This PDF file contains the front matter associated with SPIE Proceedings Volume 7829, including the Title Page, Copyright information, Table of Contents, and the Conference Committee listing.

Goal of this paper is an evaluation of Bayesian estimators: Minimum Mean Square Error (MMSE), Minimum Mean Absolute Error (MMAE) and Maximum A-posteriori Probability (MAP). Such estimations have been carried out in the undecimated wavelet domain. Bayesian estimation requires probability density function (PDF) models for the wavelet coefficients of the reflectivity and of the signal-dependent noise. In this work several combination of PDFs will be assessed. Closed-form solutions for MMSE, MMAE and MAP have been derived, whenever possible; numerical solutions otherwise. Experimental results carried out on simulated noisy images evidence the cost-performance trade off of the different estimators in conjunction with PDF models. MAP estimation with generalized Gaussian (GG) PDF for wavelet coefficients of both reflectivity and signal-dependent noise (GG - GG) yields best performances. MAP with Laplacian - Gaussian (L - G) is only 0.07 dB less performing than MAP with GG - GG. However, the former admits a closed-form solution and its computational cost is more than ten times lower than that of the latter. Results on true single look high-resolution Cosmo-SkyMed SAR images provided by Italian Space Agency (ASI), are presented and discussed.

Change detection provides a powerful means for the initial detection of small target objects. However, speckle effects mean this type of approach can be difficult to apply to Synthetic Aperture Radar (SAR) imagery. This paper examines one method for target detection using change between a registered pair of SAR images. The technique may be parameterized to detect small target objects ranging in size from a few to perhaps a few hundred pixels. The approach considered here exploits the observation that the scattering response of many target types of interest is dominated by a small number of bright scatterers, whilst natural clutter regions tend not to display this property. The variance provides a useful statistic summarizing this effect, consequently the detection method considered here is based on the ratio of the variances of corresponding patches in the pair of images. Ideally any test statistic should be characterized by a known statistical distribution; this will allow formal tests of a null hypothesis to be carried out. Here the null hypothesis corresponds to no change, and knowledge of the distribution of the test statistic enables the implementation of a Constant False-Alarm Rate (CFAR) detection process. The analysis carried out herein considers the distribution of the variance ratio under realistic operating parameterisations for target detection in SAR imagery. Synthetic data is used to characterize this distribution, and Monte Carlo techniques are applied to derive empirical formulae for use in an online application. Results are presented for synthetic data and for a registered image pair, in the form of detection maps.

Synthetic Aperture Radar (SAR) is the only remote sensing technology that can provide high resolution images in adverse weather conditions and in day or night times. It is thus is a powerful tool for Earth monitoring. Certain applications, such as disaster relief, military reconnaissance and ice-flow and ship monitoring require a continuous flow of high-resolution images covering large areas; however, given the large amount of complex data generated and system limitations of data bandwidth and processing speed, not all the requirements can be met at the same time. In addition, multiple user requests are often submitted to the SAR system platform, and not all can be addressed, again due to limitations of area coverage. Increasing the speed of SAR processors and processing on-board are two ways to improve the SAR data throughput and therefore to meet the operational needs of all users. This paper discusses an optronic SAR processor capable of rapidly processing full-scene multi-looked images. Details of the processor design and image results are discussed. Estimations for speed and image throughput are provided, all presented in the context of the requirements for operational service of the various applications.

In this paper we investigate an unsupervised neural network approach for automatically extracting objects of interest from very high resolution (VHR) SAR images. The technique is based on the use of Pulse-Coupled Neural Networks (PCNN) which is a relatively novel technique based on models of the visual cortex of small mammals. The study discusses the use of PCNN technique in different applications. In a first case the extraction procedure is focused on the detection of buildings. In the second case the segmentation of a dark spot representing an oil spill in a SAR image is considered. The performance yielded by the PCNN is evaluated and critically discussed for a set of new generation of X-band SAR images taken by COSMO-Skymed and TerraSAR-X systems.

