To investigate the potentiality of a radar ground prediction radar satellite data (SAR) have been studied for the generation of clutter estimation models. The generation of clutter statistics as well as scrutinized analysis of specific dependencies and influences of backscatter processes and behaviour is the basis for statistical clutter prediction. Clutter predition makes an adjustment to the present situation necessary. This demands for certain calibration techniques, e.g. online-calibration.
Radar raw data from the orbital SAR-systems X-SAR, Radarsat and ERS-2 found the data base for radar backscatter calculation. In combination with a digital elevation model and the ground truth of the investigated area backscatter maps representing the backscatter coefficient have been processes. The ground truth categories have to be selected in a way that each category is linked to a certain homogeneous backscatter category. Each backscatter category is represented by its probability density function or distribution function, respectively. In case of an elaborated ground truth concerning geometrical dimensions, location, invariance and homogeneity the backscatter statistics will be stable.
Parametrical dependencies of different numerous backscatter processes have been proved by theoretical considerations and practical analysis of existing satellite radar data. It has been shown that backscatter behaviour is due to statistical regularities and laws that make a clutter prediction feasible and reliable.
In recent years a new class of Synthetic Aperture Radar (SAR) systems, using low frequencies, have emerged. The combination of low frequencies with high bandwidths allows a variety of new applications. Several new fields arise in forestry, biomass estimation and in archaeological and geological exploration. The P-band SAR technology benefits from technological advances in antenna design, low noise amplifiers, band pass filters, digital receiver technology, as well as new processing algorithms.
For all the new applications of an airborne P-band SAR system, the high-resolution imaging is an important parameter, but it cannot be easily achieved with conventional processing techniques. In this paper, the performance and limitations of the Extended Chirp Scaling (ECS) algorithm and wavenumber domain Omega-K processing algorithm are analysed and discussed. Additionally, modifications of both algorithms are proposed, which optimise the respective algorithm for processing low frequency, wide-beam and wide-band SAR data. Despite of the inherent limitations of the above mentioned processing algorithms, a deterministic phase error, called "digital phase error", due to digital signal processing characteristics is formulated and its effect to the processed SAR data is analytically described. The analysis is carried out, using simulated low frequency airborne SAR data.
Synthetic aperture radar (SAR) processors generally assume that the scene reflectors are stationary over the time of integration of the image. When this is not the case, various kinds of image distortions may occur, such as target displacement or smearing. In the present paper, oscillating targets are considered. It is shown that such targets are smeared in azimuth to an extent determined by the amplitude and frequency of the oscillation. The reason may be stated in terms of the slow-time Doppler shift of the target. The Doppler shift is not constant, but varies with aperture time. We show that time-frequency analysis provides useful tools to handle problems of this kind. The choice of analysis method is often difficult. Here, we compare several methods of the Cohen's class, and show good results with the data adaptive optimal kernel method. This method, being adaptive, dispenses with some of the trial-and-error often necessary with quadratic methods. We show data from a controlled experiment where oscillating reflectors were placed within a scene imaged by an airborne SAR system. The reflectors are smeared in azimuth. We estimate the amplitude and frequency of the oscillations from the time-frequency distributions.
This paper presents a new concept for Time-Frequency estimation, which is based on algorithmic fusion. It is shown that algorithmic fusion increases considerably the detectability of signals while suppresses artifacts and noise. The paper reviews a sample of representative Time-Frequency algorithms. Their performance is studied from a qualitative and quantitative point of view. For simplicity, we have considered the Mean-Squared Error (MSE) as a measure of performance in quantitative performance evaluation studies. The algorithmic fusion is presented using a self adaptive signal and noise dependent or independent approach, while the fusion is performed using the first two terms of the Volterra Series expansion. Simplistic algorithmic fusion methods on time-frequency results (e.g. weighted averaging or weighted multiplication), are special cases of the proposed fusion technique. Experimental results are presented from simulated and real High Resolution (HR)-SAR data. Real HR-SAR data were used from the experiments performed by the Defence Research Establishment (DRDC)-Ottawa.
