After the acceptance of two small missions aimed at astrophysics and space communications, the Italian Space Agency is looking forward scheduling a third small mission, devoted to Earth observation. In this ambit a proposal for bistatic synthetic aperture radar (SAR) observations has been founded for a feasibility study. The proposed mission (BISSAT) concept consists in flying a passive SAR on board a small satellite, which observes the area illuminated by an active SAR, operating on an already existing large platform. Several scientific applications of bistatic measurements can be envisaged, such as : evaluation of bistatic radar cross section; improvement of image classification and pattern recognition procedures; acquisition of terrain elevation and slope; acquisition of velocity measurements; interferometric surveys upon completion of the nominal bistatic mission and successful baseline control; stereogrammatic applications. This paper presents these applications within the BISSAT mission, with particular reference to interferometric ones, showing the requirements on mission operation and discussing the expected results and performance.
This paper compares two different approaches for designing airborne SAR systems. The first one is the most common where conventional processing is employed, and therefore wide antenna beams are to be used in order to avoid ambiguities in the final image due to attitude variations. A second approach is proposed to lower the requirements such system imposes based on subaperture processing. The idea is to follow the azimuth variations of the Doppler centroid, without increasing the hardware requirements of the system. As it is shown in this paper, this processing procedure must be complemented with precise radiometric corrections, because the platform may experience small attitude variations, which could increase/decrease the target observation time, inducing a significant azimuth modulation in the final image. This leads to the definition of a new criterion concerning maximum attitude deviations for an airborne platform.
This paper analyses eight different remote sensing campaigns carried out from 1998 to 2001, pointing out the backscattering coefficients behaviour in dependence both to soil moisture and roughness. Our study indicates a clear dependence of backscattering coefficients on soil moisture with an average sensitivity of 0.25 dB/gr/cm3. In a subsequent step these data sets are utilised to validate an inversion procedure based on a Bayesian algorithm aimed at extracting soil moisture information from backscattering coefficients. After a first run, a priori soil moisture information deriving from the simulation of a hydrological model is introduced leading to an improvement both in extracted soil moisture values and in their uncertainties.
A three dimensional (3-D) realistic radar simulation package including imaging radar simulation concept applied to multisensor scenarios is under development as a project between the Electromagnetism and Radar Department of ONERA and the OKTAL SE Company. Taking advantage of various studies in the domain, this partnership associates the expertise of ONERA in radar phenomenology, wave interaction with targets and clutter, with that of OKTAL SE in the generation and management of realistic scene databases in the infrared and optical domains using advanced Shooting and Bouncing Rays (SBR) techniques. The objective of this program is to develop simulation tools capable of predicting the behaviour of sensors in a realistic environment. This is achieved by coupling a terrain database completed by radar and optical features and a fast SBR algorithm. This paper is focused on the specification of the radar (i.e. electromagnetic wave interaction) principles. Outputs from the simulations illustrate the effectiveness of the tool in respectively, Synthetic Aperture Radar (SAR) simulations; and in the multisensor evaluation context in airborne applications such as enhanced vision for all weather landing assistance.
This paper presents two novel approaches to speckle reduction in SAR images. The former relies on the multiplicative speckle model as an MMSE filtering performed in the wavelet domain by means of an adaptive shrinkage of the detail coefficients of an undecimated decomposition. Each coefficient is shrunk by the variance ratio of the noise-free coefficient to the noisy one. All the above quantities are analytically calculated from the speckled image, the noise variance, and the wavelet filters only, without resorting to any model to describe the underlying backscatter. Estimation of the local statistics driving the filter is expedited and layered processing allows to extend adaptivity also across the spatial scale. The latter is not model-based and provides a blind estimation of the backscatter underlying the speckled image stated as a problem of matching pursuits. The local adaptive MMSE estimator is obtained as an expansion in series of a finite number of prototype estimators, fitting the spatial features of the different statistical classes encountered, e.g., edges and textures. Such estimators are calculated in a fuzzy fashion through an automatic training procedure. The space-varying coefficients of the expansion are stated as degrees of fuzzy membership of a pixel to each of the estimators. A thorough performance comparison is carried out with the Gamma-MAP filter and with the Rational Laplacian Pyramid (RLP) filter, recently introduced by three of the authors. On simulated speckled images both the proposed filters gain almost 3 dB SNR with respect to conventional local-statistics (Lee/Kuan) filtering. Experiments carried out on widespread test SAR images and on a speckled mosaic image, comprising synthetic shapes, textures, and details from true SAR images, demonstrate that the visual quality of the results is excellent in terms of both background smoothing and preservation of edge sharpness, textures, and point targets. The absence of decimation in the wavelet decomposition avoids the typical ringing impairments produced by critically-sampled wavelet-based denoising.
