EP2160628A2

EP2160628A2 - Method for processing multi-pass radar data for sensing and analysing multiple components of non-stationary scatterers

Info

Publication number: EP2160628A2
Application number: EP08709820A
Authority: EP
Inventors: Fabrizio Lombardini
Original assignee: Universita di Pisa
Current assignee: Universita di Pisa
Priority date: 2007-02-14
Filing date: 2008-02-14
Publication date: 2010-03-10
Also published as: WO2008125929A3; WO2008125929A4; WO2008125929A2; WO2008125929A9; ITPI20070012A1

Abstract

The invention relates to a method for processing imaging radar data of distributed and/or multiple layover scatterers contained in one or more preselected radar resolution cells, acquired by a radar system having at least one receiving channel with multi-pass multibaseline acquisition and consisting of at least one set of complex valued pixels. Each set corresponds substantially to a respective resolution cell in a plurality of complex focalized radar images, and each radar image is formed through each receiving channel for each pass. The method according to the invention comprises a multidimensional separation step, in particular two-dimensional, in a domain comprising the height/displacement velocity joint domain, of multiple radio echo components coming from the distributed and/or multiple layover scatterers in said or each cell, where the separation is carried out by means of two-dimensional processing step of the data in the baseline/acquisition time joint domain, said separation comprising the extraction in the height/displacement velocity joint domain of data of a distribution on height and displacement velocity of said scatterers and/or of components of said distributed and/or multiple scatterers. This method allows forming a differential tomographic image (70) that describes the reconstructed radio reflectivity (71) vs. height (72) and the displacement velocity (73), for three layover isolated scatterers from which three dominant reflectivity peaks depend (74,75 and 76), with a maximum ambiguities level (77) that is satisfactorily low.

Description

TITLE

METHOD FOR PROCESSING MULTI-PASS RADAR DATA FOR SENSING AND ANALYSING MULTIPLE COMPONENTS OF NON-STATIONARY SCATTERERS DESCRIPTION

Field of the invention

The present invention relates to the field of remote radar sensing, for extracting maps, in any weather and light conditions, for environmental monitoring, evaluation of natural risks, assistance in case of natural disasters, and for geophysics, relatively to phenomena such as topographical survey and subsidences, glaciers retreat, deforestation, for sensing even extreme environmental events such as earthquakes, tectonic plate displacements, land preservation and investigation, civil defence. In particular, the present invention relates to a method for processing interferometric multi-pass imaging radar data for sensing and analysing multiple components of radio non-stationary scatterers at different heights and/or displacement velocities.

Particular uses of the invention are in the fields of monitoring slow deformations and sensing other dynamic parameters in complex areas, and of surveying topographic and stratigraphic maps of areas subject to temporal perturbations. The method according to the invention, for example, is adapted to topographic and stratigraphic maps survey of areas subject to temporal perturbations such as lands, forests, frozen and snowy areas, urban areas, and for monitoring deformation and displacements in complex areas such as glacier sliding, landslide displacement in areas with high grade slopes, variation of the water level in swamps or flooded forests, variation of water table subjacency, subsidences in vegetated, volcanic, or urban areas, deformations and subsidences of infrastructures, quarries, open sky mines.

Description of the technical problem

As well known, different remote sensing techniques exist for environmental and geophysic monitoring, adapted to carry out an interferometric analysis of electromagnetic signals manipulated by radar systems.

In particular, a technique for processing imaging radar signals, so-called

"SAR Interferometry", is used in radar remote sensing, for extracting maps for topographic analysis and for urban areas, monitoring the use of the land, defining sites for base radio stations for telephony and wireless networks, geophysics and ore exploration, hydrology, glaciology, sensing avalanches and forest regions.

In particular, the technique allows to obtain digital maps of the height of grounds and surfaces, so-called Digital Elevation Models (DEM), on a large scale, at low cost and high precision, in a practically completely automatic way. The precision available on such topographic surveys is metric and also decimetric.

The acronym SAR stands for Synthetic Aperture Radar, which consists of a normal coherent radar, capable of measurements both the amplitude and the phase of the radio echo, having an antenna physically arranged on an aircraft. The radar data in SAR are then computed in order to have a virtual antenna aperture much larger than the actual aperture, exploiting the fact that the radar and the antenna move with the aircraft at a determined speed. Actually, each single pulse generated by the antenna, owing to the limited antenna aperture along the travel direction of the aircraft, so-called azimuthal dimension, would have the drawback of being extended rotationally in azimuth. However, by multiplying the pulses coming from the antenna, as the aircraft moves on its trajectory, it is possible to combine the returning echo signals, so-called "radio echo", as if they had been obtained and received at the same time by an antenna whose aperture were extremely larger. Thus, by focalizing the echoes in azimuth, a "synthetic beam" is obtained that is rather thin along the azimuthal dimension, and, by exploiting also the definition capacity, so-called resolution, in range (based on the modulation of the wave shape with a possible focalization in range, so-called impulse compression), radio reflectivity maps are obtained at high resolution, structured in "pixel" that are associated with the so-called resolution cells in range and azimuth. In case the SAR processing maintains the phase information, maps are obtained at high resolution both in amplitude and in phase of the focalized radio echo, so-called complex focalized SAR images.

In figure 1 an example is shown of SAR Interferometry, implemented with a radar system with two antennas 12 and 13, also called "receiving channels", which term used also for respective so-called phase centres, said antennas moving along tracks 1 and 2, which are substantially parallel and separated by a predetermined distance 3, so-called "baseline", that can be resolved into its components, horizontal 5 and vertical 4. In particular, the radio signal emitted by the radar sensor, after having reached a target 7, so-called "scatterer", in a determined resolution cell, is reflected by it as a radio echo that is sensed by the radar at the two receiving channels 12 and 13, after having traveled on paths 10 and 11. Scatterer 7 is located at a certain height 6, measured along a vertical height axis 19 by a horizontal reference plane defined by axes 17 and 30, where axis 17 extending along the direction of tracks 1 and 2, is called "azimuth". A line 14, which passes at the middle point of the baseline and at a projection T of scatterer 7, in a horizontal reference plane spaced at a same distance from said middle point, is called "nominal line-of-sight". The projection 15 of baseline 3 on a direction orthogonal to a nominal line-of-sight is called "orthogonal baseline". The direction of orthogonal baseline 15 defines the direction of an axis of the normal heights 18. This double receiving effect can be obtained by arranging, for example, the two antennas 12 and 13 on an aircraft at a distance from each other, or using multi-pass measurements of a same aircraft or of a satellite or of a platform movable on the ground on a rail with a single antenna travelling on the two different tracks 1 and 2. This way, a couple of complex focalized SAR images is obtained at slightly different angles in the vertical plane, which are then computed with interferometric techniques.

The height 6 of the scatterer present on the earth surface 8 is detected by triangulation, starting from the phase shift between the focalized radio echo recorded in the complex "pixel" values that correspond in the two images containing said scatterer ("interferometric phase").

In case the scatterer is not of the type so-called "point-like" but is "distributed", or it consists of a plurality of equivalent scatterers arranged on a portion of surface defined by the resolution cell in range-azimuth, so-called "surface scatterer", or it consists of a plurality of equivalent scatterers in a volume defined by the resolution cell in range-azimuth, so-called "volumetric scatterer", the height detected is a mean height.

With "unwrapping" techniques of the interferometric phase map, so-called interferogram, the interferometric phase measurement is disambiguated, eliminating the limit of the "equivocation height", exploiting the continuity of its variations in the plane of the image.

It is known, furthermore, the SAR 3D technique tomography, capable of providing three-dimensional radio reflectivity maps, or radio "stratigraphy", of volumetric scatterers, such as for example forests. The resolution available in the height dimension is decametric and also metric. The technique uses an acquisition on several baselines, i.e. with more than two radar antennas that form a "composed radar interferometer", obtainable by using a radar sensor with more than two receiving channels, or by using multiple passes on different tracks of a sensor with one or more receiving channels.