Usually dedicated to ground imaging, Spot SAR processing can also be applied to moving targets, when proper autofocusing is used. In this case, the simultaneous motions of carrier and target both have an impact on the azimuth resolution. That is why these motions have to be estimated and their effects corrected. The process is then a joint SAR-Inverse SAR (ISAR) imaging. For ship targets, the typical roll, pitch and yaw motions (usually unknown for real, non co-operative targets) produce some residual uncorrected migrations of scatterers. The consequence of these migrations is a blurring of SAR/ISAR range-Doppler maps of the target.
A "snapshot" technique, based on a short imaging time, allows some robustness to these residual migrations during imaging time, but it has two main disadvantages. First, it makes it difficult to achieve high azimuth resolution. Second, it produces a series of range-Doppler maps, which include both useful and unsuitable images for extracting the outline of the ship target. The interest of a particular image depends on the moment in the unknown rotation of the target.
We developed a criterion enabling us to choose suitable snapshots in a series, and also a segmentation technique adapted to the typical shapes of ship targets. This criterion can be adapted to range undersampled data and the presumed Doppler spread of the target return. Its complexity and accuracy may therefore be adapted to the context. The criterion and the segmentation technique have both been tested on synthetic and real in-flight data. A highly effective way of producing easy recognizable height profiles of ship targets at sea has been achieved.
In some practical situations it is often necessary to process radar images for which a priori information about noise characteristics is limited. Evaluating these characteristics, in particular, estimating speckle noise variance, is quite a complicated and time consuming task. In order to carry out both tasks efficiently enough for typical practical situations an automated robust procedure for SAR image filtering and preliminary analysis is proposed. It consists of several stages: a) blind evaluation of speckle relative variance for the original image, b) pre-processing using the local statistic Lee filter, c) blind evaluation of the residual noise relative variance for the pre-processed image, d) post-filtering. It is shown that the proposed procedure provides rather accurate estimations of noise characteristics. The effectiveness of the filtering scheme is confirmed for both simulated and real scene SAR images.
The subject of this communication is the detection of roads on SAR images starting from a user-drawn graphical sketch. This sketch is considered as a road model, and the main contribution of this communication is to incorporate this model in a yet-existing blind road detection method to improve the quality and reliability of the detection. The method combines both local and global criteria for raod extraction. It consists of mainly two parts. First local information is extracted by applying a low-level operator, based on a line detector taking to account the statistical properties of speckle in SAR images. In the second step, road detection is performed by selecting the best road in a graph composed of all the detected segments and all possible connections between them. To demonstrate the contribution of the use of a model in the road detection, we first introduce an "exact" model of hte road (extracted for instance from a road map), in order to validate the method. Then the method is applied with an un-precise model (both in form and position). The obtained results from different SAR images are presented and evaluated with objective criteria.
Bayesian inference has been proved to be a valuable tool in inversion processes. In this study, it is applied in two cases, both aimed at estimating dielectric constant from radar measurements.
The first case is devoted to merge point measurements deriving from radiometer and scatterometer data on bare soils. The second case uses, in the inversion process, only active scatterometer data, but introduces supplementary information from the simulations of a hydrological model.
In this Bayesian algorithm the key point is the evaluation of a joint probability density function based on the knowledge of data sets consisting of soil parameters measurements and the corresponding remote sensed data. It is obtained by applying the maximum likelihood procedure.
As a further step, the influence of prior information about roughness is assessed within the context of the dielectric constant retrieval. At the beginning, a prior uniform distribution is assumed for all surface parameters. Subsequently, a non-uniform prior distribution, based on field measurements, is introduced in order to verify its impact on the estimates and the relative errors.
The sensitivity of microwave scattering to the dielectric properties and the geometrical structure of bare soil surfaces makes radar remote sensing a challenge for a wide range of environmental issues related to the condition of natural surfaces. Polarimetry plays an important role as it allows a direct or indirect separation of surface parameters, namely the volumetric soil moisture content and the geometrical surface properties. The aim of this study is to investigate two different distributions used for the description of rough surfaces in order to extend the validity range of the Small Perturbation Model. The first approach considered a uniform distribution of angles corresponding to a certain surface roughness range, taking only the rms height into account. The still opening question is the influence of the surface correlation length on the quantitative estimation of surface parameters which will be addressed in this study. For the investigation we chose a Gaussian and a Fractional Brownian Motion distribution which are accounting for the surface correlation length.