With the increasing availability of high resolution SAR imagery like RADARSAT-2 and TerraSAR-X, it becomes interesting to investigate the potential of this type of data for urban applications. There is however a great obstacle in using SAR imagery of urban areas: the corner reflector or "cardinal effect" problem. It is greatly problematic when multiple images of the same scene are taken from different azimuth angle. We propose a novel framework to overcome this problem by using contextual information about road orientation and building position to correct higher than normal pixel intensity caused by corner reflectors.

TerraSAR-X is the Synthetic Aperture Radar (SAR) German satellite which provides a high diversity of information due to its high-resolution. TerraSAR-X acquires daily a volume of up to 100 GB of high complexity, multi-mode SAR images, i.e. SpotLight, StripMap, and ScanSAR data, with dual or quad-polarization, and with different look angles. The high and multiple resolutions of the instrument (1m, 3m or 10m) open perspectives for new applications, that were not possible with past lower resolution sensors (20-30m). Mainly the 1m and 3m modes we expect to support a broad range of new applications related to human activities with relevant structures and objects at the 1m scale. Thus, among the most interesting scenes are: urban, industrial, and rural data. In addition, the global coverage and the relatively frequent repeat pass will definitely help to acquire extremely relevant data sets. To analyze the available TerrrSAR-X data we rely on model based methods for feature extraction and despeckling. The image information content is extracted using model-based methods based on Gauss Markov Random Field (GMRF) and Bayesian inference approach. This approach enhances the local adaptation by using a prior model, which learns the image structure and enables to estimate the local description of the structures, acting as primitive feature extraction method. However, the GMRF model-based method uses as input parameters the Model Order (MO) and the size of Estimation Window (EW). The appropriated selection of these parameters allows us to improve the classification and indexing results due to the number of well separated classes could be determined by them. Our belief is that the selection of the MO depends on the kind of information that the image contains, explaining how well the model can recognize complex structures as objects, and according to the size of EW the accuracy of the estimation is determined. In the following, we present an evaluation of the impact of the model order selection and the estimation windows size using TerraSAR-X data. We determine how many classes can be indexed depending on the Model Order and Estimation Window. The experimental results shows a good choice is model order 3 and 4, and estimation window with radius 15 × 15 pixels size.

The very-high spatial resolution provided by COSMO-Skymed products, also considering the concurrent TerraSAR-X mission, opens new challenges in the field of SAR image processing for remote sensing applications, maybe comparable to those represented by the first optical commercial satellites at the beginning of last decade. The Tor Vergata-Frascati test site, where extensive ground-truth data are available, was imaged by the COSMO constellation at two different days in summer 2010. This enabled first investigations on the potential of this type of imagery in providing a characterization of sub-urban areas by exploitation of both amplitude and phase information contained in the radar return. In particular this paper deals with the set-up of preliminary chains of automatic processing based on Multi-Layer Perceptron neural networks for pixel based analysis. Also some comments concerning the retrieval of information on the vertical properties of a single building are reported.

Soil moisture estimation is one of the most challenging problems in the context of biophysical parameter estimation from remotely sensed data. Typically, microwave signals are used thanks to their well known sensitivity to variations in the water content of soil. However, other target properties such as soil roughness and the presence of vegetation affect the microwave signals, thus increasing the complexity of the estimation problem. The latter problem becomes even more complex when we move on mountain areas, such as the Alps, where the high heterogeneity of the topographic condition further affect the signals acquired by remote sensors. In this paper, we explore the use of polarimetric RADARSAT2 SAR images for the estimation of soil moisture content in an alpine catchment. In greater detail, we first exploit field measurements and ancillary data to carry out an analysis on the sensitivity of the SAR signal to the moisture content of soil and other target properties, such as topography and vegetation/land-cover heterogeneity, that characterize the mountain environment. On the basis of the findings emerged from this analysis, we propose a technique for estimating moisture content of soils in these challenging operative conditions. This technique is based on the Support Vector Regression algorithm and the integration of ancillary data. Preliminary results are discussed both in terms of accuracy over point measurements and effectiveness in handling spatially distributed data.

