presented in Figure 7. Associated elevation values are estimated with Capon filtering in case of single PS and with sequential NLLS search in case ofdouble PS. As expected, stable targets are mainly identified on roof and facades of constructions. Double PS are identified in regions prone to layover phenomenon occurrence. Main contributions presented in this work consist in adaptation of fast-sup-GLRT detector to a ML version which uses Capon filtering for dominant scatterers position estimation and implementation of a statisticalanalysis to determine an optimal procedure which avoids multiple PS classification of same scattering contribution. This supplementary analysis is required because the multi-look leads to a widening of the main lobe of scatterer’s response, since contributions present in multiple resolution cells are averaged. Therefore, in case of the presence of a single scatterer, NLLS is prone to estimate the position of the secondary scatterer close to the one of the primary. A minimum distance equal to one third of Rayleigh resolution was imposed between doublescatterers elevation positions. The analysis indicated that the optimal way to impose this constraint is within the detection process, in points classified exclusively by NLLS search as double PS, since those are the ones placed within a distance below Rayleigh resolution.
Abstract: The estimation of deformation parameters using persistent scatterer interferometry (PSI) is limited to single dominant coherent scatterers. As such, it rejects layovers wherein multiple scatterers are interfering in the same range-azimuth resolution cell. Differential synthetic aperture radar (SAR) tomography can improve deformation sampling as it has the ability to resolve layovers by separating the interfering scatterers. In this way, both PSI and tomography inevitably require a means to detect coherent scatterers, i.e., to perform hypothesis testing to decide whether a given candidate scatterer is coherent. This paper reports the application of a detection strategy in the context of “tomography as an add-on to PSI”. As the performance of a detector is typically linked to the statistical description of the underlying mathematical model, we investigate how the statistics of the phase instabilities in the PSI analysis are carried forward to the subsequent tomographic analysis. While phase instabilities in PSI are generally modeled as an additive noise term in the interferometric phase model, their impact in SARtomography manifests as a multiplicative disturbance. The detection strategy proposed in this paper allows extending the same quality considerations as used in the prior PSI processing (in terms of the dispersion of the residual phase) to the subsequent tomographic analysis. In particular, the hypothesis testing for the detectionof coherent scatterers is implemented such that the expected probability of false alarm is consistent between PSI and tomography. The investigation is supported with empirical analyses on an interferometric data stack comprising 50 TerraSAR-X acquisitions in stripmap mode, over the city of Barcelona, Spain, from 2007–2012.
for the automated extraction of the information about the height and deformation velocity of single and multiple scatterers in the same radar cell. More specifically, the contribution of this paper is threefold. First, the originator method of superresolution adaptive Diff-Tomo is improved beyond the first tests in  in order to handle automatically a large data set by augmenting it with a new original scatterer multiplicity detector for the extraction of the height/deformation velocity information. In particular, we tackle the detection problem by combining adap- tive Diff-Tomo with a model-based least squares (LS) fitting in the complex (i.e., amplitude and phase) data domain. All the theoretical details and the discussion about the novelty of the proposed identification (i.e., detection and parameter estimation) algorithm are reported in Section II. Second, by means of the developed scatterer multiplicity detector, the extracted height/deformation velocity information is validated extensively, instead of on a number of selected cells. Third, a preliminary phenomenological analysisof the characteristics of the detected scatterers is performed, which is novel also considering the superresolution of the adaptive processing. These two points and the related results have been addressed in Section III by processing real C-band ERS-1/2 data over an area around the San Paolo stadium in the city of Naples and over the Cinecittà area in the city of Rome, with particular emphasis on single and doublescatterers.
The simulated positions of triple bounce contributions are expected to be reliable as the relative shifts with respect to the corresponding point signatures are small (Figure 4c). In contrast, pixels marked as fivefold bounce do not directly correspond to salient signatures but are located close to intensity peaks. The specular map, shown in Figure 4d reveals that almost all simulated signal contributions followed specular directions. For providing the map, the necessary information for the decision specular/non-specular has been derived during the ray tracing step. Based on a geometrical analysisof signal reflection, each signal bounce is marked in case of specular reflection. Thus, the occurrence of weak fivefold bounce signatures may be possible.
