix. This structure allows for additional constraints that reduce the computational complexity when compared to generic RPCA. We compare the performances of two algorithms, proximal gradient descent (PGD) and alternating direction method of multipliers (ADMM), on numerical simulations for the moving target imaging problem.