Synthetic Aperture Radar Interferometry (InSAR) is utilized to evaluate slope stability, revealing pronounced periodic oscillations in the deformation time series. Although such periodic patterns have conventionally been ascribed to stratified tropospheric delays or seasonal precipitation in prior studies, periodic signals persist in the deformation results even after atmospheric phase removal by a linear iterative model. This observation underscores the limitations of conventional deformation interpretations and prompts further investigation into the underlying physical mechanisms. To address this, an interferometric phase correction model accounting for variations in surface dielectric property has been proposed. This model effectively removes phase delays induced by dielectric property changes from the raw interferometric phase, enabling the extraction of linear trend signals that reflect actual surface deformation. To verify the reliability of the correction model, Sentinel-1A (C-band) and TerraSAR-X (X-band) radar data is employed to quantify deformation patterns in the slope of non-sliding section around the Huangnibazi landslide. The analysis consistently identifies periodic deformation signals in both SAR datasets after mitigating atmospheric influences. Our findings indicate that dielectric property changes constitute a critical factor in InSAR deformation monitoring that cannot be overlooked. The observed periodic fluctuations in deformation time series are attributed to dielectric-induced phase modulation effects. Furthermore, systematic correlation analyses confirm a strong coherence between soil moisture variations and deformation fluctuations, with minimal temporal hysteresis. In contrast, seasonal precipitation exhibits a weaker correlation with deformation and longer hysteresis time. These results robustly support the theoretical framework linking soil moisture, dielectric property, penetration depth, and phase delay. This insight holds significant implications for enhancing the effectiveness of InSAR technology in slope disaster monitoring and early warning systems.

This paper investigates the use of the Phase Histogram (PH) technique for the remote sensing of forested areas using Synthetic Aperture Radar (SAR) data. The PH technique assigns each pixel in a SAR interferogram to a specific height bin based on the value of the corresponding interferometric phase, thus allowing for the estimation of the forest vertical structure by accumulating pixels magnitudes within a given spatial window. In this paper, we first analyze the formation of phase histograms from a theoretical perspective, explicitly considering the role played by phase dispersion as a function of the number of targets within the SAR resolution cell. Theoretical developments are followed by further experimental analyses of L-Band data from the ESA campaign TomoSense, flown at the Eifel National Park, Germany, in 2020/2021. Theoretical and experimental results indicate that the applicability of the PH technique is subject to the assumption that Radar returns are determined by the presence of a dominant scatterer in each SAR resolution cell, in which case the phase histogram can successfully represent the forest electromagnetic structure. This leads to the conclusion that the PH technique is best suited for the analysis of high-frequency and/or highresolution data, in which conditions PH may perform satisfactorily based on single interferograms.

Tropospheric delay significantly hinders the accurate acquisition of high-precision surface deformation by Time series Interferometric Synthetic Aperture Radar (InSAR). The main challenge for current InSAR tropospheric delay estimation lies in effectively utilizing the time-dependent characteristics of the tropospheric delay for accurate atmospheric delay estimation. This paper develops a model to estimate the time-dependent and stochastic components of the delay based on periodic and random characteristics. The experiment demonstrates the effectiveness of the proposed method regardless of whether terrain-dependent delays dominate or random delays dominate at Danba-Xiaojin. The Std of the corrected decreases in 89/90 of the interferograms. The average and maximum improvement of Std is more than 40% and 80% respectively. From the time series, the proposed method can effectively suppress the periodic signals in both non-deformation and deformation regions and can obtain smoother time series. Overall, the proposed method outperforms the other four models for InSAR tropospheric delay correction.

This article evaluates the recently introduced phase histogram (PH) technique for estimating forest height and vertical structure using theoretical modeling and experimental synthetic aperture radar (SAR) data. The PH technique assigns each pixel in an SAR interferogram to a specific height bin based on the value of the corresponding interferometric phase, thus allowing for the estimation of the forest’s vertical structure by accumulating pixels magnitudes within a given spatial window. This approach is radically different from the one employed by SAR tomography (TomoSAR), which allows for direct imaging of the 3-D structure of the vegetation by jointly focusing on SAR data from multiple trajectories. Importantly, PHs can be built using as few as two images (a single interferogram), whereas TomoSAR is well-known to perform best when many images area available. Accordingly, the main question we intend to address in this article is to what extent and in which conditions single-baseline PHs can be used as a surrogate of TomoSAR (in the absence of multibaseline data). Experimental analyses are conducted using L-band tomographic SAR data from the ESA campaign TomoSense, flown in 2020 at Eifel Park in North West Germany. TomoSense data include 30 + 30 monostatic overpasses acquired along two opposite flight headings, and are complemented by airborne, terrestrial, and unmanned aerial vehicle (UAV) Lidar surveys. Lidar data are used to generate a forest canopy height model (CHM) and vertical profiles of leaf area density (LAD), taken as the main reference in the evaluation of PHs. Multibaseline tomographic data are produced and investigated to assess the actual sensitivity of radar data to forest structure at this site, as well as to provide indications about the performance of a radar instrument when multiple baselines are available. Experimental results indicate that the PH technique can only loosely approximate the vertical structure produced by TomoSAR. Still, it can produce a reasonably good estimate of forest height. In particular, TomoSAR and the PH technique are observed to have an average root mean square error (RMSE) with respect to Lidar estimate of 2.8 and 4.45 m in North-West heading data, and 1.84 and 5.46 m in South-East heading data, respectively. The observed results are interpreted in light of a simple physical model to characterize PHs depending on the number of scatterers within the SAR resolution cell, on which basis we derive analytical expressions to predict height dispersion in PHs. The proposed model indicates that the concept of PH is inherently based on the assumption of a single dominant scatterer within any single SAR resolution cell. If this is not the case, PHs produce an intrinsic dispersion that does not represent the actual vertical distribution of scatterers within the vegetation. Consistently, we conclude that the PH technique is inherently best suited for the analysis of high- or very-high resolution data, which suggests its use in the context of higher frequency SAR missions (e.g., Tandem-X) and when there are few acquisitions available.