Under the condition of a so-called "far field" and a linear geometry of the composed interferometer by a suitable calibration of the data, for corresponding pixels in the various complex focalized SAR images, the radar echo component coming from a certain height determines, by impacting on the composed interferometer, a linear distribution of interferometric phase vs. the length of the baseline, corresponding to a harmonic component with a specific frequency, so-called "spatial frequency". Then, a Fourier transform relationship, i.e. a harmonic or spectral decomposition, exists between the radio reflectivity profile in height, so-called tomographical reflectivity profile in height, and the complex data in amplitude and phase, measured by the composed interferometer.

The tomographical reflectivity profile in height then consists of a spatial spectral analysis, which allows to separate the radio echo components at the various spatial frequencies or heights. Actually, this is like applying to the data a resonating filter on a spatial frequency, that can be varied as desired, or in general, even without the conditions of "far field" and linear geometry, it is like to sum all the data measured by the composed interferometer, so-called multibaseline data, after a suitable phase reset, thus enabling in turn the interferometer to sense particular heights (forming a synthetic beam in height).

It should be noted, dually, that the spatial spectral analysis in the transformed domain corresponds to an analysis of spatial correlation of the data along the composed interferometer, being the decline of the spatial correlation (spatial decorrelation) linked to the extension and the course of the tomographical reflectivity profile in height of the volumetric scatterer. This analysis can be carried out also by an identification of parameters of a predetermined model. This method can be in particular useful for reducing phenomena of intrinsic ambiguities in a typical use of not uniformly distanced baselines, or of data distributed in the baseline domain. Such ambiguities consist of high levels in an abnormal way of the so-called lateral lobes of the so-called point spread function (PSF) of an imaging system, in this case tomographic images in height, which can lead to detecting false scatterers or concealing weak scatterers.

The SAR 3D tomographic technique allows not only to survey three- dimensional radio reflectivity maps of volumetric scatterers, i.e. distributed with continuity in a volume that is semitransparent to the radio waves, as in case of forest zones and glaciers, but also sensing separately the position and the reflectivity of isolated multiple scatterers at different heights contained in a same resolution cell in range-azimuth, i.e. perspective^ overlapped since thay are at a same slant range from the radar. This condition is called of "layover", and is typical in complex geometries observable for example in mountain areas with robust slopes and in urban areas, said multiple isolated scatterers in layover being either point-like or distributed, either surface or also volumetric scatterers.

Furthermore, the SAR 3D Tomography, like other techniques of "fusion" of interferometric multibaseline data, allows sensing DEM with a precision higher than that obtainable with a single baseline, reducing effects of noise in the data and simplifying the achievement of an unambiguous information in height of the scatterer. A particular method for SAR 3D Tomography processing is described in US2003122700A1.

It is, yet, known the technique of SAR Differential Interferometry, capable of providing maps of long term slow displacements and deformations of a surface, such as subsidences and saliences of the ground or of urban areas, structural deformation of buildings, sliding glaciers, premonitory signs of landslides, effects of earthquakes. The precision available on measuring the deformation, along the direction of slant range, is centimetric and also millimetric. The technique uses specifiically the multi-pass acquisition, where the successive tracks of the sensor either can be substantially coincident, or more in general they define negligible orthogonal bases, or can be distinct from each other, i.e. with significant orthogonal bases.

In a much easier case of SAR Differential Interferometry, with at least one couple of SAR images obtained in different times from a same track, the interferometric phase, i.e. the phase between the complex valued pixels that correspond in the two images, is determined by slight variations of the slant range that is accumulated between an acquisition and another between the sensor and the scatterer/s present in the considered pixel. The measurement of interferometric phase allows then to sense the mean displacement velocity, or the deformation occurred between two acquisitions, or the temporal history of the deformation in case more than two acquisitions were available.

If the images are obtained from different tracks with significant orthogonal bases, the interferograms depend both on the topography and on the deformation vs. time. According to the configuration of the acquisition, the preexisting topographic data could be needed, even of not radar type, for compensating the topographical phase contribution and for extracting the displacement components, or it can be possible to obtain at the same time from the interferograms, if more than two, both the map of the displacement and the DEM, useful for georeferentiating the displacement. Among the methods that are more important and common of SAR

Differential Interferometry, the so-called method of Permanent Scatterers (PS) is known, as described in WO2000EP03741, which method has been recently extended to the case of so-called partially coherent PS, and to the case of Small Baseline Subsets (SBAS).

Furthermore, the extraction of the average displacement velocity allows also compensating the relative contribution of phase, bringing the data to a condition equivalent to that of static sight, useful for SAR 3D Tomography satellite applications, where the multibaseline acquisition is obtained by using multiple passes, according to the existing satellite technologies. The technique of SAR Differential Interferometry allows finally to extract maps of temporal "coherence" of the radar echoes between successive acquisitions, by means of statistic correlation measurements that quantify the degree of random temporal changes of the focalized radio echo collected in the complex valued pixels that do not depend on stiff movements of the scatterers in the pixels, but depend on changes of the electromagnetic characteristics or of the inner position of the scatterers. Such maps of coherence are used as indicator of the precision obtainable in the topographical survey and in the measurement of the displacement by means of interferometry and differential interferometry, as well as are used as a further product of remote sensing, for example in applications of classification of areas of different nature that can be distinguished also according to a different degree of temporal steadiness of the relative radio echo, or in applications of measurements small changes, so-called coherent change detection, for surveying so-called thematic maps. The techniques here above described have however the following drawbacks.

On the one hand, since the SAR Interferometry, the technique of merging interferometric multibaseline data, and the SAR 3D Tomography technique, all use the interferometric data by computing them substantially only in the spatial domain of baselines, when they operate by means of acquisitions obtained with multiple passes, they have the drawback of being subject to:

- loss of precision in height measurements in the DEM, when the temporal coherence of the radio echo coming from the examined surface or volume is not maintained high during the acquisitions, so- called condition of temporal decorrelation;

- defocalization of the tomographic reflectivity profiles in height, in the same condition;

- loss of measurements precision of position and/or multiple reflectivity in the presence of a condition of layover where the multiple scatterers are affected by motion with different displacement velocity (not stiff motion), in a way not compensable completely before the tomographic processing; - loss of measurements precision of position and/or multiple reflectivity in the presence of a condition of layover where the multiple scatterers are affected by temporal decorrelation. Such drawbacks are due to the fact that:

- the temporal phase contribution can be extracted and compensated only with reference to a single component of displacement velocity, or compensated only partially with reference to a mean displacement velocity;

- the temporal decorrelation cannot be compensated.

The subsequent temporal perturbations on the interferometric phases, or more in general on the complex data, are incorrectly interpreted as due to a height, to a reflectivity profile in height, or to multiple heights of layover scatterers, not corresponding to real.

On the other hand, the techniques of SAR Differential Interferometry, since they substantially operate a processing in the only temporal domain, because only the spatial contribution of phase can be extracted and compensated from a single scatterer, have in general the drawback of:

- they cannot work satisfactorily in the presence of distributed volumetric scatterers, or in case of stiff displacement either, since they cannot sense precisely the only average displacement velocity, in case of not stiff displacement;

- they cannot work in the presence of multiple layover scatterers that can be also distributed, both in stiff and not stiff displacement, as above defined;

- they cannot measure, in case of not stiff displacement, the distribution in height of the displacement velocity of the various layers of a volumetric scatterer, or the different components of displacement velocity of multiple layover scatterers, nor the distribution in height or the different components of kinematic parameters or parameters defining the temporal evolution of said scatterers;

- they provide a measurement of temporal coherence relative only to the whole of the volumetric scatterer or of the multiple scatterers.

Such drawbacks are due to sensitivity of said techniques to spatial decorrelation from distributed or multiple layover scatterers, even in case they are subject to a stiff displacement, and to an impossibility to identify temporal overlapped contributes in the radio echo from distributed scatterers, in particular, volumetric, or multiple layover, with not stiff displacement.

