Resolution Enhancement for LASAR 3D Imaging via 1 Regularization and SVA

Linear array SAR (LASAR) has been attracting more and more attention for its capability of obtaining three dimensional (3D) resolutions. However, the low resolution in cross track (CT) direction limited by the length of its linear antenna array has become the bottleneck of its practical applications. To overcome this problem, we present a novel algorithm based on sparse reconstruction (SR) to improve the resolution in CT direction. First, it establishes a 1D real-valued sparse model for LASAR, which deals with the 3D image column by column along CT direction in each equi-range slice. This enables it to handle large scenes. Second, it employs the spatially variant apodization (SVA) to filter bases of the measurement matrix. As a result, the cross coherence gets suppressed as well, and it is helpful to improve the performance of sparse reconstruction algorithms (SRAs). Third, we propose the resolution enhancement ability (REA), which provides a new idea to evaluate how many times the resolution could be improved. Experimental results validate that when the signal to noise ratio (SNR) is 30 dB, LASAR could usually obtain 2 times of resolution improvement in CT direction, while the proposed method further improves the REA by a factor about 2.5. Moreover, the 3D surface terrain simulation shows a great improvement for the digital elevation map (DEM) in resolution enhancement.