Due to the side looking characteristics of the system, Synthetic Aperture Radar (SAR) images show very important changes of the local mean intensity, especially in areas with terrain variations. An algorithm for geometric and radiometric correction of SAR images using Digital Elevation Model is described. In addition, images acquired in different areas and with various sensors configurations are corrected. We study the impact of these corrections on a classification process.

An unsupervised method is first proposed to assess the variance and the spatial correlation coefficients of speckle noise in SAR images. They are obtained as regression coefficients, the former of local standard deviation to local mean, the latter of local unity-lag covariance to local variance, both calculated on homogeneous areas. For this purpose, an automatic procedure has been developed, based on that homogeneous areas produce clusters of scatterpoints that are aligned using the regression line. On true SAR images, the method is capable to carefully reject textured regions, in which speckle may be not fully developed and the variance of the signal is not negligible. On simulated speckled images, an impressive accuracy is obtained. Once the noise parameters are known, adaptive filtering is applied in a multiresolution fashion, to take advantage of increasing SNR of the noisy image at increasing scales, as well as to cope with the spatial correlation of the noise that is halved together with the resolution. Laplacian pyramids are generalized to the noise model by defining ratios of combinations of lowpass image versions, in which the dependence of the noise on the signal is largely removed, together with the nonstationarity of the mean. Experiments on both real and synthetic images demonstrate a high accuracy of results, both for noise estimation and for filtering.

This article describes a radargrammetric chain for which each step tends to be specific to SAR imagery. First, we start with two complex images acquired in radargrammetric conditions and for which the parameters of the geometric view models are refined. The images are then turned into epipolar geometry while keeping their complex feature. Afterwards a multiresolution method of registration of the two images is used that takes into account a geometric criterion as well as the radiometry specific to SAR images. Finally, a DEM and two orthoimages are computed by spatiotriangulation from the previously found matches and from the geometric view models of the original images.

The accurate knowledge of the geometry of an airborne SAR image is crucial for many applications. Repeat-pass SAR interferometry requires an accurate image matching to the 10^th of a pixel, while less demanding applications such as wide area coverage by multi-pass composite or image geocoding tolerate mismatches of pixels. Compared to spaceborne SAR, those accuracy requirements are stronger for airborne SAR sine they provide higher resolution. Moreover, irregularities in aircraft motion due to air turbulence introduce severe geometrical distortions in the images. Unfortunately, these distortions are tightly coupled with the signal processing algorithm used for computing SAR images. We have implemented a geometrical error model for our generic off-line SAR processor which provides an image distortion map. Derivatives with respect to errors in radar parameters and errors in aircraft trajectory measurements (velocity, altitude, oscillations are also provided, thus allowing the efficient estimation of the errors from distortion measurements (tiepoints). The paper is illustrated with some relevant application examples.

This paper proposes a target-detection scheme based on prior segmentation of the image. Introducing the prior knowledge of image structure provided by the previous segmentation eliminates many false target detections from background structure. The performance of the new scheme is shown to be identical to an ideal one-parameter CFAR for constant background. With real clutter backgrounds the background detection probability with the new scheme is considerably lower than with one-parameter CFAR, without any loss in target detection. We also demonstrate that, for smaller false alarm probabilities, the original segmentation yields nearly all the detections achieved by segmentation-based target detection.

This paper describes an optimized approach to identifying changes within a sequence of ESR images of Heathrow airport. We show that joint annealed segmentation avoids the false detections encountered along the edges of structural features when detecting differences between segmented scenes. This leads to an optimized change detection process, which can be applied in the detection of aircraft and vehicles around the airfield, even though the resolution of ERS image is only 25 m (range) by 6.25 m (azimuth). In addition, we show how joint segmentation of the coherence image between pairs of 35-day repeats yield an appreciable improvement in a false color change representation based on two amplitude images and their coherence image.

In this paper we compare the capability of Landsat TM optical imagery, JERS L-band and Radarsat C-band SAR for classifying rain forest into forest and not forest categories. In each case, simulated annealing provides the global optimum segmentation of the underlying variable. For the optical image the information is carried by the brightness, for JERS1 by the mean intensity and for Radarsat by the scene texture, where texture can be optimally measured in terms of the normalized log. We demonstrate that JERS1 and Radarsat provide similar classification into forest and not forest categories, when Landsat TM Band 5 imagery is adopted as the reference. Most of the discrepancies arise in regions of regeneration, where the physical difference between the imaging mechanisms of the three sensors has greatest impact.

Five new Distribution-Energy Maximum A Posteriori speckle filters are established for the following cases: single detected, multilook multi-channel detected, single look complex SAR images, separate complex looks, and fully polarimetric SAR data. As shown, these new filters are particularly efficient to reduce speckle noise, while preserving textural properties and spatial resolution, especially in strongly textured SAR images.