Considering CM-based features, the results have shown the importance of interferometric coherences as a feature. In fact, the intensities of the TomoSAR CM are not expected to carry as much information as the PolSAR ones because they all arise from the same polarimetric channel, although from a slightly different view angle. Our experiments showed the relevance of using the interferometric coherences as features. This tends to confirm the findings of previous works such as [ 39 , 47 , 48 ] exploiting the dependency of interferometric coherence on the type of target. However, simple methods considering unimodal data distribution such as Wishart will fail to capture the possible variations of class statistics across the image, which are more likely to occur on multi-baseline datasets due to a loss of correlation in the near range area and variation of the vertical resolution depending on slant range. To illustrate the previous observations, Figure 11 shows examples of coherences for two different baselines. Nevertheless, the use of RF classifier allowed to obtain a better performance than Wishart from both PolSAR and TomoSAR CM, due to its ability to handle multi-modal distributions. When using RF in combination with coherence features, TomoSAR CM classification consistently outperformed PolSAR. Therefore, the issue of range-dependent decorrelation related to multi-baseline images can be easily overcome by using a more sophisticated classifier that is able to model multi-modal distributions.
Furthermore, the separation of multiple PS within one resolution cell has been addressed to in the past years. Adam et al. (2005) proposes a method to detect the appearance of two scatterers from amplitude only data. The baseline dependent reflectivity is investigated and the decision for one or two PS is supported by Bayesian theory. In addition, the relative height between the scatterers can be estimated along elevation direction. 15% of the PS are reported to be composed of two dominant scatterers. Higher order models corresponding to three or more scatterers are not considered. Ferretti et al. (2005) shows a possible extension of the original PSI approach for the separation ofscatterers. A second order model is introduced, which can be extended to multi-scatterer cases. However, the number ofscatterers is not detected directly, but a hypothesis testing procedure has to be used in order to find the model that fits the underlying data best. This implies an increase in calculation times for each scattering model used. For typical urban areas up to 20% of PS may be detected in addition using second order models, but at higher computational costs. An initial use of first order models is proposed, because these are regarded to be sufficient as long as the highest possible number of PS is not needed. By reason of the increased complexity, these approaches have to be carefully balanced in terms of necessity for the given investigation. Fore sure, an increased number of PS is always desirable, but only inevitable for areas suffering from a sufficient number of points in order to analyze the expected deformation phenomenon or provide a good coverage of PS within the area of interest. However, the separation of multiple scatterers within one resolution cell can be achieved by recently developed tomographic methods (Lombardini, 2003; Fornaro and Serafino, 2006), which are capable of resolving layover. It is applied to ERS data by Fornaro et al. (2005) and for the first time to high resolution SAR data of TerraSAR-X by Zhu et al. (2008) (see also Zhu and Bamler (2010c)). An increase in number of PS up to 30% is reported in Zhu and Bamler (2010a), especially due to the separation of many doublescatterers at low heights with respect to ground surface.
Modern Tomographic SAR is an advanced InSAR techniques for urban mapping, which can not only retrieve 3D spatial information but also assess the 4D temporal information, such as deformation. To retrieve the information from InSAR data, several algorithms have been developed. Among them, SL1MMER algorithms is state of the art. However, it suffers from the computational expenses and it is hard to extend to large scale practice. In this work, we
Tomographic synthetic aperture radar (SAR) imaging has been recently formulated in a wavelet-based compressed sensing (CS) framework. This paper reviews the underly- ing sparsity-driven algorithms for single-channel as well as polarimetric tomography, and discusses its applicabil- ity in terms of ambiguity rejection, physical validity, ac- quisition geometry, and required a priori knowledge. In addition, we present a comparison with traditional non- parametric spectral estimators by using L-band data ac- quired by the Experimental SAR (E-SAR) sensor of the German Aerospace Center (DLR).
Fig. 13 shows the result of tomographic analysisfor the corresponding position in real TerraSAR-X data. The reflection profile has been calculated with the approach described in (Zhu et al., 2008). As input data, 16 TerraSAR-X spotlight images with an across-track baseline range of 270m have been used. The peaks in reflection profile show nice correspondence with the simulated results, which underlines the accurate geometric properties of the simulation. However, it has to be noted that accurate estimation of intensity proportions is not possible as ground truth for surface properties was not available. At this point, simulated intensity values only indicate a stronger diffuse backscattering from the plaza which is also visible in the reflectivity map extracted from real SAR data. Enhanced information about the scattering behaviour of the plaza and the convention center may enable better simulation results in the
The E-SAR Synthetic Aperture Radar (SAR) system onboard a DLR Dornier DO-228 aircraft operates in 4 frequency bands, X-, C-, L- and P-band. The polarisation of the radar signal is selectable. E-SAR measurement modes include single channel operation, i.e. one wavelength and polarisation at a time, and the modes ofSAR Interferometry and SAR Polarimetry. The system is
per polarimetric channel in the elevation direction obtained by processing the signals individually according to the single- signal approach (SSA). Additionally, Fig. 4(c) and (f) show the jointly processed profiles found by means of the multisignal approach (MSA). Noticeably, unlike the SSA, the MSA allows for a correct identification of the support throughout polariza- tions. Then, once the positions of the nonzero elements have been identified, an accurate complex reflectivity can be readily retrieved by the method of least squares . Experimentally, we have found that, even when dealing with irregular baseline distributions, we can still find the support, albeit at the expense of more passes. In this case, we needed at least two more baselines.
