Esteban Aguilera

Very-High-Resolution Airborne Synthetic Aperture Radar Imaging: Signal Processing and Applications

During the last decade, synthetic aperture radar (SAR) became an indispensable source of information in Earth observation. This has been possible mainly due to the current trend toward higher spatial resolution and novel imaging modes. A major driver for this development has been and still is the airborne SAR technology, which is usually ahead of the capabilities of spaceborne sensors by several years. Todays airborne sensors are capable of delivering high-quality SAR data with decimeter resolution and allow the development of novel approaches in data analysis and information extraction from SAR. In this paper, a review about the abilities and needs of todays very high-resolution airborne SAR sensors is given, based on and summarizing the longtime experience of the German Aerospace Center (DLR) with airborne SAR technology and its applications. A description of the specific requirements of high-resolution airborne data processing is presented, followed by an extensive overview of emerging applications of high-resolution SAR. In many cases, information extraction from high-resolution airborne SAR imagery has achieved a mature level, turning SAR technology more and more into an operational tool. Such abilities, which are today mostly limited to airborne SAR, might become typical in the next generation of spaceborne SAR missions.

Wavelet-Based Compressed Sensing for SAR Tomography of Forested Areas

Synthetic aperture radar (SAR) tomography is a 3-D imaging modality that is commonly tackled by spectral estimation techniques. Thus, the backscattered power along the cross-range direction can be readily obtained by computing the Fourier spectrum of a stack of multibaseline measurements. In addition, recent work has addressed the tomographic inversion under the framework of compressed sensing, thereby recovering sparse cross-range profiles from a reduced set of measurements. This paper differs from previous publications, in that it focuses on sparse expansions in the wavelet domain while working with the second-order statistics of the corresponding multibaseline measurements. In this regard, we elaborate on the conditions under which this perspective is applicable to forested areas and discuss the possibility of optimizing the acquisition geometry. Finally, we compare this approach with traditional nonparametric ones and validate it by using fully polarimetric L-band data acquired by the Experimental SAR (E-SAR) sensor of the German Aerospace Center (DLR).

Multisignal Compressed Sensing for Polarimetric SAR Tomography

In recent years, 3-D imaging by means of polarimetric synthetic aperture radar (SAR) sensors has become a field of intensive research. In SAR tomography, the vertical reflectivity function for every azimuth-range pixel is usually recovered by processing data collected using a defined repeat-pass acquisition geometry. The most common approach is to generate a synthetic aperture in the elevation direction through imaging from a large number of parallel tracks. This imaging technique is appealing, since it is very simple. However, it has the drawback that large temporal baselines can severely affect the reconstruction. In an attempt to reduce the number of parallel tracks, we propose a new approach that exploits structural correlations between neighboring azimuth-range pixels and/or polarimetric channels. As a matter of fact, this can be done under the framework of distributed compressed sensing (CS) (DCS), which stems from CS theory, thus also exploiting sparsity in the tomographic signal. Finally, results demonstrating the potential of the DCS methodology will be validated by using fully polarimetric L-band data acquired by the E-SAR sensor of the German Aerospace Center (DLR).

A Data-Adaptive Compressed Sensing Approach to Polarimetric SAR Tomography of Forested Areas

Super-resolution imaging via compressed sensing (CS)-based spectral estimators has been recently introduced to synthetic aperture radar (SAR) tomography. In the case of partial scatterers, the mainstream has so far been twofold, in that the tomographic reconstruction is conducted by either directly working with multiple looks and/or polarimetric channels or by exploiting the corresponding single-channel second-order statistics. In this letter, we unify these two methodologies in the context of covariance fitting. In essence, we exploit the fact that both vertical structures and the unknown polarimetric signatures can be approximated in a low-dimensional subspace. For this purpose, we make use of a wavelet basis in order to sparsely represent vertical structures. Additionally, we synthesize a data-adaptive orthonormal basis that spans the space of polarimetric signatures. Finally, we validate this approach by using fully polarimetric L-band data acquired by the E-SAR sensor of the German Aerospace Center (DLR).