This paper compares two techniques for obtaining forest height and vertical structure from synthetic aperture radar (SAR) data, namely SAR tomography (TomoSAR) and phase histogram (PH). The comparison is carried out on an experimental basis by using the tomographic SAR dataset from the TomoSense campaign, flown by ESA in 2020/21 to support investigation of forested area for future low frequency spaceborne SAR missions. Both techniques were tested using data at full and degraded resolution. Experimental results show that the PH technique can only loosely approximate the forest vertical structure produced by SAR tomography, although it was able to achieve a fairly good estimate of forest height if the appropriate range of height of ambiguity is used. A degraded performance of the PH technique when applied to low-resolution data indicates that this technique is best fit for the case of high-resolution data, consistently with the assumption of the presence of dominant scatterers. Overall, these findings indicate the PH technique as an interesting option in the context of high-resolution spaceborne missions.

The new ESA TomoSense campaign aims at investigating a temperate forest located at the Eifel natural park, north-west Germany by means of Synthetic Aperture Radar (SAR) Tomography (TomoSAR). Multifrequency (P, L, C), tomographic SAR data were acquired by repeated flights at different heights. Particularly in L- and C-band surveys, two aircrafts were simultaneously flying operated in a single-pass bistatic interferometric configuration. Such dataset could motivate advanced SAR technologies, and support scientific applications for future spaceborne SAR missions. However, a direct tomographic reconstruction on TomoSense L-band data presented unwanted artifacts due to: i) uncertainties in provided navigational data, and ii) a potential presence of clock mismatches as no dedicated communication link was employed for time synchronization. A dedicated calibration approach is therefore developed to compensate above disturbances using natural scatterers. Experimental results indicate that our proposed calibration approach is able to remarkably enhance the interferometric and tomographic performances on TomoSense L-band data.

Synthetic Aperture Radar Interferometry (InSAR) technique shows huge potential in the deformation monitoring of coastal seawalls, which is essential to guarantee the safety of people's lives and social property. However, a single view InSAR dataset can only identify satellite-oriented point-like targets (PTs), and the one-dimensional deformation would cause deformation estimation deviation and potential risk undetected. This study developed a multi-view InSAR analysis method to present the first 3D deformation monitoring and structural-level deformation analysis of coastal seawalls. The density and accuracy of selected PTs are improved by a spatial-temporal similarity-based multi-view PTs joint extraction method, and the deformation accuracy is improved through a parallax entropy weighted 3D deformation inversion method. Taking the Donghai Seawall as an example, the 3D deformation of the seawall is revealed, indicating a rapid-slow-steady deformation process.

Synthetic aperture radar tomography (TomoSAR) at different radar wavelength can be used to measure different structural elements of forests. In this letter, we compared the airborne P- and L-band synthetic aperture radar (SAR) tomograms and TomoSAR-measured canopy height model (CHM) and above-ground biomass (AGB) over a tropical forest in Lopé, Gabon. The SAR data sets were acquired by German Aerospace Center (DLR)’s F-SAR system during the AfriSAR2016 campaign. First, the Weighted covariance fitting-based Iterative Spectral Estimator (WISE) was applied to obtain tomograms. CHM was then retrieved based on the canopy phase center derived from the tomograms. Finally, AGB was estimated via an empirical logarithmic model developed from field measurements and the tomographic backscatter power of vegetation layers between 40 and 50 m above ground. Compared with the classical approaches of Capon and Wavelet-based Compressed Sensing, the WISE method can achieve better resolution with higher computational efficiency and reduce the ambiguity level of L-band tomograms successfully. The experimental results also show that there is no substantial difference between P- and L-band TomoCHM, while P-band tomographic intensity is more sensitive than the L-band for the inversion of tropical forest AGB at a resolution of 50 m$\times $50 m.

The new ESA TomoSense campaign aims to explore the retrieval of biophysical quantities over forests for different acquisition geometries and radar parameters. Tomographic SAR acquisitions are currently being carried out using different wavelengths, both monostatic and bistatic systems and opposite views. This work presents the current advances in the analyses and calibration of the TomoSense data stacks to make them suited for scientific analyses. Airborne monostatic P-band acquisitions as received by MetaSensing presented artifacts connected to the acquisition geometry. Coherence and phase fluctuations were compensated thus obtaining clean tomographic reconstructions and a clear identification of the terrain level. Bistatic L-band data are expected to be available in short time as well.