It is well known that C Band SAR data is not adequate for land cover classification especially in tropical environments. The purpose of this paper is to study selected feature extraction techniques to improve the C band usefulness. ERS1/2 tandem mode data was acquired over the Tapajos National Forest, Brazil, and gave the opportunity of testing the C band coherence map for classification purposes. The use of coherence for land cover classification is justified since it is expected low coherence in forest areas, in comparison with bare soil and sparse vegetation, which have high coherence. Texture is also a standard feature normally employed and from a set of fourteen of the Haralick's features, one was selected to be tested jointly with C band coherence map and the C band backscatter itself. Two sets of classes, one with 10 classes, the other with 4 classes were defined for training the classifiers. These classes represent typical classes in the region, and mainly include primary forest, several stages of regeneration and pasture. Iterative Conditional Mode (ICM), which is a contextual supervised per point classification scheme, and a supervised region classifier from a segmentation map produced by region growing type of segmentation were used to produce the classification maps. Statistical tests based on Kappa statistics were used to test the precision and significance of the results. The result shows that coherence map alone is adequate for the case of four classes, and the inclusion of any other feature either does not change the precision significantly or worse the results.
The principal land characteristics that can be estimated by means of airphoto interpretation are bedrock type, landform, soil texture, site drainage conditions, susceptibility to flooding, and depth of unconsolidated materials over bedrock. In addition, the slope of the land surface can be estimated by airphoto interpretation and measured by phptpgrammetric methods. The aim of this paper is to show an experimental use of satellite images in determining soil quality affected by anthropic activities as rock crushing, or scarifying. Scarifying activities began, in Murgia area, Apulia Region (Italy), as land improvement for agriculture uses. Scarifying is defined as loosening (the surface of soil) by using an agricultural tool or a machine with prongs. This kind of activity is facilitated by the availability, on the market, of scarifying machines and the objective is to get a stratum of agriculture-useful loose material on the soil surface. Apulia Region Government has permitted calcareous stone scarifying with Regional Law n.54 (August 31, 1981) according to National Law n.984 (Dicember 27,1977), that provides for encouraging to transform grazing in sown land in order to create new possibility of forage production to increase zootecnical facilities. We have used ERS-2/SAR images as contribution in the process of soil characterization.The area we have considered is in Puglia Region and is subject to soil transformation due to rocks crushed on land for agricultural facilities. European Union, through the same Apulia Region Government, has renewed funds for the improvement of meadow and grazing for an overall surface of 2000 hectares. In this way it is
clear to understand the importance of qualitative and quantitative evaluation of rock crushing or scarifying by using airphoto interpretation. We have evaluated the soil quality by introducing a multicriteria, analysis by using a qualitative and quantitative methodology, so that it will be possible to prevent damages on soil, sub-soil and hydrology. Decision analysis in Impact assessment is a set of procedures for analyzing complex decision problems. The strategy is to divide the decision problem into small, understandable parts; analyze each part; and integrate the parts in a logical manner to produce a meaningful solution. The terms multicriteria decision making (MCDM) and multicriteria decision analysis (MCDA) are used interchangeably.
This paper describes a non-parametric algorithm based on fuzzy-reasoning concepts and suitable for land use classification, either supervised or unsupervised, starting from pixel features derived from SAR observations. Pixel vectors composed by simple features calculated from the backscattering coefficient(s) in one or more bands and/or polarizations are iteratively clustered. At each iteration step, pixels in the scene are classified based on the minimum attained by a weighted Euclidean distance from the centroid representative of each cluster. Upgrade of centroids is iteratively obtained both from the previously obtained classification map and by thresholding a membership function of pixel vectors to each cluster. Such a function is inversely related to the weighted Euclidean distances from the centroid representative of each cluster. To yield the weighted distances from a pixel vector, its features are weighted by means of progressively refined coefficients, whose calculation still relies on the membership function through a least squares algorithm. Possible "a priori" knowledge coming from ground truth data may be used to initialize the procedure, but is not required. Experiments on MAC-91 NASA/JPL AIRSAR data on the Montespertoli test site show that a total of nine features derived from C-HV, L-HV, and P-HV observations are capable to discriminate seven agricultural cover classes with an overall pixel accuracy of 70%, with one tenth of the ground truth data used for training and the remaining nine tenths for testing the classification accuracy.
