SAR imaging via modern 2-D spectral estimation methods

Abstract

This paper discusses the use of modern 2D spectral estimation algorithms for synthetic aperture radar (SAR) imaging. The motivation for applying power spectrum estimation methods to SAR imaging is to improve resolution, remove sidelobe artifacts, and reduce speckle compared to what is possible with conventional Fourier transform SAR imaging techniques. This paper makes two principal contributions to the field of adaptive SAR imaging. First, it is a comprehensive comparison of 2D spectral estimation methods for SAR imaging. It provides a synopsis of the algorithms available, discusses their relative merits for SAR imaging, and illustrates their performance on simulated and collected SAR imagery. Some of the algorithms presented or their derivations are new, as are some of the insights into or analyses of the algorithms. Second, this work develops multichannel variants of four related algorithms, minimum variance method (MVM), reduced-rank MVM (RRMVM), adaptive sidelobe reduction (ASR) and space variant apodization (SVA) to estimate both reflectivity intensity and interferometric height from polarimetric displaced-aperture interferometric data. All of these interferometric variants are new. In the interferometric contest, adaptive spectral estimation can improve the height estimates through a combination of adaptive nulling and averaging. Examples illustrate that MVM, ASR, and SVA offer significant advantages over Fourier methods for estimating both scattering intensity and interferometric height, and allow empirical comparison of the accuracies of Fourier, MVM, ASR, and SVA interferometric height estimates.

SAR imaging via efficient implementations of sparse ML approaches

“…required by the SMLAs, resulting in the approximative algorithms here termed the QN-SMLA SMLAs, stems from the fact that the complexity of computing the displacement representation of the approximate inverse covariance in (75) using the LWWR algorithm requires approximately 1:5N 3 1 ðM Q 2 Þ 2 operations as opposed to the 1:5N 3 1 N 2 2 operations required by the displacement representation of R À1 N 1 N 2 , where, typically,…”

“…The LWWR algorithm requires approximately C LWWR ¼ 1:5N 3 1 N 2 2 operations. This figure may be reduced to C LWWR1 ¼ N 3 1 N 2 2 , if the persymmetric property of the TBT matrix R N 1 N 2 is taken into account, according to which…”

“…Typically, SAR images are formed using periodogram-based estimators, thereby suffering from the well-known limitations in resolution and high leakage levels. Various forms of data-adaptive and often non-parametric approaches can enhance the resolution while reducing these shortcomings [1][2][3][4][5]. Generally, these methods require large data sets to offer reliable estimates of the second-order statistics, a requirement that is hard to satisfy in practice.…”

Through the Wall Radar Imaging with MIMO beamforming processing - Simulation and Experimental Results

“…Note that in our MIMO data signal matrix the steering operation will depend not only on antennas positions but also on frequency so the processing can be done at every bin (f l ) and sum over frequency. The output power of conventional beamforming at f l is expressed as: (9) where w l is the steered vector to position x at frequency f l . In far field we would have w = a l (x) but in order to take into account near field we multiply by a vector d that compensates the 1/(r mt r tn ) attenuation and w is given by the element wise multiplication of d by a l (x).…”

“…This processings are applied here for TTW imaging. Other spectral estimation techniques have already been used for imaging [9] like in SAR, but here the implementation should take into account near field considerations and TTW propagation. Imaging with simulated numerical data for different scenarii illustrates first results.…”