Tomography SAR (TomoSAR) methods recover the 3D structure of targets by processing several SAR images simultaneously. Depending on the degree of approximation, recovering the vertical structure can amount to a 1D problem (processing a vector of pixels), a 2D problem (processing a matrix) or a 3D problem (processing the whole 3D stack at once). The computational burden decreases from the 3D to the 1D, but the constraints for a proper reconstruction are tighter. Hence, this paper discusses the limit and criterion for the feasibility of each method. The huge computational burden of TomoSAR 3D method is addressed by a fast implementation on GPUs. Theoretical analyses and our approach are demonstrated on simulated data, as well as on real data from the ESA AlpTo-moSAR campaign.

Synthetic aperture radar (SAR) images inherently present random and complex spatial patterns, which makes the land-cover classification from SAR images a challenging task. A convolutional neural network (CNN) has been applied to the land-cover classification. However, the statistical properties of an SAR image have not yet been explicitly considered by CNN for feature extraction. To address this problem, this letter presents a statistical CNN (SCNN) for land-cover classification from SAR images, which enables the representation of learning and statistical analysis to be implemented with a unified framework. In the proposed SCNN, the distribution of mid-level primitive features, extracted by representation learning, is characterized by their first- and second-order statistics. These statistics are used to fit the land-cover representations, which encode the statistical properties of the SAR image in the feature space. Experiments on the TerraSAR-X data demonstrate that the SCNN is effective and efficient for the land-cover classification from SAR images.

In this letter, we present an experimental assessment of vegetation height retrieval in tropical forests based on P-band synthetic aperture radar (SAR) acquisitions. Two approaches are implemented and compared: 1) parametric height estimation by minimizing the least-square problem between random volume over ground (RVoG) model predictions and multibaseline SAR data and 2) thresholding the vertical backscattering profiles that are focused by SAR Beam-forming tomography. The data set under analysis is from the ESA AfriSAR campaign that was flown over Gabon in 2016. Results show that at a resolution of 25 m × 25 m, which corresponds to about 80 independent looks, both of the two approaches are able to retrieve forest height to within an accuracy of about 3 m or better over the interval of forest height between 30 and 50 m when compared to Light Detection and Ranging (LiDAR) measurements.

Persistent scatterers (PSs) are coherent measurement points obtained from time series of satellite radar images, which are used to detect and estimate millimeter-scale displacements of the terrain or man-made structures. However, associating these measurement points with specific physical objects is not straightforward, which hampers the exploitation of the full potential of the data. We have investigated the potential for predicting the occurrence and location of PSs using generic 3-D city models and ray-tracing methods, and proposed a methodology to match PSs to the pointlike scatterers predicted using RaySAR, a ray-tracing synthetic aperture radar simulator. We also investigate the impact of the level of detail (LOD) of the city models. For our test area in Rotterdam, we find that 10% and 37% of the PSs detected in a stack of TerraSAR-X data can be matched with point scatterers identified by ray tracing using LOD1 and LOD2 models, respectively. In the LOD1 case, most matched scatterers are at street level while LOD2 allows the identification of many scatterers on the buildings. Over half of the identified scatterers easily correspond to identify double or triple-bounce scatterers. However, a significant fraction corresponds to higher bounce levels, with approximately 25% being fivefold-bounce scatterers.

The primary bottlenecks that hinder the widespread application of Differential Synthetic Aperture Radar Interferometry (DInSAR) technique for bridge monitoring are the difficulties in reverting three-dimensional (3D) deformation of these complex structures and achieving detailed structural risk assessment for end users. To address these challenges, we developed an improved time series InSAR analysis approach that can retrieve 3D deformation of bridges and enable a detailed structural risk assessment. The proposed strategy effectively integrates the observations from multiple SAR sensors to reveal the 3D deformation. Moreover, the selected point-like targets (PTs) are classified and linked to specific bridge features based on their temporal deformation models to make a detailed structural risk assessment. Our results demonstrate the effectiveness of integrating observations from multiple SAR sensors in studying bridge 3D deformation and investigating the structural risk.

Three dimensional displacements are more convenient for people understanding ground processes. However, only azimuth or line of sight (LOS) measurements can be directly extracted from SAR images which restrict further applications, especially for people who are not expertise in radar remote sensing. Aiming at this problem, we proposed an approach of retrieving time series three-dimensional displacements from multi-angular SAR datasets. Firstly, time series displacements in the azimuth and LOS direction can be estimated using traditional methods of SAR interferometry (InSAR) and SAR pixel offset tracking. Then, a fitting and interpolation procedure was applied to parameterize the displacement history and interpolate the displacements from different datasets with identical dates. Thus, three dimensional displacements can be obtained by making use of the different observation geometry of different SAR datasets. Our method was applied to retrieve three dimensional displacements history of the Shuping landslide, Three Gorges area, China.

Forest biomass is a most important parameter in the context of the global carbon cycle. Mapping above ground biomass (AGB) at a global scale contributes to understanding the dynamics of climate change. Tropical forests are extremely important as they store more biomass. In recent years, SAR tomography has been introduced as a new technique that has shown enormous potential in AGB retrieval. A strong linear relationship between in-situ measurements and tomographic power from 30 m above the terrain was discovered by previous studies carried out in French Guiana. However, the two parameters that determine the linear relationship might vary for different tropical forests. Due to the great difficulty in measuring tropical forest AGB by field surveys, in-situ measurements is unfeasible to relate SAR tomography for mapping global forests AGB. For purpose of solving this problem, we investigate the possibility to use LiDAR derived AGB to find the two parameters of the fit line. Experimental results obtained by processing data from the TropiSAR campaign support the feasibility of the proposed concept.

