Accelerating CS radar imaging by NUFFT

Abstract

In this paper, a new algorithm based on Nonuniform Fast Fourier Transform (NUFFT) is proposed to accelerate compressed sensing (CS) radar imaging and reduce the memory requirement. This new algorithm reduces the computation complexity of matrix-vector multiplication in CS recovery algorithms from O(LN) to O(3N) and O((log2 2N +1)N), and the memory consumption from O(LN) to O(3N) and O(1.5N). Through-the-wall radar imaging results by Spectral Projection Gradient l1-norm (SPGL1) method demonstrate the efficiency and accuracy of this method.