The support of Earth Observation Systems to flood monitoring and damage assessment has been evaluated by integrating skills and knowledge of hydrologists and experts on image processing. Detailed flood maps are generated from the analysis of images pairs acquired by Cosmo-Skymed on the same area at different times. The adopted methodology applies image pre-processing and segmentation techniques with a multi-temporal approach. After noise reduction by despeckling, the user manually localizes few water points in one image. From selected points, segmentation distinguishes the classes of flood, permanent-water and no-change areas. Then, an anisotropic scanning analyzes the images to define its content. The resulting connected regions are converted into vectors to be entered as constraint for a Physically based 2-D hydraulic flood model. The model recursively varies the unknown boundary conditions to match at best the areas extracted from Cosmo-Skymed. The product is a hydraulic consistent report of the flooded area including information on water depth and velocities. By combining these information with vulnerability maps, extracted from optical satellite images with a supervised approach, an estimation of the damage is provided. The reported results refer to the monitoring of the flood event occurred in the Scutari Lake area (Albania, January 2010).

In general algorithms for soil moisture retrieval from high resolution satellite data cannot be easily extended to areas where they have not been calibrated and validated. This paper presents the application of an innovative approach for the detection of soil moisture from high resolution SAR images in order to overcome this main limitation by introducing a priori information. During the training phase, extensive data sets of SAR images and related ground truth on four areas characterized by very different surface features have been analyzed in order to understand the ENVISAT/ASAR responses to different soil, environmental and seasonal conditions. From preliminary analyses, the comparison of the backscattering coefficients in dependence of soil moisture values for all the analyzed datasets indicates the same sensitivity to soil moisture variations but with different biases, which may depend on soil characteristics, vegetation presence and roughness effect. These bias values have been used to introduce an adaptive term in the electromagnetic formulation of the backscattering responses from natural bare surfaces. The simulated data from this new model have been then used to train a neural network to be used then as an inversion algorithm. Preliminary results indicate an improvement in the accuracy of soil moisture retrieval with respect to the use of a traditional neural network approach. The results have been also compared with the estimates derived from the application of a Bayesian approach.

This paper presents the analysis of C and X band images in the scope of soil moisture detection in agricultural fields. Archived data have been analyzed in order to understand the SAR signal behavior of vegetated fields in comparison to bare soils. The results indicate that the sensitivity to bare fields of C and X band signatures is very close, while it changes in presence of vegetation. In particular the effect is directly proportional to amount of vegetation that in this preliminary analysis has been evaluated through the NDVI variable. After this analysis, a statistical approach has been applied to SAR images to infer the information on the soil moisture values. Several experiments have been carried out by considering only C band data, only X band data and a combination of C and X band data. For bare soils, C and X band data determine very similar results and in good agreement to ground measurements. For vegetated fields, C band data tend to underestimate soil moisture due to the vegetation attenuation, while X band data, mainly influenced by vegetation, determine very poor results. Encouraging results are obtained by the combination of C and X band data, thus indicating that X band data can be used in combination to C band data in order to compensate the effect of vegetation.

The PSInSAR technique, invented by Ferretti et. al. [1], [2], [3] ten years ago, meanwhile has proven its capability for very precise measurement of surface deformations. To achieve this, the influence of the atmospheric phase screen (APS) has to be removed. We investigated the APS for two series of TerraSAR-X high resolution spotlight data of a scene in Bavaria. In order to account for stratified troposphere and turbulence we augmented the APS estimation of StaMPS (Stanford Method for Persistent Scatterers) [4], that is we consider the APS as composed of a phase ramp, a part stratified with height and a turbulent component. The turbulent component is estimated via kriging. The necessary variograms can be computed under the assumption of isotropy as well as allowing for anisotropy. For short distances the variograms show a regime which is not visible for lower resolutions. In this paper we discuss the choice of appropriate variogram models with respect to our data.