Speckle inherent in synthetic aperture radar (SAR) images usually complicates visual interpretation and brings difficulty to information extraction for applications. Current speckle filters are mainly developed for single SAR image or an image pair (InSAR or PolInSAR). Although some multichannel filters are proposed, they only exploit pixel intensity to identify statistically homogeneous pixels (SHPs). In this paper, we present a new unified approach to filter multitemporal SAR images by adaptively estimating complex covariance matrix-based multitemporal filtering, named CCM-MTF. The key idea is to employ generalized likelihood ratio (GLR) test on the Wishart distributed initial CCM to evaluate the similarity between two pixels. A special design is given to the initial CCM estimation, in which temporal samples are used instead of spatially neighboring samples. Then, a threshold determined by the asymptotic distribution of the logarithm of GLR test statistics at a fixed significance level is used to select spatial SHPs for the reference pixel. Subsequently, the filtering is implemented by estimation of the final CCM from original SAR scattering vector over SHP pixels, and all filtered target information channels including intensity, interferometric phase, and coherence can be explicitly derived from the final CCM. The effectiveness of the proposed CCM-MTF method is validated by experiments on both simulated and real multitemporal SAR images. Both qualitative and quantitative comparisons between CCM-MTF and four state-of-the-art SAR filters are carried out to demonstrate its advantages in terms of speckle suppression as well as detail preservation for all the three information channels.

Feature extraction is an important issue for image interpretation, many valuable feature extraction methods have been proposed to address synthetic aperture radar (SAR) image registration. However, the current methods care little about changes over multi-temporal SAR images, which result in unstable output features. The unstable property is a latent factor to affect the accuracy of SAR image registration. To overcome this problem, we propose a new method to detect stable features by intersecting Coherent Scatters (CS) and SAR-FAST corners. Thus the stable features are not only corners but also located at the time-invariant area. Then, the stable features are described by SIFT descriptors and the coarse registration is achieved by matching these stable points. Finally, the Powell algorithm is used to search the optimal Mutual Information location for precise registration. Experimental results demonstrate the effectiveness of the proposed method.

Unsupervised classification plays an important role in understanding polarimetric synthetic aperture radar (PolSAR) images. One of the typical representations of PolSAR data is in the form of Hermitian positive definite (HPD) covariance matrices. Most algorithms for unsupervised classification using this representation either use statistical distribution models or adopt polarimetric target decompositions. In this paper, we propose an unsupervised classification method by introducing a sparsity-based similarity measure on HPD matrices. Specifically, we first use a novel Riemannian sparse coding scheme for representing each HPD covariance matrix as sparse linear combinations of other HPD matrices, where the sparse reconstruction loss is defined by the Riemannian geodesic distance between HPD matrices. The coefficient vectors generated by this step reflect the neighborhood structure of HPD matrices embedded in the Euclidean space and hence can be used to define a similarity measure. We apply the scheme for PolSAR data, in which we first oversegment the images into superpixels, followed by representing each superpixel by an HPD matrix. These HPD matrices are then sparse coded, and the resulting sparse coefficient vectors are then clustered by spectral clustering using the neighborhood matrix generated by our similarity measure. The experimental results on different fully PolSAR images demonstrate the superior performance of the proposed classification approach against the state-of-the-art approaches.

The orbit of TerraSAR-X is strictly controlled and well known. Based on this information, a very high absolute accuracy in geo-locating TerraSAR-X data is possible and has been demonstrated before. In this manuscript, we present an experimental validation of this, demonstrating the very high level of accuracy achievable using only the atmospheric phase delay information provided in the header files of TerraSAR-X.

Multiple InSAR techniques are increasingly being developed for earth observation. However, among them, persistent scatterers-based InSAR (PSI) techniques fail to obtain enough measurement points (MPs) in rural mountainous area due to the lack of persistent scatterers (PSs), such as ma-made structures, rocks, and outcrops, etc. In this paper, Distributed Scatterers-based InSAR (DS-InSAR) algorithm, exploiting both persistent scatterers and distributed scatterers (DSs) widely spreading in rural areas, is proposed to make up the limitation of persistent scatterers-based technique when monitoring mountainous landslides. There are two key steps to preprocess DSs in DS-InSAR algorithm: selecting DS candidates and estimating optimal interferometric phases. The selected DSs and PSs are combined for further processing using traditional PSI procedure. A qualitative and quantitative simulation analysis was operated to validate the DS-InSAR algorithm. Then, both PSI and DS-InSAR were implemented to monitor Xishan landslide in western mountainous region of Sichuan province, based on high-resolution TerraSAR-X images. The obtained results demonstrate that DS-InSAR could detect much more MPs and provide more reliable deformation information.