In conclusion, all the above described techniques have the drawback of using, in the processing, a simply monodimensional domain, only spatial or only temporal, or to use the dual domain (temporal or spatial, respectively), but corresponding to a model for a single scatterer and so in an incomplete way.

Only recently an extension of the technique of Differential lnterferometry with PS method has been proposed, capable of working in the presence of two point-like layover scatterers subject to a stiff displacement, separating the temporal and spatial components of the phase history by a specific model for the indicated condition, but with the cost of a substantive increase of the computational load. This extension, however has the drawback of not allowing to work practically with more than two multiple scatterers, nor with not rigid displacements, and of not allowing to work with multiple not point-like scatterers, nor with a volumetric scatterer, preferably affected by temporal decorrelation, nor they allow identifying the distribution in height of the displacement velocity of various layers of the volume.

It is felt, then, the need to provide a method for processing SAR interferometric data, acquired in multi-pass configurations, adapted to obtain a more reliable and complete, or new, informative content, with respect to the above described known techniques, by exploiting more extensively and completely the SAR data archives.

Summary of the invention It is then a feature of the present invention to provide a method for processing interferometric multi-pass multibaseline imaging radar data, acquired by one or more receiving channels for each pass, capable of improving the functionality of the techniques of SAR interferometry and of "merging" interferometric multibaseline data, SAR 3D tomographic data and SAR Differential Interferometry data.

Another feature of the invention is to provide a method for processing interferometric multi-pass multibaseline imaging radar data capable of forming new images of a joint distribution of height and displacement velocity and/or of other kinematic parameters and/or parameters defining temporal evolution of volumetric scatterers, such as glaciers, snowy surfaces and vegetated areas, and of multiple layover scatterers.

A further feature of the invention is to provide a method for processing interferometric multi-pass multibaseline imaging radar data capable of extracting new data of temporal coherence of the various components of volumetric scatterers and of multiple layover scatterers, such as forest or mountainous terrain, for example for classification puposes for making thematic maps of the terrain.

Another feature of the invention is to provide a method for processing interferometric multi-pass multibaseline imaging radar data capable of extracting new data of displacement velocity and/or of other kinematic parameters and/or parameters defining temporal evolution of various components of volumetric scatterers.

It is, yet, a feature of the invention to provide a method for extracting, from interferometric multi-pass multibaseline imaging radar data, multiple parameters of displacement velocity and/or other kinematical characterizations and/or parameters of temporal evolution, of height, and of radio reflectivity, of multiple scatterers in regions at risk affected by layover, such as high slope landslide areas, volcanic and/or vegetated areas with high salience/subsidence, infrastructures (for example dams, bridges, oil and pipelines).

It is, yet, a feature of the invention to provide a method for processing interferometric multi-pass multibaseline imaging radar data capable of forming digital maps of the height (so-called Digital Elevation Models or DEM) of natural or urban surfaces and of tomographic reflectivity profiles in height of volumetric or multiple scatterers, or extracting multiple height and radio reflectivity parameters of the latter, which are robust or are not much affected by loss of precision and defocalization from temporal signal decorrelation, or of virtually reducing phenomena of temporal coherence loss.

These and other objects are achieved by a method for processing imaging radar data, said radar data being relative to distributed and/or multiple layover scatterers contained in one or more preselected radar resolution cells, said radar data being acquired by a radar system with a multi-pass multibaseline acquisition so that input radar data are obtained of multi-pass multibaseline type, said input radar data consisting of at least one set of complex valued pixels, said or each set of pixels corresponding substantially to a respective resolution cell in a plurality of complex focalized radar images, said radar system having at least one receiving channel, and each radar image being formed through said or each receiving channel of said radar system for each pass, comprising a multidimensional separation step, in particular two- dimensional, of focalized multiple radio echo components, in a joint hybrid output domain of parameters, which comprise at least one spatial parameter and a kinematic parameter and/or a parameter defining a temporal evolution, wherein said joint output domain comprises the height/displacement velocity joint domain, said focalized multiple radio echo components originating from said distributed and/or multiple layover scatterers in said or each cell, said separation step being carried out by an at least two-dimensional processing step of said input radar data, or of radar data obtained from said input radar data, in a processing joint domain comprising the baseline/acquisition time joint domain, said separation step using functions, which are responsive to said parameters, describing components of said input radar data, or describing data obtained from said input radar data, in said processing joint domain, and/or using functions, which are responsive to said parameters, or describing correlations or correlation components of said input radar data, or describing data obtained from said input radar data, in said processing joint domain, said separation step comprising an inference of data, in said joint hybrid output domain and in particular in the height/displacement velocity joint domain, on a distribution with respect to said parameters, in particular height and displacement velocity, of said distributed and/or multiple scatterers and/or of components of said scatterers and/or of said components of focalized radio echo.

In this way a method is obtained for processing interferometric multi-pass multibaseline imaging radar data adapted to carry out a two-dimensional or multidimensional spatial-temporal separation, preferably at high resolution and/or with ambiguity suppression, of multiple components on complex type and distributed in the baseline-acquisition time domain, i.e. a method for processing data by a sparse spatial-temporal composed radar interferometer, comprising the production of new images of radio reflectivity distribution in the height/displacement velocity joint domain, i.e. new "tomographic differential" images, or of joint distribution of height and displacement velocity, and/or of radio reflectivity distribution in the joint domain of height - displacement velocity of one or more kinematic parameters and/or parameters defining temporal evolution, i.e. new differential generalized tomographic images, or images of joint distribution of height and one or more kinematic parameters and/or parameters defining temporal evolution.

In particular, said height of said hybrid joint domain of parameters, which comprise at least one spatial parameter and a kinematic parameter and/or a parameter defining a temporal evolution, is selected from the group comprised of:

- normal height, or height measured in a direction orthogonal to a nominal line-of-sight, or even to an actual line-of-sight;

- vertical height. In particular, said displacement velocity of said hybrid joint domain is selected from the group comprised of:

- line-of-sight displacement velocity, in a direction of a nominal or actual line-of-sight;

- vertical displacement velocity; - displacement velocity in horizontal range.

In particular, said kinematic parameters of said hybrid joint domain are relative to displacements according to directions selected from the group comprised of:

- line-of-sight; - vertical;

- horizontal.

These different components of height, displacement velocity and kinematic parameters are linked to each other by a so-called angle of sight. Furthermore, said displacement velocity and said kinematic parameters can also refer to a direction tangential to the surface. This way, since the acquisition of multi-pass radar data is typical of the mode for processing the SAR Differential lnterferometric data and since the acquisition of radar multibaseline data is typical of the mode for processing the SAR 3D Tomography, the present method is adapted to carry out an innovative synergistic combination of both modes, in such a way to obtain distribution images of distributed and/or multiple scatterers or radio reflectivity distribution images of distributed and/or multiple scatterers in a joint domain of parameters, which comprise at least a spatial parameter and a kinematic parameter and/or a parameter defining a temporal evolution, comprising the height/displacement velocity joint domain, exploiting fully the information of the data, by using in the process the two-dimensional baseline and time spatial domain in a complete way. Furthermore, the synergistic combination of at least two-dimensional processing modes in the present method allows to separate undesired temporal effects in the extraction of data in the height domain. In particular, said method comprises a step of definition a nominal value of a complex response, of the multi-pass multibaseline radar acquisition system, to the radio echo coming from a single point-like scatterer at a specific height and having a specific line-of-sight displacement velocity, in absence of noise and temporal decorrelation.

In particular, said complex response consists of complex nominal values of corresponding pixels in said focalized images formed by various channels in various passes, said pixel values originating from a single radio echo component, vs. height and displacement velocity of said single point-like scatterer in a preselected two-dimensional domain.