SARTomography (SARTom) is an imaging technique that al- lows multiple phase centre separation in the vertical (height) direction, leading to a 3D reconstruction of the imaged scene. It is usually performed after standard 2D SAR processing and operates on a stack of coregistered SAR images. In  the first demonstration of airborne SARtomography, using Fourier beamforming techniques, has been carried out and the main constraints in terms of resolution and ambiguity rejec- tion have been analysed. If the number ofscatterers to be solved inside a resolution cell is a priori known, it is possi- ble to reduce the number of acquisitions , anyhow, for the general case this information is not known and a generic vol- umetric target has to be assumed. In this case, the ambiguity height V defines the baseline dN yq between the acquisitions
Distributed Compressed Sensing (DCS) theory enables the joint recovery of multi-signal ensembles by exploiting inter- signal correlations. It generalizes the concept of a signal be- ing sparse in some basis to the concept of an ensemble of signals being jointly sparse. In this paper, we demonstrate how to apply a multiple measurement vector model that has been thoroughly studied and can be found in the literature , . One of the crowning achievements of this model is that it allows us to reduce the number of measurements needed for reconstruction. When it comes to SARtomography, this translates into reducing the number of 2-D SAR images.
Note that, for MSF and DCRCB, the set of tomograms depicting the wings of the selected edifice (Figures 11 – 13 ) present high ambiguity levels among the PLOS height range from −35 m to 60 m and from −10 m to −20 m, with a pseudo-power up to −3 dB and −5 dB, respectively, in the VV polarization. CS and WISE perform reduction of the ambiguity levels; however, the presence of ambiguities is still significant in some positions, especially for VV, which may lead to false detections. The analysisof all polarizations and the succeeding tomograms helps discriminating these ambiguities. The latter makes more sense for the single-look case, since the spatial mixture of sources is avoided. Moreover, we can also compare the single-look response against the multilook response. By instance, the ambiguity levels for CS and WISE in Figure 11 (HH polarization), among the aforesaid PLOS height ranges, are weaker. Furthermore, these same ambiguities do not appear in the succeeding tomograms in Figures 15 – 17 . Without having a priori information on the ROI, we could infer that the sources within such ranges, for the VV polarization in Figure 13 , are indeed ambiguities. Thus, the PLOS height range can be set accordingly, e.g., from −10 m to 30 m.
Due to the unavailability of suitable Tandem-L bistatic test data, we applied the proposed framework to a small TanDEM- X pursuit monostatic stack. The pursuit monostatic mode was temporarily put into practice from October 2014 to February 2015 during the TanDEM-X Science Phase . In order to avoid RF interference between radar signals, the along-track distance was set to approximately 76 km, which corresponds to a temporal baseline of circa 10 seconds. During this five-month period, 12 staring spotlight scenes of the City of Las Vegas were acquired. Out of these, 6 pursuit monostatic interfero- grams were generated and their baselines are plotted in Fig. 2a. As can be observed, relatively large values in magnitude are available, whereas in the usual cases of TSX and TDX 2 the baselines are bounded between ±250 m. As a matter of fact, in order to favor TomoSAR and other applications in polar regions, cross-track perpendicular baselines were programmed to slowly drift (in magnitude) from 0 to 750 m . Since all baselines but one are negative, we applied the sign flipping procedure that was introduced in Sec. II-A. The baselines after sign flipping are plotted in Fig. 2b. The sign was indeed flipped for two baselines and the standard deviation increased from approximately 286.7 to 308.3 m. As a consequence, the CRLB was improved by 7.5%.