Urbanization is an ongoing process that need constant monitoring and observation. In this manuscript we showcase the build-up changes within and around the city of Shanghai by using space-borne microwave remote sensing data. Specifically, we employ a small baseline time series analysis using coherence change detection from two consecutive image stacks from 2003 to 2015 with data from ENVISAT ASAR and TerraSAR-X. Coherence is a measurement of relative phase change and serves as an indication for changes on the surface. We have generated a set of maps that show the spreading of the urban build-up area and its changes within.

Terrain measurement and surface motion estimation are key applications for SAR missions. These applications drive several SAR satellite missions in the Dragon partner countries in Europe and China. In this context, we present our work in the Dragon program on terrain measurements with results from several test sites in China.

High-speed railways play important roles in connecting populated cities. Strict attention to ground subsidence is applied to ensure its transportation safety. Shanghai-Hangzhou high-speed railway is an important part of China's “Four Vertical and Four Horizontal” Passenger Line Network. In this study, TerraSAR-X and ASAR images were employed to recover the surface evolution along the high-speed railway using advanced InSAR technique. A spatial-temporal criterion is updated for PS identification. Nonlinear deformation is modeled and removed by a linear least squares fit. Measurements from TSX and ASAR during a closer period as well as the transverse subsidence profiles showing a fairly consistent agreement. According to the longitudinal subsidence profiles, although the subsidence gradient now is within the guidelines of high-speed railway, the increasing cumulative subsidence can still be a potential threat for the line.

In this paper, a new approach developed to retrieve temporally-differenced maps of the spatial distribution of water vapor from interferometric synthetic aperture radar (InSAR) data with a horizontal resolution as fine as 20 m is described. To get a more accurate map of wet delay, we used the ERA-Interim to predict the dry delay component in the total atmospheric delay from InSAR. In addition, the ERA-Interim was used to compute the conversion factors required to convert the wet delay to precipitable water vapor (PWV). The InSAR-derived differential PWV maps were calibrated by means of the ground-based GPS PWV measurements. The approach was applied to map the temporal change of PWV over the Southern California, USA. We validated our results against measurements of the PWV acquired from a Medium-Resolution Imaging Spectrometer (MERIS) onboard the ENVISAT satellite. The results show strong spatial correlation with values of uncertainty of less than 2 mm.

The radiometric quality of polarimetric SAR (PolSAR)/SAR images is affected by terrain undulations, and the resultant radiometric distortions should be calibrated to facilitate quantitative applications as land cover classification. This paper presents a terrain-related radiometric calibration method to a quad-polarimetric Advanced Land Observing Satellite phased array type L-band synthetic aperture radar (ALOS PALSAR) image. A digital elevation model (DEM) was used for accurate detection of layover and shadow areas. Precise calibration was done subsequently. Polarimetric features were extracted and a supervised random forest (RF) classifier was then employed. Five classes were extracted as waterbody, bare soil, farmland, forest, and man-made objects. Accuracy assessment was performed and the results were analyzed. Improvement of overall accuracy from 78.67% to 82.67% and that of kappa coefficient from 0.73 to 0.78 was achieved using the radiometric-terrain-calibrated (RTC) features, which shows great necessity of RTC processing for PolSAR land cover classification in mountainous areas.

A novel Polarimetric-Texture-Structure descriptor for high-resolution PolSAR image is presented in this paper. More precisely, a PolSAR image is represented by a tree of shapes, each of which is associated with several polarimetric and texture attributes. We first extract the texture properties and polarimetric characteristics from each shape, then use the shape co-occurrence patterns (SCOPs) to characterize the shape relationships, and finally use the resulting SCOPs distributions as features for PolSAR image classification. The proposed method not only has the strong ability to depict the texture and polarimetric properties, but also encodes the shape relationships on the tree. We compare the proposed method with the cluster based statistical feature (CSF) and the scattering mechanism based statistical feature (SMSF). Experimental results on high-resolution PolSAR sample dataset and a large scene for classification demonstrate the effectiveness of the proposed method.

As an advanced technique, synthetic aperture radar (SAR) tomography makes it possible to overcome the layover problem induced by the intrinsic side-looking geometry of SAR sensors. However, traditional nonparametric spectral estimators, e.g., truncated singular value decomposition, are limited by their poor elevation resolution. On the other hand, the compressive-sensing-based approaches using the basis-pursuit strategy to find an L1-norm minimization solution for SAR tomography (TomoSAR) are extremely time consuming. Therefore, a fast and robust tomographic algorithm with super-resolution capability is needed. In this letter, we propose a new approach for TomoSAR based on two-step iterative shrinkage/thresholding (TWIST). TWIST uses a two-step strategy to speed up the L1-norm minimization procedure and can achieve an exceptionally fast convergence speed for TomoSAR. Experimental studies with simulated signals and a spotlight-mode TerraSAR-X data set were carried out to demonstrate the merits of the proposed TWIST approach in terms of robustness, fast convergence speed, and super-resolution capability.

SAR interferometry has long been regarded as an effective tool for wide-area topographic mapping in hilly and mountainous areas. However, quality of InSAR DEM product is usually affected by atmospheric disturbances and decorrelation-induced voids, especially for data acquired in repeat-pass mode. In this paper, we proposed an approach for improved topographic mapping by optimal fusion of multi-orbit InSAR DEMs with correction of atmospheric phase screen (APS). An experimental study with highresolution TerraSAR-X and COSMO-SkyMed datasets covering a mountainous area was carried out to demonstrate the effectiveness of the proposed approach. Validation with a reference DEM of scale 1:50,000 indicated that vertical accuracy of the fused DEM can be better than 5 m.