Advantageously, each of said complex nominal values can be defined in amplitude and phase, in particular said amplitude being unitary, said phase being obtained by evaluating changes of radio echo phase shift in its path from said point-like scatterer towards a corresponding receiving channel with respect to a radio echo phase shift in its path between the same scatterer and a predetermined reference receiving channel of a predetermined reference pass, each of said phase shifts being obtained by evaluating the length of said path accounting for its variation with respect to said reference pass and for wavelength.

In particular, the length of said path is obtained by evaluating a relative geometry between channel and scatterer, and said variation with respect to said reference pass is determined by the line-of-sight displacement velocity and by the temporal delay after said reference pass. In particular, said relative geometry is determined by the geometric configuration and by the location of said multiple baselines with respect to said range-azimuth resolution cell and by the height of the scatterer.

Advantageously said complex response is computed including also other kinematic parameters and/or parameters defining temporal evolution. In particular, said other parameters can be line-of-sight acceleration and/or the line-of-sight acceleration derivative and/or the line-of-sight step displacement or step phase variation and/or the temporal delay between the step displacement or phase variation and the reference pass.

In this way the complex response describes an equation of dependence of a generic component of said input radar data in said processing joint domain vs. said height parameters and displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution.

Advantageously, said values of the complex response are structured in a "steering vector". Advantageously, the multi-pass multibaseline acquisition forms a composed spatial-temporal interferometer, and in conditions equivalent to the far field and linear geometry of the composed spatial-temporal interferometer, said complex response consists of complex data in the baseline/acquisition time domain corresponding nominally to a specific spatial frequency, and to a specific linear distribution of an interferometric phase temporal component responsive to the temporal delay between said acquisitions, corresponding to a "temporal frequency", forming a specific two-dimensional spatial-temporal harmonic.

Advantageously, in conditions equivalent to the far field and linear geometry of the composed spatial-temporal interferometer, said complex response consists of complex data in the baseline/acquisition time domain corresponding nominally to a specific spatial frequency, and to a specific non- linear distribution of an interferometric phase temporal component responsive to the temporal delay between said acquisitions.

Advantageously, in conditions equivalent to the far field and linear geometry of the composed spatial-temporal interferometer, said complex response corresponds to a specific spatial frequency, and to a specific temporal frequency and/or chirp rate and/or rate of change of the chirp rate and/or step phase variation and/or temporal delay between the step displacement or phase variation and the reference pass.

Advantageously, said step of definition the complex response includes computing an effect of refraction of the radio propagation in volumetric scatterers, so-called dense scatterers, such as glaciers.

In particular, said two-dimensional separation step in the joint output domain of height/displacement velocity is obtained with a two-dimensional processing technique in the baseline/acquisition time processing joint domain, chosen among, or obtained from combinations of: - two-dimensional spacial-temporal spectral analysis of said complex data in the baseline/acquisition time processing joint domain, even sparse data;

- resonant or passband two-dimensional filtering of said data in said processing joint domain; - forming a "hybrid" two-dimensional synthetic beam in the height/displacement velocity joint output domain;

- signal processing of two-dimensional array sensors, so-called 2D "array processing", for sensing arrival directions of waves propagating in a three-dimensional space, so-called 2D direction of arrival (DOA) estimation, and in particular for sensing also their amplitude, where one of the two spatial directions is formally replaced by said displacement velocity parameter;

- reconstruction of images by means of variational methods. In particular, said multidimensional separation step in the joint output domain of height/displacement velocity and/or of other kinematic parameters and/or parameters defining temporal evolution is obtained with a two- dimensional processing technique in the baseline/acquisition time processing joint domain, chosen among, or obtained from combinations of: forming a "hybrid" multidimensional synthetic beam in the joint output domain of height/displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution; signal processing of two-dimensional array sensors or array processing, for sensing multiple parameters of waves, Advantageously, said technique chosen for the two-dimensional separation step is obtained by a method for two-dimensional analysis with high resolution and/or ambiguity suppression, selected from the group comprised of, or obtained from a combination of:

- adaptive methods; - apodization methods;

- model-based methods or parametric methods;

- Fourier or beamforming methods followed by two-dimensional deconvolution methods for reducing ambiguities;

- maximum entropy methods. In particular, said parametric methods are based on specific or approximate models of radio reflectivity distribution in the height/displacement velocity joint domain, or of spatial-temporal spectra, for surface and/or volumetric and/or multiple scatterers, with distribution of displacement velocity for the radio echo component at a certain height of null extension, and/or of a not null extension, fixed (decoupled) or variable (coupled) with the height of the various radio echo components.

In particular, said specific or approximate models model radio echo components without and/or with temporal decorrelation.

In particular, said parametric methods are of a class or are obtained from combinations of classes selected among:

- subspace decomposition of a correlation matrix of said data;

- iterating alternated estimation and cancellations;

- least squares amplitude fitting;

- autoregressive models or autoregressive moving average models;

- models identification correlation or covariance matrix fitting.

In a possible embodiment, for said two-dimensional separation step of a two-dimensional analysis based on Fourier transform is used. Advantageously, for said multidimensional separation step of a technique of two-dimensional analysis with high resolution and/or ambiguity suppression is used with an output in said joint hybrid output domain, selected from the group comprised of or obtained from a combination of: adaptive methods; model-based methods or parametric methods; methods of beamforming followed by multidimensional deconvolution methods.

Advantageously, said methods are based on the knowledge of the complex response, or of the steering vector.

Advantageously, said two-dimensional or multidimensional separation includes the extraction of a number of multiple layover scatterers, based on least squares amplitude fitting.

Advantageously, said method for processing imaging radar data comprises a preliminary calibration step of said multi-pass multibaseline radar data. In particular, said calibration step comprises at least one phase selected from the group comprised of:

- coregistering said complex focalized radar images deriving from said multi-pass multibaseline acquisition;

- obtaining the geometric configuration and location of said multiple baselines;

- calibrating amplitude and/or phase of various receiving channels;

- compensating the effects due to a possible non-linearity of the geometry of said composed spatial-temporal interferometer and at the curvature of the electromagnetic radio echo wave fronts, so-called deramping;

- compensating parasitic phase shifts due to variation of the radio propagation velocity in the atmosphere and/or in the ionosphere during the multi-pass acquisition, so-called atmospheric compensation. Advantageously, said method for processing imaging radar data comprises a preprocessing step of said multi-pass multibaseline radar data for bringing said data in conditions adapted to the successive two-dimensional or multidimensional separation.

In particular, said preprocessing step comprises at least one phase selected from the group comprised of:

- minimizing parasitic spatial decorrelation effects due to variation of geometric conditions for optimal coregistration of the images vs. an examined height; - minimizing random scintillation effects of the radio echo, which are due in case of distributed scatterers or distributed multiple scatterers, at an interaction of the radio waves with a surface or volume microstructure of said scatterers, or the inner position of equivalent scatterers, so-called complex "speckle" phenomenon or "fading", that adds to additive noise of so-called "thermal" type;

- minimizing ambiguities effects, in a successive two-dimensional or multidimensional separation, due to two-dimensional sparse non uniform sampling of the baseline-acquisition time domain carried out in said multi-pass multibaseline radar data acquisition step;

- stabilizing the correlation matrix of the multi-pass multibaseline complex data by means of so-called diagonal loading, in case of two- dimensional or multidimensional separation with adaptive method, based on said correlation matrix, and/or in case of preprocessing comprising an extraction step of the number of multiple scatterers by means of eigenvalue-based methods applied to said matrix;

- extracting the number of multiple scatterers, for example in case of two-dimensional or multidimensional separation of multiple scatterers with a class method based on subspace decomposition, or more in general calculating a so-called data model order.

Advantageously, said step of minimizing parasitic spatial decorrelation effects is obtained carrying out a compensation of the so-called migration in range by means of a coregistration in variable range with the height of interest. Said preprocessing step for minimizing said effects of spatial decorrelation with that of separation being a method for three-dimensional processing in the processing joint baseline-acquisition time - range domain.

