Wavelet-Based Compressed Sensing for SAR Tomography of Forested Areas

Abstract

SAR tomography is a thriving three-dimensional imaging modality that is commonly tackled by spectral estimation techniques. As a matter of fact, the backscattered power along the vertical direction can be readily obtained by computing the Fourier spectrum of a stack of multi-baseline measurements. Alternatively, recent groundbreaking work has addressed the tomographic problem from a parametric viewpoint, thus estimating effective scattering centers by means of covariance matching techniques. In this paper, we introduce a compressed sensing based covariance matching approach that allows us to retrieve the complete vertical structure of forested areas. For this purpose, we employ sparse representations in the wavelet domain and propose suitable pre-filtering techniques. Finally, we validate this approach by using fully polarimetric L-band data acquired by the E-SAR sensor of DLR.