Three-Dimensional Reconstruction of Man-Made Objects Using Polarimetric Tomographic SAR

The growing interest in 3-D reconstruction of targets, particularly man-made objects, has made synthetic aperture radar (SAR) tomography a hot topic in recent years. It obtains a 3-D reflectivity map by coherently processing multiple images acquired from slightly different views. This paper investigates the use of polarimetric tomographic SAR to reconstruct 3-D images of man-made objects and proposes a polarimetric 3-D reconstruction technique, based on the multiple-measurement vector compressive sensing (CS) (MMV-CS) model and Tikhonov regularization theory. The main feature of the technique is that of allowing joint reconstruction of polarimetric 3-D reflectivity maps with high resolution. For X-band tomographic SAR, we demonstrate that the phase error induced by an array element error is severe enough to prevent data focusing and propose a phase calibration technique employing prominent scatterers to adapt the proposed technique to practice. To check the plausibility of our algorithms, a ground-based SAR system is introduced, and high-resolution 3-D images obtained by processing the real collected data are presented. As special cases of the new proposed technique, both the Tikhonov regularization and the MMV-CS method reconstruct the targets successfully and give an accurate estimate for the vehicle size. Compared to the Tikhonov regularization method, the MMV-CS technique exhibits much better superresolution power and lower sidelobe level. The advantage of the MMV-CS technique over the single-measurement vector CS technique is also demonstrated, and the phase calibration algorithm is verified.