PolSAR image segmentation has long been an important problem in the PolSAR remote sensing community. Many segmentation algorithms describe images in terms of a hierarchy of regions has attracted particular attention in recent years. However, they often contain more data than is required for an efficient description. In this paper, we propose an effective measure to extract hierarchical semantic structures from PolSAR images. First, we construct the Binary partition tree (BPT) which is a multi-scale image representation to obtain a hierarchy of regions. Once the tree has been constructed, every hierarchy can be considered as a region adjacency graph (RAG). Second, we use a Non-Parametric Graph Entropy as a measure of graph complexity to identify semantic structures within BPT hierarchies. Experimental results on NASA/JPL AIRSAR and DLR E-SAR images demonstrate the effectiveness of the proposed approach.

The conditions in the Altai Mountains make it a difficult area for archaeological on-ground surveys. Automatic surveying could facilitate decision making and planning as well as the building of a comprehensive database of archaeological monuments. We have tried three different approaches towards automatic detection of grave mounds in high-resolution optical data and found an object-class specific Hough forest algorithm the most suitable for the purpose. Through adaption and testing of the algorithm on IKONOS-2 data we created a tool for mapping archaeological features in the Altai Mountains, hence contributing to future steps towards a sustainable cultural heritage management.

In this paper, an algorithm applied for road extraction on SAR image is proposed, which is based on a multi-scale linear feature detector and beamlet framework, and then a quadratic kernel is introduced to offer optimal representation for the circle roads, aiming at improving the extraction quality. Firstly, a multi-scale pyramid is built on the input image and at each level the image is subdivided into a series of dyadic squares that constructs a quadtree. Then the multi-scale linear feature detector and beamlet are employed to compute pixels' responses. Finally, a quadratic kernel for non-linear candidates is introduced and adaptively selects the generating direction of segments. Experiments on TerraSAR images prove that the proposed approach significantly improves the extraction quality and performance when compared to several methods.

Spaceborne D-InSAR system is a kind of SAR satellite system aiming at differential interferometry application. It provides us a capability of measuring subtle deformation on the ground with phase difference. In this work, we focused on the system performance analysis theory as a first step to launch the research work on the overall spaceborne D-InSAR system. After the theoretical model and major error analysis work, we realized that coherence is really important. The qualitative and quantitative analysis showed that the situation becomes worse due to volumetric and temporal decorrelations compared with single-pass InSAR system such as TanDEM-X system. The in-orbit SAR satellite data experiment study validated the quantitative analysis. In order to obtain high-precision D-InSAR measurement, there are some new problems to solve rather than just using an existing SAR satellite to implement D-InSAR.

Modern spaceborne SAR sensors provide geometric resolutions well below one meter. In data of this kind, many features of urban objects become visible. However, because of the intrinsic side-looking geometry of SAR sensors, layover and foreshortening issues inevitably arise, especially in dense urban areas. SAR tomography provides a new way of overcoming these problems by exploiting the back-scattering property for each pixel. However, traditional non-parametric spectral estimators are limited by their poor elevation resolution, which is not comparable to the azimuth and slant-range resolution. In order to improve the estimated elevation resolution, super-resolution techniques, like compressive sensing, are introduced to SAR tomographic processing. In this paper, we analyze the performance of the compressive sensing approach in SAR tomographic analysis. Numerical experiments on simulated signals and real TerraSAR-X spotlight data are given, which demonstrate the robustness and super-resolution power of compressive sensing.

In this paper, a new Attributed Scattering Center(ASC) feature extraction model is proposed. Together with normalization procedure, optimization of amplitude and the length of scattering center feature extraction, we can get a fine estimation of ASC parameter. The image reconstruction experiment demonstrates that with fewer scattering center can we get a satisfied description of SAR image. Moreover, we also do classifaication experiments on TerraSAR-X data base, the result demonstrate that KNN classification method with ASC feature can obtain a better result than GLGM and GMRF. In this way the usage of ACS is exterded.

Offset tracking is an important complement to measure large ground displacements in both azimuth and range dimensions where synthetic aperture radar (SAR) interferometry is unfeasible. Subpixel offsets can be obtained by searching for the cross-correlation peak calculated from the match patches uniformly distributed on two SAR images. However, it has its limitations, including redundant computation and incorrect estimations on decorrelated patches. In this letter, we propose a simple strategy that performs offset tracking on detected point-like targets (PT). We first detect image patches within bright PT by using a sinc-like template from a single SAR image and then perform offset tracking on them to obtain the pixel shifts. Compared with the standard method, the application on the 2010 M 7.2 El Mayor-Cucapah earthquake shows that the proposed PT offset tracking can significantly increase the cross-correlation and thus result in both efficiency and reliability improvements.

Energy is the major limiting factor for high-performance computing. Nowadays, modern Graphics Processing Units (GPU) use less power per floating-point operation. With the CUDA on ARM architecture, a low-power high performance architecture is available and we present first results on using this architecture for fast SAR geocoding and demonstrate the use of fast SAR geo-coding on GPU for a flexible SAR stereo-radargrammetry approach. We demonstrate the speed-up of SAR geo-coding with GPU implementations on an NVIDIA Tesla C2070 and the CARMA DevKit.