The Cassini mission constitutes the last big spatial enterprise of this century. It will investigate the Saturnian system, with particular interest for its larger satellite, Titan. A Ku band radar is included in the instrumentation. The radar will acquire data on 90% of Titan surface, operating as imaging radar, scatterometer, altimeter and passive radiometer. The possibility of retrieving information about physical and morphological properties of the surface is expected. In this work, a preliminary study about the simulation of expected e.m. response of Titan surface is performed. A fractal based model is employed to describe surface morphology, while Kirchhoff approximation is used to model e.m. response. The simulated data suggest the possibility of discriminating between three of the expected scenarios: a water ice surface, a water ice surface in presence of ammonia, liquid ethane surfaces.

In order to obtain a more accurate topographic model, a noise filtering step must be performed before the unwrapping of phases. We propose an iterative process that involves a filter in the spatial domain and a multi-resolution phase unwrapping method. The approach is based on the generation of an approximate phase model, which is iteratively refined. In the aim of preserving fine details in the interferogram that are directly related to the ground topography, an edge- preserving smoothing method has been applied. The local spatial content of the phase image is exploited in order to determine for each pixel the best matching shape of the filtering mask, searching in a set of different neighborhood systems. Then, a non-linear adaptive filtering function, based on the local estimation of noise and signal standard deviation, is adopted. The results obtained by processing several noisy simulated and real interferometrical images with the described method show clearly how the combination of a detail preserving adaptive filter with a noise robust phase unwrapping approach appears as a good means for reducing the influence of distributed noise and low coherence areas on the determination of a digital ground elevation model.

This paper investigates the coherence properties of a variety of terrain types to determine the extent that high coherence is also a characteristic of natural terrain as well as man-made objects. The paper also considers whether ratio techniques can be employed to avoid the inherent loss of resolution due to the computation of coherence. It also investigates the benefit of using data from interferometric baselines for ratio techniques.

Differential SAR interferometry measurements provide a unique tool for low-cost, large-coverage surface deformations monitoring. Limitations are essentially due to temporal decorrelation and atmospheric inhomogeneities. Though temporal decorrelation and atmospheric disturbances strongly affect interferogram quality, reliable deformation measurements can be obtained in a multi-image framework on a small subset of image pixels, corresponding to stable areas. These points, hereafter called Permanent Scatterers, can be used as a `natural GPS network' to monitor terrain motion, analyzing the phase history of each one. In this paper, results obtained using 45 ERS SAR images gathered over the Italian town of Camaiore (within a time span of more than 6 years and a range of normal baseline of more than 2000 m) are presented. The area is of high geophysical interest because it is known to be unstable. A subterranean cavity collapsed in October 1995 causing the ruin of several houses in that location. Time series analysis of the phase values showed the presence of precursors three months before the collapse.

We present a high frequency model for scattering present in SAR imagery. The model retains dominant terms of the electromagnetic scattering response of canonical scattering objects using solutions from both Physical Optics and the Geometric Theory of Diffraction. Both frequency and aspect dependence of scattering centers are modeled. It is applicable to single polarization as well as multiple polarization data, and it provides an effective, physics- based description of complex SAR imagery. The model parameters provide a concise, physically relevant description of a complex object and are thus good candidates for use in target recognition, radar data compression, and scattering phenomenology studies. Algorithms for estimating the parameters from measured SAR imagery are presented, and the problem of model structure selection is addressed.

In order to test a multi-frequency polarimetric scatterometer based on a Vector Network Analyzer, calibration measurements have been performed over point targets. Two trihedral corner reflectors with different dimensions have been employed. The radar cross sections have been measured at different frequency bands (L, C and X) and for different look angles between 23 degree(s) and 50 degree(s). Satisfactory results have been obtained in all three bands, however in the L-band the electromagnetic smog, due to mobile phones and airport radars, caused some difficulties in the extinction of the radiometric information. Other calibration tests have been planned before using the instrument as a ground-truth data acquisition device on the test-sites envisaged for the spaceborne SRTM and ENVISAT SAR missions.

In this paper, a much more rapid method of raw data generation is applied, starting by simulating the SAR image itself. Conventional methods of translating a thermal IR- or photographic image into a SAR image are first briefly reviewed. 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 center focusing, which ties together SAR data and sensor geometry. This method of generating raw data is shown to be several orders of magnitude faster than the direct method described above, and raw data from the I- and Q-channels can readily be simulated even on a PC. The effects of internal noise and phase errors, due to imperfect motion compensation, can also be included and simulated in detail. The method is exemplified, showing the phase error effect on the SAR-image of an urban scene, the use of autofocusing to restore the image resolution, and finally the degradation due to additive noise on the collected SAR raw data.