It is well known that a SAR image is composed of two types of information: amplitude and phase. Nevertheless, the information contained in the phase is hardly exploited on its own. Indeed, the number of processes at work and the scale difference between the image resolution and the wavelength induce, with regard to the phase, a quasi-random spatial behavior. However, our recent work shows that the phase of one image can be spatially correlated. First, we define an estimator for the spatial correlation of the phase, and study its behavior with real data. We assess the phase correlation according to the resolution and the type of surface. Then, we set down the theoretical bases of a statistical model of this behavior. We highlight the conditions required with regard to the resolution, the sampling rate, and the impulse response. We therefore identify the best kinds of surfaces, so that the phenomenon occurs. Hence, we simulate the phase correlation for different cases according to the phase model defined. We choose suitable parameters to the conditions of the real data and compare measurements and simulations. Finally, we propose possible applications related to the use of this new source of information.
The upcoming launches of new satellites like ALOS, Envisat, Radarsat2 and ECHO will pose a significant challenge for many ground stations, namely to integrate new SAR processing software into their existing systems. Vexcel Corporation in Boulder, Colorado, has built a SAR processing system, named APEX -Suite, for spaceborne SAR satellites that can easily be expanded for the next generation of SAR satellites. APEX-Suite includes an auto-satellite-detecting Level 0 Processor that includes bit-error correction, data quality characterization, and as a unique feature, a sophisticated and very accurate Doppler centroid estimator. The Level 1 processing is divided into the strip mode processor FOCUST, based on the well-proven range-Doppler algorithm, and the SWATHT ScanSAR processor that uses the Chirp Z Trans-form algorithm. A high-accuracy ortho-rectification processor produces systematic and precision corrected Level 2 SAR image pro ducts. The PALSAR instrument is an L-band SAR with multiple fine and standard resolution beams in strip mode, and several wide-swath ScanSAR modes. We will address the adaptation process of Vexcel's APEX-Suite processing system for the PALSAR sensor and discuss image quality characteristics based on processed simulated point target phase history data.
A fundamental pre-cursor to synthetic aperture radar (SAR)interpretation is the segmentation of the image into statistically homogeneous regions for which very reliable algorithms are now available. The aim of the work reported in this paper has been to build on the initial SAR segmentation to produce a low-level description of the SAR scene and then to demonstrate the use of high-level processing applied to the low-level components. To this end, feature-based classification of segments into different terrain types has been implemented. Furthermore, algorithms for linear feature detection and classification have been developed. These use measures of length and thinness to find candidate starting segments from which networks of potential lines are grown using a Kalman filter to identify potential extensions to the current line whilst also providing a measure of confidence for the detected line. Once the image constituents have been identified with associated degrees of confidence, Bayesian techniques can be used to exploit prior contextual information. This is demonstrated with respect to the target detection application for which prior probabilities are introduced given terrain type, hedge proximity and proximity of other targets. It is shown how enhanced target detection can be obtained by utilising this contextual information in a rigorous statistical framework.
We present results of large-scale experimentation with the enhanced model-based despeckling (EMBD) filter aiming at its validation from a pragmatical point of view. Furthermore, we point out criteria for the choice of a prior model for synthetic aperture radar (SAR) images. These criteria rely on an evidence maximization step -part of the EMBD algorithm itself- and on the verification of the obtained speckle statistics against the assumptions made in the filtering.