Polarimetric 3-D reconstruction from multicircular SAR at P-band

Multicircular synthetic aperture radar (SAR) (MCSAR) is an extension of circular SAR (CSAR) characterized by the formation of a synthetic aperture in elevation with several circular flights. This imaging mode allows an improved resolution in the plane perpendicular to the line of sight ( LOS⊥), thus suppressing the 3-D cone-shaped sidelobes that are formed when focusing with CSAR. This letter presents the first polarimetric MCSAR airborne experiment acquired at P-band by the German Aerospace Center (DLR)s F-SAR system over a forested area in Vordemwald, Switzerland. This letter also includes a phase calibration method based on the singular value decomposition (SVD) using ground signatures to estimate constant phase offsets within a stack of 2-D images. Focusing methods, such as fast-factorized back projection (FFBP), beamforming (BF), and compressive sensing (CS), described in previous publications are used to solve the complex reflectivity in the (x, y, z) space.

Multi-signal compressed sensing for polarimetric SAR tomography

In recent years, three-dimensional imaging by means of SAR tomography has become a field of intensive research. In SAR tomography, the vertical reflectivity function for every azimuth-range pixel is usually recovered by processing data collected using a defined repeat pass acquisition geometry. The most common approach is to generate a synthetic aperture in the elevation direction through imaging from a large number of parallel tracks. This imaging technique is appealing, since it is very simple. However, it has the drawback that large temporal baselines, which is the case for space-borne platforms, can severely affect the reconstruction. In an attempt to reduce the number of parallel tracks, we propose a new tomographic focusing approach that trades number of SAR images for correlations between neighboring azimuth-range pixels and polarimetric channels. As a matter of fact, this can be done under the framework of Distributed Compressed Sensing (DCS), which stems from Compressed Sensing (CS) theory, thus also exploiting sparsity in our tomographic signal. In addition, we address the problem of measurements affected by additive as well as multiplicative speckle noise. Results demonstrating the potential of the DCS methodology will be validated by using fully polarimetric L-band data acquired by the E-SAR sensor of DLR.

Joint Sparsity Model for Multilook Hyperspectral Image Unmixing

Recent work on hyperspectral image (HSI) unmixing has addressed the use of overcomplete dictionaries by employing sparse models. In essence, this approach exploits the fact that HSI pixels can be associated with a small number of constituent pure materials. However, unlike traditional least-squares-based methods, sparsity-based techniques do not require a preselection of endmembers and are thus able to simultaneously estimate the underlying active materials along with their respective abundances. In addition, this perspective has been extended so as to exploit the spatial homogeneity of abundance vectors. As a result, these techniques have been reported to provide improved estimation accuracy. In this letter, we present an alternative approach that is able to relax, yet exploit, the assumption of spatial homogeneity by introducing a model that captures both similarities and differences between neighboring abundances. In order to validate this approach, we analyze our model using simulated as well as real hyperspectral data acquired by the HyMap sensor.

A data adaptive compressed sensing approach to polarimetric SAR tomography

Super-resolution imaging via compressed sensing (CS)-based spectral estimators has been recently introduced to synthetic aperture radar (SAR) tomography. In the case of partial scatterers, the mainstream has so far been twofold, in that the tomographic reconstruction is conducted by either directly working with multiple looks and/or polarimetric channels or by exploiting the corresponding single-channel second-order statistics. In this letter, we unify these two methodologies in the context of covariance fitting. In essence, we exploit the fact that both vertical structures and the unknown polarimetric signatures can be approximated in a low-dimensional subspace. For this purpose, we make use of a wavelet basis in order to sparsely represent vertical structures. Additionally, we synthesize a data-adaptive orthonormal basis that spans the space of polarimetric signatures. Finally, we validate this approach by using fully polarimetric L-band data acquired by the E-SAR sensor of the German Aerospace Center (DLR).

Spectral-spatial joint sparsity unmixing of hyperspectral data using overcomplete dictionaries

Sparse spectral unmixing can be modeled as a linear combination of endmembers contained in an overcomplete dictionary weighted by the corresponding sparse abundance vector. This method exploits the fact that there is only a small number of endmembers inside a pixel compared to the overcomplete endmember spectral dictionary. Since the information contained in hyperspectral pixels is often spatially correlated, in this work we propose to jointly estimate the sparse abundance vectors of neighboring hyperspectral pixels within a local window exploiting joint sparsity with common and noncommon endmembers. To demonstrate the efficiency of our framework, we perform experiments using both simulated and real hyperspectral data.