Various techniques for detecting anomalous reflective objects that extend a few tens of pixels in a SAR image are discussed. These techniques assume some prior knowledge of the object, to make the detection process robust and minimize false alarms. The basis of the method is a CFAR technique that assumes Weibull clutter statistics. The method is then augmented using knowledge of the shadow cast by the object, change detection, and information from terrain databases. Examples using a simulated SAR image are given.
Synthetic Aperture Radar (SAR) images are extensively used for the determination of oil slicks in the marine environment, as they are independent of local weather conditions and cloudiness. Oil spills are recognized by the expert's eye as dark patterns of characteristic shape in particular context. However, the major difficulty to be dealt with is to differentiate between oil spills and look-alikes of natural origin. A fully automated system for the identification of possible oil spills that imitates the expert's choice and decisions has been developed. The system's architecture comprises several distinct modules of supplementary operation (georeferencing, land masking, thresholding, segmentation) and uses their contribution to the analysis and assignment of the probability of a dark image shape to be an oil spill by means of a fuzzy classifier. The output consists of several images and table providing the user with all relevant information as well as supporting decision making. The case study area was the Aegean Sea in Greece. The system responded very satisfactorily for all 35 images processed. The complete procedure described is a fully automated stand-alone application running under Windows operating system.
Ocean waves properties propagating in grease ice composed of frazil and pancakes as observed by SAR images are discussed. An ERS-2 SAR scene relevant to the Greenland Sea in an area where the Odden ice tongue developed in 1997 is considered as case study. The scene includes open sea and ice covered waters where a wave field is traveling from the open sea region. Wind induced features known as "wind rolls" can be distinguished, allowing the estimation of the wind vector. Hence the related wind generated ocean waves can be retrieved using a SAR spectral inversion procedure. The wave field is tracked while it propagates inside the ice field, thus allowing the estimation of the wave changes. Under the assumption of continuum medium, physical ice properties are then retrieved using a special SAR inversion procedure in conjunction with a recently developed wave propagation model in sea ice. The model assumes both the ice layer and the water beneath it as a system of viscous fluids. As a result, the changes suffered by the ocean wave spectrum in terms of wave dispersion and energy attenuation are related to sea ice properties such as concentration and thickness. Although the free parameters to be inverted are the ice thickness and viscosity and the water viscosity, the ice thickness is the only parameter of geophysical interest. Results are finally compared with external ice parameters information.
In this paper, the basic performance of MTI/SAR radar in ship and ground vehicle applications is analysed. If a single antenna beam scans a sector, significant degradations in MTI sensitivity and SAR resolution occur due to the reduced dwell time on target. Improved performance can be achieved by digital array beamforming with multiple beams, or high speed scanning. SAR needs accurate phase error compensation by inertial measurements or autofocus due to the non-linear movement path of the antenna.
Many applications of SAR interferometry and differential interferometry lead to a set of sparse phase measurements that usually have to be unwrapped, and then interpolated on a regular grid. We investigate the utility of the scaling information available on the absolute phase, in the process of unwrapping a set of sparse, wrapped phase measurements. Scaling information is an important tool for the description of natural processes exhibiting fractal-like behaviour. One notable example is the interferometric phase contribution due to tropospheric inhomogeneities. Scaling properties can be estimated experimentally on a set of points through computation of the variogram. If it can be assumed that the absolute phase field obeys a defined scaling power law, then the wrapping operator will cause the variogram to depart from the modelled behaviour. Under these hypotheses, the difference between actual and modelled variogram can be used as an optimization Hamiltonian. In this work, we investigate whether the scaling information can be used as a constraint in retrieving the absolute (i.e. unwrapped) phase field from a set of sparse measurements. In particular, we consider here the problem of constructing a cost function which embodies the scaling requirement, and we test several strategies to optimise the cost.
Classification of Earth surface using SAR observations constitutes an important application of polarimetric SAR. The most promising and investigated polarimetric parameters for such a task have been the Entropy, alpha and Anisotropy (H, α and A) parameters. A similar use of coherent methods, however, appears to have been scarcely considered and remained essentially untested. In this contribution, we wish to address this issue, testing and comparing a wide range of polarimetric SAR parameters, coherent and incoherent, by means of different classification algorithms.