The recent availability of wide-bandwidth, high-frequency, high-resolution SAR data is contributing to improved monitoring capabilities of spaceborne remote sensing instruments. In particular, the new COSMO/SkyMed (CSK) and TerraSAR- X (TSX) X-band sensors allow better performances in multitemporal DInSAR and PSI applications than legacy C-band sensors such as ENVISAT ASAR, with respect to both target detection and terrain displacement monitoring capabilities. In this paper we investigate about the possibility of achieving performances of PSI displacement detection comparable to those of C-band sensors, by use of reduced numbers of high-resolution X-band acquisitions. To this end, we develop a simple model for phase and displacement rate measurement accuracies taking into account both target characteristics and sensors acquisition schedule. The model predicts that the generally better resolution and repeat-time characteristics of new-generation X-band sensors allow reaching accuracies comparable to C-band data with a significantly smaller number of X-band acquisitions, provided that the total time span of the acquisitions is large enough. This allows in principle to contain the costs of monitoring campaigns, by using less scenes. Indications are more variable in the case of short-time acquisition schedules, such as those involved in the generation of so-called "rush products" for emergency applications. In this case, the higher uncertainty given by shorter total time spans lowers X-band performances to levels mostly comparable to those of the legacy medium-resolution C-band sensors, so that no significant gain in image number budget are foreseen. These theoretical results are confirmed by comparison of three PSI datasets, acquired by ENVISAT ASAR, CSK and TSX sensors over Assisi (central Italy) and Venice.

Satellite SAR Interferometry (InSAR) has been already proven to be effective in the analysis of seismic events. In fact, the surface displacement field obtained by InSAR application contains useful information to define the fault geometry (such as dip and strike angles, width, length), the extension of the rupture, the distribution of slip on the fault plain. However, the solution of the inverse problem, which means to recover the source parameters from the knowledge of InSAR surface displacement field, is rather complex. In this work we propose an inversion approach for the seismic source classification and the fault parameter quantitative retrieval based on neural networks. The network is trained by using a simulated data set generated by means of a forward model. The application of the methodology has been validated with a set of experimental data corresponding to different types of seismic events.

Spectral clustering has become one of the most popular modern clustering algorithms in recent years. In this paper, a new algorithm named entropy ranking based adaptive semi-supervised spectral clustering for SAR image segmentation is proposed. We focus not only on finding a suitable scaling parameter but also determining automatically the cluster number with the entropy ranking theory. Also, two kinds of constrains must-link and cannot-link based semi-supervised spectral clustering is applied to gain better segmentation results. Experimental results on SAR images show that the proposed method outperforms other spectral clustering algorithms.

Two kinds of doubly peaked ocean wave spectra such as Torsethaugen spectrum and Ochi-Hubble spectrum are used to simulate the mixed ocean waves with both swells and wind seas. The Envisat ASAR (Advanced Synthetic Aperture radar) image cross spectra of mixed ocean waves in different significant wave height (SWH), wave direction, wave component and peak period are then simulated by using Engen's nonlinear transformation formula. Analysis based on the simulation indicate that (1) in addition to the contribution of wind wave part and swell part of the mixed waves, the cross spectra of mixed waves consist of an extra term; (2) the cross spectra of mixed ocean waves dilate in range direction and shrink in azimuth direction (the so-called azimuth cutoff effect) and the cutoff effect increases for waves with larger wave height, or for waves propagating closer the azimuth direction, or for waves containing more wind wave component, or for waves with shorter peak period; (3) the cross spectra split into two parts for waves propagating along range direction; (4) the direction ambiguity of ocean waves can be removed by using the imaginary part of cross spectra.