Efficient Imaging Algorithm for Rail-SAR with K-Space Undersampling

In rail-SAR systems, the synthetic aperture length is always much shorter than the azimuth scope of the imaging scene. To implement wavenumber domain algorithms, the aperture has to be zero-padded to avoid undersampling in the azimuth spectrum. In this paper, the space and frequency coupling in rail-SAR geometry is analyzed and an efficient ω K algorithm is proposed which uses demodulation, interpolation and modulation steps to retrieve information from the undersampled spectrum. We also propose to fold up the azimuth spectrum to avoid oversampling in image azimuth direction. This algorithm greatly reduces the memory occupation and computation load. Measurements of an outdoor scene verify its advantages in computational efficiency, image quality and phase preservation ability.