Elevation-Dependent Frequency-Domain Imaging for General Bistatic SAR

In the general bistatic SAR (GBiSAR) for imaging of fluctuant terrains, e.g., hilly or mountainous areas, the diverse platform’s types and various geometric configurations of the GBiSAR system lead to a 3-D spatial variance with complicated couplings and significant variation in its frequency-domain matching filter (FDMF) kernel, which brings great challenges to the conventional frequency-domain (FD) imaging techniques. In this paper, an efficient FD imaging algorithm is proposed to compensate the 3-D spatial variance. First, the discrete linear least square (DLLS) is employed to simplify the FDMF kernel. Then a linear mapping converts the spatial-frequency couplings to irregular real numbers. The scaled 2-D nonuniform fast Fourier transform of type 3 (2-D NUFFT-3) is finally performed to speed up the compensation of the 3-D spatial variance. The computational complexity of the algorithm is O(MNlogMN) where MN is the number of the image pixels. The proposed algorithm is straightforward regardless the complicated couplings. In addition, it is robust to accommodate the variation of the FDMF kernel phase. Simulations on several BiSAR systems show that the proposed algorithm can yield high-quality images with good focusing performance and fine geometric registration performance for diverse BiSAR configurations within a widescope fluctuant terrain.