Advantageously, said step of minimizing scintillations effects is carried out by using so-called multiple looks for each corresponding pixels in the various focalized complex images, or for groups of pixels adjacent to and comprising a pixel of interest or reduced resolution multiple focalized versions of the same.

Advantageously, said step of minimizing scintillations effects is obtained by calculating the correlation matrix of the multi-pass multibaseline complex data, through a coherent average on said multiple looks, on which a successive two-dimensional or multidimensional separation is based.

In particular, said complex multi-pass multibaseline radar data are structured in a data vector with the same structure of said steering vector, or in more data vectors with the same structure of said steering vector in case of multiple looks.

Advantageously, said step of minimizing ambiguities effects is obtained by two-dimensional interpolation of the data with a priori information on the so- called support in the domain of height/displacement velocity and/or other kinematic parameters or parameters defining the temporal evolution, where the radio echo components are expected, or on their average statistic reflectivity distribution in the height-displacement velocity domain, or extension of methods of monodimensional interpolation with a priori information, to obtain interpolated multi-pass multibaseline data. Advantageously, said two-dimensional interpolation can be obtained by a linear transformation of the data, defined for minimizing the square modulus of the deviation between the complex response relative to the interpolated baselines and/or acquisition times and the response obtained from said linear transformation applied to the complex response relative to the available baselines and acquisition time, in particular, said square modulus of the deviation being cumulated, or weighed and cumulated, for a grid of values of height and displacement velocity in said support.

Advantageously, said step of minimizing ambiguities effects is carried out by two-dimensional windowing methods. In particular, said two-dimensional windowing methods are applied as desired at least to one of the following types of said data and matrix:

- complex multi-pass multibaseline radar data;

- interpolated data;

- calculated correlation matrix in case of two-dimensional or multidimensional separation based on the correlation matrix.

In a possible embodiment, said preprocessing step comprises a phase of computation of the correlation matrix of the complex multi-pass multibaseline radar data through a coherent average on the baselines of identical length and/or on the temporal ranges between acquisitions of identical duration, or on partitions of the data, in particular, after deramping and two-dimensional interpolation, and advantageously two-dimensional windowing, in case of two- dimensional or multidimensional separation based on said correlation matrix and/or in case of preprocessing comprising an extraction step of the number of multiple scatterers by said eigenvalue based methods, and use of a so-called single look for keeping a full capacity of resolution in range-azimuth.

In particular, said diagonal loading for stabilizing the correlation matrix is fixed or adaptive.

Advantageously, said step of extracting the number of multiple scatterers is obtained by eigenvalue-based methods applied to the correlation matrix, in particular a stabilized correlation matrix.

Advantageously, the complex response is calculated for ideal geometric configuration and/or acquisition times which the data can be brought to, in case a deramping step and/or interpolation is carried out.

Advantageously, in case of a separation based on a calculated correlation matrix after interpolation using a single look, the complex response that can be used at the separation is obtained as a partition of the complex response corresponding to said configuration and/or acquisition times of the interpolated data.

Advantageously, said two-dimensional or multidimensional separation is carried out on said calibrated and/or preelaborated multi-pass multibaseline radar data, in case a calibrating and/or preprocessing step has been carried out previously. Advantageously, for minimizing parasitic spatial decorrelation effects by migrations in range, the separation step can be obtained also by means of three-dimensional processing, in a processing joint domain comprising in addition to the baselines and to the acquisition time also the range, of said data, preferably calibrated and/or preelaborated with preliminary inverse Fourier transform in range, and refocalization in range jointly to said separation by three-dimensional processing techniques. In this way, in case of two- dimensional separation the output of the three-dimensional processing techniques is in the joint domain of height/displacement velocity jointly to the domain in range. In particular, said method for processing imaging radar data comprises at least one phase selected from the group comprised of:

- preparing an image of height and displacement velocity joint distribution for an actual range-azimuth resolution cell;

- preparing an image of reconstructed radio reflectivity distribution in the height/displacement velocity joint domain for the actual range-azimuth resolution cell (hereinafter called "differential tomographic image");

- preparing an image of height and displacement velocity joint distribution and/or other kinematic parameters and/or parameters defining temporal evolution for said cell; - preparing an image of reconstructed radio reflectivity distribution in a joint domain of height and displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution for said cell (hereinafter called generalized differential tomographic image). In particular, said reflectivity can be given also in an equivalent scale such as amplitude, intensity, a so-called radar cross section (RCS)₁ a normalized RCS.

In a possible exemplary embodiment, said method for processing imaging radar data comprises a further post-processing step in the joint domain of height/displacement velocity, or of spacial-temporal frequencies, for extracting additional data from said differential tomographic image.

In particular, said post-processing step comprises at least one phase selected from the group comprised of: - extracting multiple parameters of height and/or displacement velocity and/or radio reflectivity of multiple layover scatterers;

- extracting the number of multiple layover scatterers;

- extracting from the differential tomographic image a radio reflectivity distribution, so-called marginal, in the only height domain, for obtaining a robust tomographic reflectivity profile in height with respect to a temporal signal decorrelation, or for virtually reducing phenomena of temporal coherence loss;

- extracting multiple parameters of height and/or radio reflectivity of multiple layover scatterers from said robust tomographic reflectivity profile in height, or from the differential tomographic image, for obtaining said parameters that are robust vs. a temporal signal decorrelation, or for virtually reducing phenomena of temporal coherence loss;

- extraction of a single height parameter of a single scatterer, either a point-like or a surface or a volumetric scatterer, from the differential tomographic image or from said robust tomographic reflectivity profile in height, to obtain a robust DEM having a temporal signal decorrelation, or for virtually reducing phenomena of temporal coherence loss;

- extracting temporal coherence measurements of the various components of volumetric scatterers or of multiple layover scatterers, or preparing a profile of temporal coherence vs. height, (hereinafter called

"tomographical temporal coherence profile in height"), and/or obtaining the temporal coherence multiple parameters of the extracted multiple layover scatterers;

- extracting displacement velocity measurements of various components of volumetric scatterers, or preparing a profile of displacement velocity vs. height, (hereinafter called "tomographical displacement velocity profile in height");

- minimizing random speckle scintillation effects by using multiple looks, to each of which, or to subsets of which, the two-dimensional separation has been applied separately, and by an incoherent average of the various differential tomographic images obtained from them, or by an average of various multiple parameters of height and/or displacement velocity and/or radio reflectivity, or various parameters of only height, and/or various robust tomographic reflectivity profiles in height, and/or various height tomographic coherence profiles and/or various temporal coherence multiple parameters and/or various tomographic displacement velocity profiles in height extracted therefrom. Advantageously, said step of extracting multiple parameters of height and/or displacement velocity and/or radio reflectivity is obtained from dominant peaks of said differential tomographic image and/or from their neigbourhoods.

Advantageously, said step of extracting the number of multiple scatterers is obtained by a threshold test on said multiple reflectivity parameters. Advantageously, said step of extracting a marginal radio reflectivity distribution is obtained by integration along the displacement velocity domain and/or by extraction of a monodimensional maximum along said domain of said differential tomographic reconstructed image in the height/displacement velocity joint domain, restricted to each of the various actual heights. Advantageously, said step of extracting multiple parameters of height and/or radio reflectivity from said robust tomographic reflectivity profile in height is obtained from dominant peaks of said profile and/or from their neigbourhoods.

Advantageously, said step of extraction of a single height parameter from said robust tomographic reflectivity profile in height is obtained from the highest peak value of said profile and/or from a neigbourhood thereof.

Advantageously, said step of extracting temporal coherence measurements of the various scatterer components is obtained by extraction of a displacement velocity band and/or a band of temporal frequencies and/or so-called correlation time and/or coherence measurements and/or by extraction of the course of a temporal decorrelation through a Fourier reverse transform from said differential tomographic reconstructed image in the height/displacement velocity joint domain restricted to each of the various actual heights, and/or to said multiple height parameters, or to the heights comprised in their neighbourhoods with following average operations for each neighbourhood.