The aim of this article is to explore new methods to enhance the results of automatic interpretation of SAR images by combining images acquired from different viewing directions (multi-aspect SAR images). Using the combined information extracted from multi-aspect images allows to resolve problems of obscurance, by for instance the borders of a forest, to increase the resolution and to augment the confidence in detection as compared to detection in single images. The article focuses on high-resolution polarimetric images for the automatic interpretation of an airfield scene. Specifically for this type of images we have developed a set of new image interpretation tools such as edge detectors and bar (line) detectors, both based on multi-variate statistics. These detectors are briefly described in the article. The main part of the proposed article will focus on how the use of multi-aspect images can enhance the results of these detectors. The multi-aspect images are supposed to be accurately registered. It is thus possible to warp them into a common coordinate system. Because the spatial resolution of a SAR system is usually not the same in range and azimuth, it is sometimes better to detect objects in each image separately and fuse the results of the detection at the object level. This is particularly true for the detection and delimitation of the buildings. On the other hand, edge detectors can benefit from combined information on a pixel-level. In particular edge detectors based on multi-variate statistical methods can be applied on registered images, thus increasing the confidence level of detection and reducing the false alarm rate, by combining the information at a low level. For edge detectors we will compare results of combining the information available from multi-aspect polarimetric images at different levels. In particular we will compare the results of applying them directly to the registered image set with these obtained when applying them on each individual image and fusing the results at the object level or intermediate (edge-strength) level. Similar investigations will be presented for the bar detectors. Results will be shown on a set of polarimetric L-band images of an airfield.
This work presents a method for detection, localization, classification and pose estimation of objects in SAR-image sequences. Such methods have to deal with strong noise in SAR-images and have the challenge that shadows, which may occur, should not affect the recognition process. The disturbing effect of noise is significantly reduced in the presented method by temporal integration of the SAR-images, using a motion-model of the sensor. Thus it is possible to perform a segmentation on the integrated images with quantile-thresholds and a region growing algorithm using an edge image created by a Canny-edge detector. To be independent of the number of objects in the image and the brightness of the image, a multi-threshold approach is used. By accumulating the segmented images, following an analysis of the homogeneity of the accumulated segments, it is possible to identify stable segments as possible objects. An optimization process is used to fit a generic model of a house into the stable segments. As initial values for the optimization process the results of a connected-pixel algorithm are used. An application example is presented, in which house-objects can be separated from shadows in a village formation and their pose can be determined correctly.
This paper deals with automatic extraction of 3-D buildings from stereoscopic high-resolution images recorded by the RAMSES sensor. Roofs are not very textured whereas typical strong L-shaped echoes are visible. These returns generally result from dihedral corners between ground and structures. They provide a part of the building footprints and the ground altitude, but not building heights. Thus, we present an adapted process including two main steps: - stereoscopic structure extraction from L-shaped echoes. Buildings are detected on each image using the Hough transform. Then they are recognised during a stereoscopic refinement stage based on a criterion minimisation. - height measurement. As most of previous extracted footprints indicate the ground altitude, building heights are found by monoscopic and stereoscopic measures. Between structures, ground altitudes are obtained by a dense matching process. This processing is applied on an industrial scene. Results are compared with a ground truth. Advantages and limitations of the method are brought out.