An original aspect of this work is also the study of the dependence of the classification results on the varying size of averaging windows of pixels. Such an analysis will permit to prove if the chosen polarimetric parameters provide a description only of "point-like" physical properties of the targets or if they also contain "extended", local information. The investigations to be reported should, therefore, not only reveal a systematic and quantitative assessment of the classification efficacy of different methodologies but also afford their comparison, an exercise, hitherto unavailable in the literature in common knowledge.
Many features of interest in SAR images cast shadows. Exploiting the information provided by these shadows may enhance the detection of such features. Here, algorithms of shadow processing schemes are derived and the efficacy of such schemes is investigated by analysis and simulation.
Pulse compression radar is used in a great number of radar applications. Excellent range resolution and high ECCM performance can be achieved by wide-band modulated long pulses, which spread out the transmitted energy in frequency and time. If a random noise waveform is used, the range ambiguity is suppressed as well. The range processing then correlates the received signal and a delayed reference. When the delay of the target signal coincides with that of the delayed reference a strong correlation peak occurs. In this paper, the theory of noise radar for Doppler/range indication is first described. Then the possible use of binary or low-bits ADC is analysed, which highly improves the signal-processing rate and reduces the costs. The application of noise radar to digital beam forming is finally discussed. Similar results are obtained using a frequency chirped waveform and low-bits ADC.
Segmentation and restoration of highly noisy images is a very challenging problem. There are a number of methods reported in the literature, but more effort still need to be put on this problem.
In this paper we describe the development and implementation of a new effective approach to segmentation and restoration of imagery with pervasive, large amplitude noise. The new approach is based on the recently developed stabilized inverse diffusion equations (SIDE) and mathematical morphology. First, we find an optimized SIDE force function. Secondly, we segment the image to several regions accurately using the SIDE method. Finally a grayscale mathematical morphological filter combined with SIDE is assigned to the initial image data in each region to suppress the noise and to restore the total image. A test study based on available database is presented, and the results so far indicate that this approach to highly noisy imagery segmentation and restoration is highly effective.
A ground-based Synthetic Aperture Radar (GB-SAR) interferometric technique is proposed for the topographic mapping. It is based on a coherent continuous-wave step-frequency radar moved along a horizontal rail to generate the synthetic aperture. Antennas are placed on a mechanical arm whose rotation enables to create an interferometric baseline. The synthetic aperture is 3 m long and the average radar-to-scene distance is 1200 m. Multifrequency measurements are carried out at Schwaz, Austria, using different baselines. The interferometrically derived topography is compared with an existing Digital Elevation Model (DEM) of the area.
The aim of this study is to evaluate the performances of a polarimetric scatterometer. This sensor can measure the module of the electromagnetic backscattering matrix elements. The knowledge of this matrix permits the computation of all the possible polarisation combinations of transmitted and received signals through a Polarisation Synthesis approach.
Scatterometer data are useful for monitoring a large number of soil physical parameters. In particular, the sensitivity of a C-band radar to different growing conditions of vegetation depends on the wave polarisation. As consequences, the possibility of acquiringi both polarisation components presents a great advantage in the vegetarian studies. In addition, this type of ground sensor can permit a fast coverage of the areas of interest.
A first test of the polarimetric scatterometer has been performed over an asphalt surface, which has a well-known electromagnetic response. Moreover, a calibration procedure has been tested using both passive (Trihedral Corner Reflector, TCR) and active (Active Radar Calibrator, ARC) radar calibrator.
Each year thousands of hectares of forest burnt across Southern Europe. To date, remote sensing assessments of this phenomenon have focused on the use of optical satellite imagery. However, the presence of clouds and smoke prevents the acquisition of this type of data in some areas. It is possible to overcome this problem by using synthetic aperture radar (SAR) data. Principal component analysis (PCA) was performed to quantify differences between pre- and post- fire images and to investigate the separability over a European Remote Sensing (ERS) SAR time series. Moreover, the transformation was carried out to determine the best conditions to acquire optimal SAR imagery according to meteorological parameters and the procedures to enhance burnt area discrimination for the identification of fire damage assessment. A comparative neural network classification was performed in order to map and to assess the burnts using a complete ERS time series or just an image before and an image after the fire according to the PCA. The results suggest that ERS is suitable to highlight areas of localized changes associated with forest fire damage in Mediterranean landcover.