Modern spaceborne synthetic aperture radar (SAR) sensors, such as TerraSAR-X (TSX), TanDEM-X (TDX) and COSMO- SkyMED, deliver SAR data with very high spatial resolution of up to 1 m. With these meter-resolution data, advanced multi- pass interferometric techniques like persistent scatterer interferometry (PSI) and tomographic SAR (TomoSAR) allow retrieving not only the 3D geometrical shape but also the undergoing temporal motion in the millimeter scale of individual buildings –. In particular, sparse reconstruction based methods , e.g. SL1MMER , give robust TomoSAR inversion with very high elevation resolution, and can offer so far ultimate 3D, 4D and 5D SAR imaging .
Abstract—The world’s first ADS-B over Satellite (AOS) In- Orbit Demonstrator (IOD) within ESA’s PROBA-V mission is operational since May 2013 and has successfully validated the principle of detecting weak Mode S transponder transmissions from a Low Earth Orbit (LEO). A special feature was included in the receiver’s firmware that allows to upload new configurations and to activate these by remote access. During mission runtime so far, this has been successfully tested several times. In the meanwhile an improved Mode S correlation mechanism was developed that benefits from the phase coherence of the pulse train from the first Mode S preamble pulse to the fifth format bit. In lab tests it could be shown that the telegram detection rate increased significantly. Moreover, by generating and saving ”Low-Confidence Bits” for the 112 Mode S data bits in DF17, there is an additional chance to increase the success rate for error-free demodulation of the telegram in post-processing. The improvement on ADS-B data from space in comparison with the results gained with the non coherent approach will be shown.
We first demonstrate the performance of LPSI for in vivo structural retinal imaging. The subject was a healthy volunteer. The location of the measurements was controlled with a fixation target presented to the measured eye via dichroic beam splitter. All presented images in this section were acquired at an acquisition speed of 600kA-scans/s and an optical power of 4.6 mW at the cornea (Config. A1). Figure 4 shows a representative 2D wide-field image obtained after stitching 7 successive overlapping tomograms. The lateral FOV is close to 30°. The tomograms were acquired at 2mm intervals with sufficient overlap to allow proper stitching. The stitching allows to better evaluate and compare the performance of the system for the various regions of the retina. The retinal imaging performance in Fig. 4(a) is obtained without any averaging. The detection sensitivity and resolution are high enough to visualize the external limiting membrane (ELM), small blood vessels in the region around the fovea centralis, and to contrast the various layers of the inner retina. We labeled the different retinal layers and anatomical details for better appreciation. Despite the loss of confocality in one lateral dimension and the shorter center wavelength of 840nm as compared to systems operated in the 1060nm wavelength range, the signal intensity from the choroid below the highly scattering RPE is remarkably strong maintaining structural details even in the non- averaged tomograms. Also the depth range as well as the depth sampling is sufficient for supporting both the axial resolution as well as the visualization of all retinal structures including the choroid and even at the optic nerve head. Figure 4(b) shows the same stitch obtained after averaging over 4 successive tomograms for speckle and noise reduction. Due to the speckle averaging effect, Fig. 4(b) exhibits further contrast improvementof choroidal structure and allows better delineation of inner retinal layers. All tomograms were acquired by placing the structural components and complex conjugate artifacts across the zero path length delay. In both wide-field images, the CC artifact was completely removed. The DC term is successfully removed in nearly all of the images, however, the suppression was not complete in regions around the fovea centralis. This is attributed to the corneal reflex that leads to strong fluctuations of the non-interferometric background. Furthermore, we can appreciate the effect of the inverse Gaussian weighting method, since the signal intensity has a smooth continuation across the stitch locations.
S INCE the first practical demonstration of synthetic aper- ture radar (SAR) tomography (TomoSAR), the volumet- ric analysisof forested areas by this technique has been an important research topic . In this context, most of the lit- erature has focused on long-wavelength radar, such as L- or P-band –. Only few experiments have investigated the potential of shorter wavelength SAR using X-band sensors . However, recently, a TomoSAR inversion method aiming at the reconstruction of discrete scattering profiles  has been proposed, which has already been used to generate detailed 3-D point clouds of forested areas using Ka-band data with a wavelength in the millimeterwave domain . Based on these point clouds, even the 3-D reconstruction of individual trees could be demonstrated . The results of these studies indicated that millimeterwave SAR provides the advantage of showing almost no canopy penetration and therefore providing accurate height estimates almost comparable to LiDAR remote sensing. In contrast, it is still an open question whether millimeterwave signals do provide any canopy penetration at all, and whether they could potentially be employed for a TomoSAR analysisof the whole forest volume. This letter provides the first- ever demonstration of volume SARtomography using airborne multiantenna millimeterwave SAR data.