SAR (Synthetic Aperture Radar) image matching is the most critical step in a radargrammetric chain. The presence of speckle noise makes classical methods (used in optics) inefficient, and some specific techniques must be developed. In this paper, we firstly present the Alcatel's radargrammetric chain : the two images are first resampled in epipolar geometry, to reduce the space search of homologous points, then the matching results are optimized using a TABU search with a regularity constraint. Having the disparity map, we obtain the 3D information by a space triangulation. We also generate a confidence coefficient to determine the most robust disparities. This radargrammetric chain gives encouraging results, but the matching step still raises problems : the area based matching method suffers from speckle, and processing time is considerably increased by the optimization method. So we present secondly the new matching module we are working on, integrating feature based methods. To detect edges we use the ROEWA operator (Ratio Of Exponentially Weighted Average) which is well adapted to SAR images. Edges can be extracted by different ways : watershed algorithm or maximum tracking. The objective is to find the most robust edges in both stereoscopic images, in order to match them and create a first set of matched couples. This will guide our search in the generation of a dense disparity map. We finally propose the global SAR image matching module including edge extraction and matching, radiometric correlation, and eventually user control, to generate an accurate disparity map.
The application of interferometric techniques to polarimetric SAR data is a relative new and promising research field. Noteworthy examples of its potential have been reported by Cloude and Papathanassiou for the retrieval of forests height. In general, polarimetric analysis approaches have been considered for optimizing the interferometric coherence, mainly in order to improve the generation of digital elevation models. In this paper, we will present the first results that we obtained by combining interferometric analysis with coherent target decomposition methods (in particular, the one proposed by Krogager); the different coherence properties of target models will be investigated and a provisional evaluation of the usefulness of this approach will be given.
A several kilometres thick sequence of mostly marine salt with inter-bedded gypsum, shale and dolomite rock of Pliocene to Pleistocene age build several salt diapirs in the Dead Sea area. The Lisan Peninsula salt diapir is elongated in the N-S direction, and includes several sub-domes and a structural depression. Differential interferograms were generated for several time intervals of seven to ninety three months between 1992 and 1999 and show a large diversity of uplift and subsidence features in the peninsula. The uplift rate, which has been measured, is in correspondence to the geological rate evaluated by other geological researches. The subsidence, mainly in the south dome and the cape are much more significant. Inversion deformation in the cape between the year 1995-1996 suggested to be linked to the 22 November 1995 Nuweiba earthquake. This paper suggested a tectonic mechanism connecting the salt deformation in the Lisan Peninsula with the activity of Boqeq fault.
The reconstruction of urban structures from InSAR (Interferometric Synthetic Aperture Radar) observations is a complex task. Until now it has been tipically approached using the methods of radargrammetry and SAR interferometry, in a direct extension of what had been done in the past for the reconstruction of natural surfaces from generally much lower resolution data. In this work, we present a new concept aiming at the accurate and detailed reconstruction of the observed city scenes for metric SAR observations. We use a model-based approach for the synergetic analysis of the different sources of information in InSAR data. We define a hierarchical model of the InSAR observation that is both deterministic and stochastic. While the deterministic section describes the SAR imaging geometry and its effects and expresses the different scene structures, the stochastic part incapsulates instead prior knowledge about the signal and defines its attributes while also describing incertitude over the parameters of the geometrical model. Bayesian inference is used to couple the diffent levels of the model, and to further define parameter estimation algorithms.
In this paper we investigate a least square algorithm to retrieve forest parameters from interferometric, fully polarimetric radar remote sensing P-band data, based on an interferometric optimization scheme which is applied to maximize the separation of scattering phase centers related to the pertinent interferometric coherences in order to obtain most accurate parameter inversion results. A recently developed approach is especially adopted to airborne P-band data, introducing a least square minimization scheme in which synthetic interferometric coherences computed from a scattering model, which is based on a randomly oriented volume over a non-penetrable ground, are compared with three interferometric coherences measured by the P-band sensor. Through fitting of the synthetic and the measured interferometric coherences, the pertinent candidate parameters of the optimization problem can be retrieved. These parameters are the forest volume thickness, the volume extinction coefficient, the interferometric phase related to the underlying topography, and the effective ground-to-volume amplitude ratios related to the interferometric coherences. Through a weighted superposition of all the interferometric coherences provided by the fully polarimetric radar sensor these coherences can be maximized and introduced as the right-hand side of the parameter optimization problem. Experimental results obtained from P-band fully polarimetric single baseline interferometric data acquired over the amazon rain forest are shown in order to demonstrate the potential of the proposed approach. Furthermore, a (chi) 2-test is performed on the data to prove the validity of the introduced scattering model for rain forest vegetation.