This letter proposed a Part-based CFAR Model for object detection of power tower on high-resolution SAR images. Firstly, Part-based Model is used to describe the structure feature of the target, then Compressing Sensing approach is added to reduce the speckle by means of rebuilding background clutter, next, CFAR method is used to extract local shape and scale parameters, at last, Part-based CFAR Model combines these procedures together to form the finally algorithm, not only includes the distribution features, but also considers the structure relationship in the proposed approach. The algorithm is tested on TerraSAR-X data set with the resolution of 1m and 3m. Experiments show that unlike the CFAR method can only gives the high-light points of the targets; Part-based CFAR Model illuminates the target and its local components by plotting the bounding boxes around them.

This paper addresses the problem of change detection in high-resolution multi-temporal synthetic aperture radar (SAR) images (e.g. TerraSAR-X SAR images). Given two images, the proposed method first computes a difference map between them, by taking into account both the spatial and temporal correlations. Change detection is then formulated as a binary (changed/unchanged) segmentation problem of the difference map. A hierarchical Markov model (HMM) is defined on the multi-scale over-segmented regions of the difference map. The change map is finally inferred by relying on the hierarchical marginal posterior mode (HMPM) of the HMM. Experimental results on multi-temporal TerraSAR-X SAR images demonstrate the effectiveness and the reliability of the proposed approach.

This paper presents a frame for classifying polarimetric synthetic aperture radar (PolSAR) data. The frame is based on the combination of wavelet polarization information, textons, and sparse coding. Polarimetric synthesis unites with the discrete wavelet frame to obtain wavelet polarization variance through the calculation of the wavelet variance in the space of polarization states. The K-means cluster algorithm is implemented to cluster the wavelet polarization variance vectors of the training samples for the purpose of constructing a texton dictionary. A patch, in which all the wavelet polarization variance vectors match those in the texton dictionary, is used to obtain a statistical histogram. Sparse coding is applied to describe the histogram feature and generate a new texture feature called sparse coding of a wavelet polarization texton. Finally, support vector machine is used for the classification. All experiments are carried out on five sets of PolSAR data. The experimental results confirm that the proposed method effectively classifies PolSAR data.

Using high-resolution SAR data stacks, a large amount of persistent scatterers (PS) can be found in urban areas, which are used for very detailed surface deformation monitoring. However, in dense urban areas, many of these points consist of more than one dominant scatterer in elevation direction. Using tomographic SAR, points with more than one scatterer are detected.

This paper, we propose a novel over-segmentation method Feature Geometry Space Fusion (FGSF) for polarimetric SAR (POLSAR) data classification, which uses the polarimetric feature and geometric feature. In order to exam its performance, experiments on the data acquired by AIRSAR show that the over segment regions segmented by FGSF method performs better than meanshift when used for classification.

Persistent Scatterer Interferometry (PS-InSAR) is applied to derive displacement information with millimetric precision. Analyzing stable persistent scatterers from a large stack of SAR images,helps to overcome the geometrical and temporal decorrelation, which occur when using differential interferometry. The removal of the topographic phase with an external DEM seems to cause problems. In our experiment, we select three different DEMs: ASTER GDEM, a DEM derived from a digitized topographic map, and SRTM-3 in order to analyze the influence of the input DEMs for PS-InSAR processing in the Three Gorges area. We find that differential interferogram generation is related to the topographic influence for the PS-InSAR processing and different DEMs get us different PS-InSAR results.

In this paper, the algorithm of K-SVD learning dictionary is applied to target decomposition for polarimetric SAR (PolSAR) images. This algorithm can obtain a set of bases self-adaptively according to the data on each channel of PolSAR, to make polarimetric data become more differentiated on this set of bases. Experiments on the data acquired through polarization SAR equipment developed by China for the first time show that features decomposed through K-SVD algorithm perform better than features based on the physical mechanism of PolSAR when used for classification.

This paper presents a novel approach for super-resolution (SR) reconstruction in Synthetic Aperture Radar (SAR), based on multi-dictionary. In comparison with conventional SR via sparse representation, the algorithm combines the classification with sparse representation. After classifying the training image, we jointly train the low and high resolution dictionaries for each class. And then, the image patches are reconstructed according to different dictionaries, which are chosen in conformity with the class of the image patches. The effectiveness of this method is demonstrated on Terra-SAR datasets.

The RPC (Rational Polynomial Coefficient) model can be used as a replacement sensor model for geo-coding spaceborne SAR data. A hybrid method, combining the L-curve and the IMCCV (Iteration method by correcting characteristic value) method, for solving ill-conditioned equations in the RPC model is proposed. The hybrid method can get higher accuracy at low cost of calculation time. Based on an example in Malaysia, the application for fast RPC geocoding for stereo radargrammetry is shown.

A method for building height determination via a deterministic approach is tested using three TerraSAR-X images from Barcelona, Spain. Using this method, the height of a building wall can be determined based on the strength of the double-bounce backscattering. The approach requires knowledge of the material properties of the measured building wall and the area in near range of the building wall. For certain test buildings the approach provides good results, while for other buildings the results are erroneous.