Advantageously, said step of extracting displacement velocity measurements of different scatterer components is obtained from the highest monodimensional peak value and/or from a monodimensional centroid of said differential tomographic reconstructed image in the height/displacement velocity joint domain restricted to each of the various actual heights.

In case of multidimensional separation in a joint output domain of height/displacement velocity and/or of other kinematic parameters and/or parameters defining temporal evolution, the post-processing step can comprise a post-processing step in the joint domain of height/displacement velocity and/or of other said parameters, for extracting additional data from said generalized differential tomographic image. In particular, said post-processing step in the joint domain of height/displacement velocity and/or of other kinematic parameters and/or parameters defining temporal evolution comprises at least one phase selected from the group comprised of:

- extracting a marginal radio reflectivity distribution in the only domain of height/displacement velocity from the generalized differential tomographic image, for obtaining a differential tomographic image, wherein this differential tomographic image can then be computed in one or more post-processing steps in the joint domain of height/displacement velocity.

- extracting multiple parameters of height and/or displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution and/or radio reflectivity of multiple layover scatterers.

- extracting the number of multiple layover scatterers.

- extracting displacement velocity measurements and/or of other kinematic parameters and/or parameters defining temporal evolution of the various components of volumetric scatterers, or preparing one or more profiles of kinematic parameters and/or parameters defining temporal evolution vs. height (also called "profiles of kinematic parameters and/or parameters of tomographic temporal evolution in height").

- calculating an incoherent average of the various differential generalized tomographic images or an average of the various multiple parameters and/or various profiles extracted therefrom, in case of a multidimensional separation applied separately to multiple looks or subsets thereof.

Advantageously, said extraction of marginal radio reflectivity distribution in the only domain of height/displacement velocity is obtained by integration of the generalized differential tomographic image in the domain of said kinematic parameters and/or parameters defining temporal evolution, and/or by extraction of the monodimensional or multidimensional peak of the generalized differential tomographic image restricted to each different couples of actual values of height and displacement velocity.

Advantageously, said step of extracting multiple parameters of height and/or displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution and/or radio reflectivity is obtained from dominant peaks of said generalized differential tomographic image and/or from their neigbourhoods.

Advantageously, said step of extracting the number of multiple scatterers is obtained by a threshold test on said multiple reflectivity parameters extracted from the generalized differential tomographic image. Advantageously, said step of extracting displacement velocity measurements and/or of other kinematic parameters and/or parameters defining temporal evolution of the various scatterer components is obtained from the highest monodimensional or multidimensional peak value and/or from a centroid of said generalized differential tomographic image restricted to each of the various actual heights.

In particular, said method for processing imaging radar data can comprise, in case of two-dimensional or multidimensional separation with parametric method or method based on models, a step of obtaining from the model at least one said parameter and/or profile for the actual range-azimuth resolution cell, selected from the group comprised of:

- multiple parameters of height and/or displacement velocity and/or other kinematical characterizations and/or temporal evolution and/or radio reflectivity;

- number of multiple layover scatterers; - a single height parameter to obtain a robust DEM;

- multiple parameters of robust height and/or radio reflectivity;

- a robust tomographic reflectivity profile in height;

- a tomographical temporal coherence profile in height and/or temporal coherence multiple parameters of the multiple layover scatterers; - a tomographical displacement velocity profile in height and/or one or more profiles of other kinematic parameters and/or parameters defining temporal evolution.

Advantageously, in case of use of multiple looks, to each of which, or to subsets of which, the two-dimensional or multidimensional separation with parametric method has been applied separately, various results of identification or various of said multiple or single parameters and/or profiles are averaged out.

Advantageously, said method for processing imaging radar data can be applied to multi-pass multibaseline data of a synthetic aperture radar (SAR).

Advantageously, said multi-pass multibaseline data are acquired by a radar imaging system of a type selected from the group comprised of or obtained from a combination of: - radar with a single receiving channel forming a single complex focalized image for each pass;

- multichannel radar of multiantenna type, forming at least two focalized complex images for each pass;

- multichannel radar of multiantenna type with commutation of the transmitter, forming at least three focalized complex images for each pass.

In particular, said commutation phase of the transmitter uses a technique, so-called "ping-pong" in case of two antennas, capable of forming for each pass more images of the number of receiving channels so-called real, owing to the synthesis of receiving channels so-called bistatic equivalent additional channels.

Advantageously, said multichannel radar of multiantenna type is of a type selected from the group comprised of or obtained from combination of:

- multiantenna co-located radar, having antennas connected to a same vehicle or platform;

- multiantenna distributed radar, having antennas arranged separately on vehicles or different platforms, which travel along substantially parallel tracks in case of said vehicles or mobile platforms. Advantageously, said radar imaging system is transported by one or more means selected from the group comprised of:

- an aircraft or other avionic platform;

- a satellite or other spatial platform;

- a flight of aircrafts and/or other multiple avionic platforms;

- a flight of satellites or so-called satellite "cluster" and/or other spatial platforms;

- a ground based motorized rail vehicle;

- a plurality of ground based motorized rail vehicles. Advantageously, said multi-pass multibaseline data are acquired by a radar imaging system of passive type with a single receiving channel or with a multichannel of multiantenna type, in particular, using radar pulses or in general radio signals transmitted by an external source, said passive radar imaging system being in particular based fixed on ground or based on a cluster of satellites, the source of the pulses or signals used by said passive radar imaging system being in movement in case of a SAR static passive radar system.

Brief description of the drawings.

The invention will be made clearer with the following description of some exemplary embodiments, exemplifying but not limitative, with reference to the attached drawings wherein:

- Figure 1 shows diagrammatically a perspective view of the operation of a known radar imaging technique of SAR lnterferometric type with one baseline; - Figure 2 shows a perspective view of a single generic cell in range-azimuth resolution of the radar imaging system, the relative focalized radio echo being acquired by one or more receiving channels for each of more consecutive phases, said cell defining with the normal height direction a volume where a plurality of equivalent scatterers is contained making up a volumetric scatterer, having different line-of-sight displacement velocity;

- Figure 3 shows the characterisation of two configurations of multi-pass multibaseline acquisition in the orthogonal baseline - acquisition time domain, for acquisition with one (figure 3a) or three (figure 3b) receiving channels for each of the consecutive phases, or two configurations of sparse spatial-temporal composed radar interferometer or configurations of two-dimensional sparse sampling of the baseline-acquisition time domain;

- Figures 4 and 4B show a block diagram, including also optional steps, which describes the method according to the invention;

- Figure 5 shows an example of a differential tomographic image of radio reflectivity distribution, reconstructed according to the invention in the joint domain of height/displacement velocity, for a resolution cell in range- azimuth containing three isolated scatterers at different heights in layover, with different displacement velocity, obtained with two-dimensional separation by a resonating two-dimensional filtering with an adaptive method, with fixed diagonal loading, computed by multi-pass multibaseline data with one receiving channel for each pass, according to the configuration of figure 3a and multiple looks, simulated by a computer;

- Figure 6 shows an example of a reconstructed differential tomographic image according to the invention for a resolution cell in range-azimuth containing a volumetric scatterer, with line-of-sight displacement velocity increasing with the height, obtained like figure 5 but using three receiving channels for each pass according to the configuration of figure 3b; - Figure 7 shows an example of a radio reflectivity tomographical profile, normalized at the maximum value, reconstructed according to the invention in the height domain, robust, having a temporal signal decorrelation, for a resolution cell in range-azimuth containing a distributed surface scatterer affected by temporal decorrelation for change of the inner position of equivalent scatterers, with null average displacement velocity, obtained with two-dimensional separation by a resonating two-dimensional filtering phase with an adaptive method, with fixed diagonal loading, and integration of the differential tomographic image along the domain of the displacement velocity, computed by multi-pass multibaseline data with one receiving channel for each pass according to the configuration of figure 3a and multiple looks, simulated by a computer, and comparison with radio tomographical reflectivity profile in height obtained with classic Fourier monodimensional processing, in the same conditions. Description of the preferred embodiments.