SAR-mapping is usually performed along a straight path. A curved path might increase the mapping rate significantly. Drawbacks are more complex signal processing and that defocusing may occur. In this paper, curved SAR-mapping is analysed in more detail including more forward look geometry. Relationships are derived how SAR-resolution and mapping rate are influenced by the curved SAR-path. Examples show how the non-linear phase error depends on side acceleration and scene geometry.
In the analysis of SAR, there is a need of simulating complex scenes without having measured radar raw data available. The common method is based on a set of point reflectors, which model the response from an extended target. The received I,Q-signals along the SAR aperture are computed, and the effects of internal noise and uncompensated phase errors are easily included. This method becomes extremely cumbersome for complex scenes with radar response from each resolution cell. In this paper, a much more rapid method of raw data generation is applied. The complex numbers representing the SAR image are then converted to radar raw data using the SAR-processing algorithms backwards. The key transform is the scene centre focusing, which ties together SAR data and sensor geometry. This method of generating raw data is several orders of magnitude faster than the direct method described above. The effects of internal noise and phase errors, due to imperfect motion compensation, can easily be included. The method can be applied as well to simulation studies of auto-focusing, DPCA and STAP.
This paper presents a new near-field 3-D synthetic aperture radar (SAR) imaging algorithm. This algorithm is an extension of the 2-D chirp scaling algorithm (CSA). First, the original formulation of the CSA has been extended to the 3-D case. Then, some processing steps have been reformulated in order to incorporate additional terms (up to n-th order) in the approximations assumed by the algorithm. These extra terms are required to maintain the accuracy of the method in the so-called near-field conditions (large coherent integration angle and/or high bandwidth-to-center-frequency ratio), but do not entail a significant increase of the computation time. The algorithm has been also optimized for stepped-frequency radars. The performance of the method has been illustrated with numerical simulations and experiments.
A model for the signal returns from terrain features for a Synthetic Aperture Radar is developed. A standard range-azimuth geometry is used to divide the surface area into cells for which clutter returns are generated. This geometry is extended to a spherical Earth to introduce effects due to varying terrain height and slope, and to determine areas of shadow. The calculation of phase variation in the returned pulses at the IQ level required to successfully form a SAR is discussed. The technique is illustrated with an example which demonstrates the effect of look angle on SAR images.
The signal to clutter ration of targets in Synthetic Aperture Radar (SAR) images is a function of the resolution cell and the backscattered radar cross section (RCS) of the target where both the resolution cell and the RCS of the moving target are a function of the time. Due to the long integration time both the SAR look angle and the spatial reflection pattern vary during the integration. Hence, there exist a complex interaction between the spatial reflection pattern, the temporal variation of the moving target, the long integration time and the varying platform look angle. This complex interaction gives an upper and lower limit to the size of simple scatterers such as flat plates. An attempt has been made to try and assess the size and geometry of typical main scatterers of a generic maritime target in order to predict the imaging capabilities of the soon to be launched ENVISAT SAR.
The recognition and classification of urban structures from SAR observations is a particularly complex task. In this article we present a new concept aiming at the accurate and detailed classification of the city scenes observed with metric resolution SAR sensors. SAR images of build-up areas at resolution of 2-3 meters are characterized by strong patterns induced by the geometry of buildings and the phenomenology of scattering of the radar signals. Thus, resulting in high complexity images. The accuracy of image interpretation relies on the descriptive power of the low level image information extraction. The article presents a method based on the Bayesian concepts. A hierachical 3 layers model is used for the SAR observations. The first layer describes the speckle effect as a Gamma distribution, the second, the cross-section, is modeled as Gibbs Random Field (GRF), the third layer the parameters of the Gibbs random field is considered a Jeffrey's prior. The GRF describes the cross-section structures induced by the geometry of the buildings. The model is non-stationary, its parameters adapt locally to the image structures.