The phase coherence in synthetic aperture radar interferometry is often used in classification algorithms to detect possible temporal changes of the imaged terrain. However, in mountain areas, the interferometric coherence is also sensitive to the slight variations of the acquisition geometry. In this letter, we propose a very simple but effective method to separate the temporal decorrelation from the geometrical one. Assuming the imaged terrain can be modeled as a distributed target, the geometrical coherence can be estimated by exploiting a topographic model and the sensor acquisition parameters. The discrepancy between the geometrical coherence and the observed one can then be ascribed to temporal changes. Moreover, in presence of pointlike targets, the hypothesis of distributed terrain is no longer valid, and higher values of the observed coherence with respect to the synthetic geometrical one can be used to detect such targets. The proposed approach allows then in mountain areas the following conditions: (1) a simple and very fast rough estimation of the temporal coherence, and (2) the identification of pointlike targets using just two images. The method has been applied and tested in the Badong (China) site using European Remote Sensing satellite tandem data.

SAR image from sea can constantly contain ships and their ambiguities in azimuth and range directions. For maritime applications, the ambiguities are visible due to their strong intensities in a low backscattering background of sea environment. Thus, the ambiguities can be often mistaken as ships and cause false alarms. Many approaches have been proposed for reducing the azimuth ambiguities in single channel SAR image. This paper analyzed scattering mechanisms of the azimuth ambiguities for PolSAR images and proposed a method for detecting ships from PolSAR images. By using eigenvector-eigenvalue decomposition, three eigenvalues can be used to differentiate ship targets, azimuth ambiguities and sea clutter. One C-band JPL AIRSAR polarimetric data have been chosen to evaluate the method. The experimental results show that the proposed method can effectively reduce false alarms caused by the azimuth ambiguities.

After the Three Gorges Dam began to function in China in 2003, the water level of the Yangtze River in Three Gorges area rose more than 100 meters. The impact of the man-made reservoir caused by the dam on the surroundings is becoming the object of several studies. In this paper we make use of two long term D-InSAR techniques, the quasi-permanent scatterer technique (QPS) and the Stanford Method for Persistent Scatterer (StaMPS), to measure the deformation trends in Badong, Three Gorges area, China. The results obtained by the two processing tools with the same focused and co-registered data sets are analyzed and compared. Two subsidence areas are identified by both techniques. However, since the QPS analysis is able to process partially coherent targets, many more points are extracted than in StaMPS, and more information can be retrieved.

An innovative constant false alarm rate (CFAR) algorithm was studied for ship detection using synthetic aperture radar (SAR) images of the sea. Two advances were achieved. An alpha-stable distribution rather than a traditional Weibull or$K$-distribution was used to model the distribution of sea clutter. The distribution of sea clutter in a SAR image was typically heterogeneous, caused mainly by variable wind and current conditions. Image segmentation was carried out to improve the homogeneity of the distribution in each subimage or region. In comparison with ship detection using the CFAR algorithms based on the Weibull or$K$-distribution, our algorithm detected the most number of ships with the smallest number of false alarms.

A new approach is presented in this paper to produce Digital Elevation Model (DEM) in mountainous areas with steep slope using ERS-1/2 tandem data. In order to reduce the impact of phase errors on the Interferometric Synthetic Aperture Radar (InSAR)-generated DEM, an external DEM such as that from Shuttle Radar Topography Mission (SRTM) is utilized in this approach. The proposed algorithm includes two steps: The first step is to model and remove phase trends with a linear regression analysis before converting phase to height; the second step is to filter unreliable height points before interpolating the DEM from the InSAR height map. The critical points are the following: 1) determining the one-to-one correspondence between the interferogram and the SRTM DEM before knowing the InSAR-derived elevation values and 2) estimating the elevation range of every pixel from SRTM DEM. To solve the first problem, an iteratively geocoding algorithm is performed. A DEM interpolation error model solves the second one. For InSAR data processing, the SRTM DEM is not only usable for modeling systematic phase errors but also for filtering gross height errors. The experiments in Zhangbei and the Three Gorges areas in China show that our approach has improved the accuracy of the resulting DEMs significantly without any ground control points.

D-InSAR technique has been widely adopted to monitor land subsidence caused by withdrawal of water, oil, gas, and other minerals. Though many cities in China have seriously suffered from land subsidence caused by ground water over-extracting, few of them have enough money to do the leveling control of the subsidence. Compared with leveling and GPS surveying, D-InSAR is more cost-efficient and precise. As Tianjin city has scores of ERS-1/2 SAR data, thousands of valuable leveling data, smooth topography and severe subsidence, it is an ideal place to test and analyze D-lnSAR technique. However, D-InSAR is liable to be contaminated by atmosphere delay, temporal decorrelation and baseline errors. As these errors cannot be removed by SAR data processing, some auxiliary data such as leveling data and DEM data have been introduced into D-InSAR data processing. Therefore, This work accurately analyzes the features of atmosphere delay and temporal decorrelation with the auxiliary data in Tianjin urban area. The results demonstrate that D-InSAR can detect subsidence within three months. Further actions will be undertaken to improve the D-InSAR subsidence monitoring system, such as using Envisat data, gathering GPS zenith delay data to eliminate the atmosphere delay, and so on. The ultimate interest of our research is to establish a robust, cost efficient city subsidence monitoring system by using D-InSAR technique and other measurement.