In the following description an example will be illustrated of a method for processing imaging radar data of multi-pass multibaseline type, according to the invention, for creating images of radio reflectivity distribution in the joint domain of height/displacement velocity - and/or other kinematic parameters and/or parameters defining temporal evolution of distributed and/or multiple scatterers (such new type of images being called also differential tomographic images or generalized differential tomographic images), and/or of height tomographic reflectivity profiles which are robust to phenomena of temporal decorrelation and/or not rigid displacements, and/or of height tomographic coherence profiles, and/or of profiles of displacement velocity and/or of other kinematic parameters and/or parameters of tomographic temporal evolution in height, and/or for extracting multiple height parameters, and/or displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution, and/or reflectivity, and/or temporal coherence, and/or of DEM robust to phenomena of temporal decorrelation.

By forming differential or generalized differential tomographic images, the method according to the invention is, in particular, capable of providing information as complete as possible on a distributed scatterer of volumetric type during an even not stiff displacement and/or with temporal decorrelation, preferably considered to represent a mostly general and critical case of a scatterer for these interferometric techniques.

In figure 2 an example is shown of a distributed scatterer of volumetric type during a not stiff displacement, consisting of several equivalent scatterers 41-44 and 50-54, with respective components of different line-of-sight displacement velocity shown by vectors 45-47 and 55-57. Said scatterer is located in a space corresponding to a resolution cell 40, defined by its range 16 and azimuth 17, and the above described space is defined in height along the normal height 18.

The method according to the invention is described by the block diagram of figure 4, and by that of figure 4b, including also optional steps. It starts from a data acquisition step 20, figure 4, of imaging radar data of multi-pass multibaseline type, obtained acquiring one or more complex focalized radar images for each pass, consisting of one or more receiving channels for each pass. This acquisition step 20 corresponds to a sampling, normally sparse, in the baseline/acquisition time domain, the radar system effecting said acquisition forming a sparse spatial-temporal composed radar interferometer. Such multi-pass acquisition 20 that forms multiple baselines can be carried out with radar imaging systems. Advantageously, this radar system can be of synthetic aperture (SAR) type. In particular, the radar system can be of the type with a single receiving channel, or multichannel of multiantenna type co-located and/or distributed on more aircrafts or platforms, or multichannel of multiantenna type co-located and/or distributed on more aircrafts or platforms with commutation of the transmitter, so-called ping-pong technique in case of two antennas, forming one, at least two, or at least three complex images for each pass, respectively. The radar system can be transported by an aircraft or more aircrafts or platforms, preferably an avionic or space platform, such as an aircraft or a satellite, or a flight of aircrafts and/or other multiple avionic platforms, or a cluster of satellites, or one or more ground based motorised rail systems. The radar imaging system can be also of passive type, for example based fixed on the ground or based on a cluster of satellites, with one or more receiving channels, or multiantenna, using radar pulses or in general radio signals coming from other systems, which are in movement unless the passive radar system is moving instead and is of SAR type.

The case of multi-pass multibaseline acquisition with one receiving channel for each of the successive phases is given as an example in figure 3a, whereas the case with acquisition with more receiving channels for each of the phases is given as an example in figure 3b, with reference to a radar system with three equispaced antennas or with two antennas and commutation of the transmitter. The two configurations of acquisition 60 and 61 of figures 3a and 3b, or of sparse spatial-temporal composed radar interferometer, or sparse two-dimensional sampling, are characterised in the orthogonal baselines 63 - acquisition time 62 domain, expressed in units normalized to the length of the minimum orthogonal baseline and to the minimum temporal delay between acquisitions, respectively, and referred to the first channel of the first acquisition.

Such acquisition step 20 of figure 4 is followed by a possible calibration step 21 of the multi-pass multibaseline data, for bringing them back to an ideal condition, depurating them from possible parasitic effects, for being computed in a successive preprocessing step 22 or 22b, and/or two-dimensional separation 23 of figure 4, or multidimensional separation 23b of figure 4b, and/or for creating the data necessary to the following definition step 26 and/or 26b of the complex response when these are not already known a priori.

Such calibration step 21 can comprise a step of co-registering said complex focalized radar images deriving from said acquisition step of multi- pass multibaseline data, defined as the action of causing the range-azimuth of the resolution cells to collimate for each focalized complex image detected in the different acquisitions. This coregistration can be made by known methods, and has the object of minimizing parasitic spatial decorrelation effects due to a possible not perfect coincidence of the resolution cells in range-azimuth relative to the corresponding pixels in the various images.

The above described calibration step 21 can comprise also a step of obtaining the geometric configuration of said multiple baselines relative to the cell of interest in range-azimuth and to a height of reference for this cell. This detection of the geometric configuration can be made by known methods, and has the object of determining the data necessary to the following step of definition the complex response, and possibly to a preprocessing step.

Furthermore, the calibration step 21 can comprise also a step of calibrating amplitude and/or phase to the various receiving channels, defined as the action of measurements and also of equalizing possible unbalancing of the sensitivity to the radio reflectivity and/or to the level of additive thermal noise and/or undesired phase shifts between various channels. The above described calibration of phase and/or of amplitude can be made by known methods, and has the object of determining the information necessary to the following step of definition the complex response, and/or of bringing the data back to a nominal condition according to which the complex response is defined.

Yet, the calibration step 21 can comprise also a step of compensating the effects due to a possible non-linearity of the geometry of said composed spatial-temporal interferometer and at the curvature of the electromagnetic radio echo wave fronts, so-called deramping, defined as the action of compensating the phase shifts at various receiving channels due to such geometric effects for the actual cell in range-azimuth and a height of reference, as well as to calculate the orthogonal baselines corresponding to the radar line-of sight for said cell and height of reference. This deramping step can be carried out through known methods, and has the object of bringing the data back to the ideal condition of linear geometry of the interferometer and so- called far field or straight wave fronts, preferably necessary or advantageous in some steps of the process, which are comprised in the following steps of preprocessing 22 or 22b and/or of definition the complex response 26 and/or 26b and/or of two-dimensional 23 or multidimensional 23b separation.

Finally, the calibration step 21 can comprise also a step of compensating parasitic phase shifts due to variation of the radio propagation velocity in the atmosphere and/or in the ionosphere during the multi-pass acquisition, so- called atmospheric compensation. This atmospheric compensation can be made by known methods, and has the object of bringing the data back to the ideal condition of constant radio propagation velocity for all the receiving channels. Calibrations of amplitude and phase can be obtained even with methods of autofocalization integrated in the separation step.

After the calibration step 21 a phase 22 or 22b can be provided of preprocessing the multi-pass multibaseline data in the baseline/acquisition time domain, or baseline/acquisition time-range, preferably calibrated for bringing said data in the most appropriate conditions, and/or for transforming them in a form adapted to the successive two-dimensional or multidimensional separation.

The preprocessing step 22b can comprise a step of minimizing parasitic spatial decorrelation effects due to variation of geometric conditions for optimal coregistration of the images vs. an examined height, subsequent to the effect of migration in range. A compensation of said migration in range can be made by a coregistration in variable range with the position set in the domain of interest, or with predetermined range intervals that divide this domain, according to predetermined geometric functions that can be deducted from said relative geometric configuration, preferably detected in the calibration step. Said step of minimizing said effects of spatial decorrelation, along with the step of separation 23b, is a three-dimensional processing in the baseline- acquisition time - range domain.

In case a step of compensation of the migration in range is carried out, according to this method, the multi-pass multibaseline data cannot be computed in block in the successive two-dimensional or multidimensional separation, and in possible other preprocessing steps, but they can be computed separately for each of the conditions of compensation of the migration in range vs. the height or the height interval.

Said step of minimizing parasitic spatial decorrelation effects can also be obtained carrying out the successive two-dimensional or multidimensional separation 23b by means of the at least three-dimensional processing of said data in a joint domain comprising the joint domain baselines - acquisition time - range, said at least three-dimensional processing comprising a refocalization in range jointly to said separation.

Another step that can be carried out in the preprocessing step 22b and 22 is the step of minimizing random scintillation effects, or speckle, of the radio echo coming from distributed scatterers or distributed multiple scatterers due to an interaction of the radio waves with the surface or volume microstructure of said scatterers. This step of minimizing speckle effects can be carried out by using multiple looks and also computation of the correlation matrix of the multipass multibaseline complex data through a coherent average on the multiple looks, on which the successive two-dimensional or multidimensional separation is based.

Such computation of the correlation matrix can be made through known methods, such as the so-called sampling estimation of the correlation matrix, as described below. By defining n_p the number of the multiple passes, n_c(i_p) the number of receiving channels for the i_p ^"th pass, and n_L the number of the look, where

is a column vector of complex elements, called "data vector", in which the corresponding pixels can be structured in the various focalized complex images for the i_L ^"th look, being yii_p, i_c, i_L) the pixel relative to the image formed in the i_p ^"th pass with the i_c ^'th channel.

The correlation matrix can be calculated as:

where H is the Ηermitian operator, or transposition and conjugation operator.

Furthermore, the preprocessing step 22 (and 22b) can comprise a step of minimizing ambiguity effects, in a successive two-dimensional or multidimensional separation, due to sparse sampling of the baseline- acquisition time domain carried out in said acquisition step, forming lateral lobes, abnormal in the PSF, in the joint domain of height/displacement velocity and/or of other kinematic parameters and/or parameters defining temporal evolution, by means of a two-dimensional interpolation of the data based on an a priori information on the support in the joint domain of height/displacement velocity and/or of other kinematic parameters and/or parameters defining temporal evolution, where the signal components are expected, or based on their average statistic reflectivity distribution in the height-displacement velocity domain.

Such two-dimensional interpolation can be made by means of direct extension of known monodimensional methods of interpolation based on an a priori information on the relative support or on the so-called spectral density.

In particular, said interpolation can be obtained advantageously by a linear transformation of the data Y₁U₁) = M₁YU₁) , where Y₁U₁) is a column vector of interpolated data, with structure similar to the data vector, relative to baselines and/or acquisition times and/or number of phases and/or number of channels for each pass, so-called interpolated, even different from those available relative to the data. Advantageously, a transformation matrix M₁ is defined for minimizing the quadratic modulus of the deviation between the complex response relative to the interpolated baselines and/or acquisition times and the complex response obtained from said transformation applied to the complex response relative to the available baselines and acquisition time, said square modulus of the deviation being cumulated or weighed and cumulated in said support in the height-displacement velocity domain. This transformation can be used also for extrapolation, to increase the resolution, and is applicable not only to data, but also to data correlations (the correlations could be also of data relative to multiple carrier frequencies).

The preprocessing step 22 (and 22b) can comprise also a step of minimizing said ambiguity effects by means of two-dimensional windowing methods, applied to the multi-pass multibaseline complex data, and/or to the interpolated data, and/or to the calculated correlation matrix in case of two- dimensional or multidimensional separation based on said matrix. Said windowing can be made by known methods.

The preprocessing step 22 (e 22b) can, furthermore, provide, in case of two-dimensional or multidimensional separation based on the correlation matrix and/or of preprocessing, comprising an extraction step of the number of multiple scatterers made by methods based on the eigenvalues of the correlation matrix, and use of a single look for keeping a full capacity of resolution in range-azimuth, a computation step of said correlation matrix through a coherent average on the baselines of identical length and/or on temporal ranges between acquisitions of identical duration, or on divisions of the data. This computation of the correlation matrix can be carried out after deramping and two-dimensional interpolation, and advantageously after two- dimensional windowing.

An example of computation of the correlation matrix in case of use of a single look is described below. By defining y = yd) the data vector, Y₁ = M_xy an interpolated data vector relative to uniform interpolated baselines and acquisition time, and y_I(iβ) a i_B ^"th division of said interpolated data vector with a structure similar to the data vector, relative to n_PIB successive interpolated acquisition times and n_CIB successive interpolated baselines, the correlation matrix can be calculated through a coherent average on divisions, as R_y = n^∑""^ Y_IUB)Y"_UB) , where n_B is the number of said divisions, even partially overlapped.

In case of a successive two-dimensional 23 or multidimensional 23b separation step, with adaptive method, and/or of a preprocessing step comprising an extraction step of the number of multiple scatterers made by methods based on the eigenvalues of the correlation matrix, the preprocessing step 22 or 22b can comprise a phase of stabilizing the correlation matrix, on which said methods are based for minimizing ambiguities effects and increasing the resolution in the separation, and for calculating the model order of the data, by means of diagonal loading of said calculated correlation matrix. This diagonal loading can be made through known methods, with fixed loading or adaptive loading.

The fixed diagonal loading can be carried out as R_{y cd} = R_y + f_cfσ^.i , where R_{y cd} is the correlation matrix with loaded diagonal, R_y is the calculated correlation matrix, f_cl is a non-negative fixed loading factor that is expressed versus a level of thermal noise, is the level of thermal noise, or its value in square mean modulus known a priori or from the calibration step, and i is the identity matrix. An object of said stabilization is to solve problems of numerical ill-conditioning of the adaptive method and/or to reduce its sensitivity to residue errors of calibration of the data and/or statistic errors in the computation of the correlation matrix, due to the limited number of the looks available and/or the number of baselines of identical length and/or of temporal ranges of identical duration on which to compute the coherent average, errors which, as well known, can cause effects of reduction of sensitivity to the reflectivity of the radio echo components or loss of so-called radiometric precision and/or problems of numerical ill-conditioning. Another object of said stabilization is to reduce problems of overestimation on the model of the data in the methods based on the eigenvalues of the correlation matrix. In case of an adaptive diagonal loading, which depends also on the examined height and displacement velocity, the correlation matrix with loaded diagonal cannot be computed as a block in the successive two-dimensional or multidimensional separation, but can be computed separately for each of the considered conditions of diagonal loading responsive to the height and the displacement velocity.

Advantageously, the choice of the loading factor, in case of fixed loading, or of possible factors affecting the rate of adaptivity of the adaptive loading, is influenced, in case of successive two-dimensional or multidimensional separations with adaptive method, by a compromise solution between reduction of loss of radiometric precision and undesired reduction of the capacity of said method for minimizing ambiguities effects and increasing the resolution, and/or is influenced by data on the level of errors or residue calibration errors of the data, or on the precision obtainable in the calibration steps 21 , such as obtaining the geometric configuration of the multiple baselines, the calibration of amplitude and/or of phase, the atmospheric compensation. In case of preprocessing 22 or 22b comprising an extraction step of the number of multiple scatterers made by methods based on the eigenvalues, said choice is influenced by a compromise solution between a reduction of problems of overestimation and undesired accentuation of problems of underestimation.

The preprocessing step 22 or 22b can finally comprise a step of extraction of the number of multiple scatterers, or more in general a step of calculaton of a so-called model order of the multi-pass multibaseline data, possibli after calibration, and/or after other preprocessing steps. This phase of computation of the model order can be made by known methods, for example based on the eigenvalues of the calculated correlation matrix, preferably stabilized with diagonal loading, and has the object of determining the data necessary to the successive two-dimensional or multidimensional separation, when this is carried out with method based on models, for example with methods based on a model for multiple scatterers with radio echo components without temporal decorrelation of the class based on subspace decomposition, and this order cannot be considered known a priori, nor it is obtained within the separation step same, and/or obtains the information of the number of multiple scatterers, preferably distributed, in layover.

Advantageously, in case of computation of the order of the model made through methods based on the eigenvalues, with stabilization of the correlation matrix calculated by means of adaptive diagonal loading, said adaptive loading can be carried out as R_{y cd} = R_y + f_caσ_s ²τ , where f_ca is an adaptive not negative loading factor expressed according to the level of the signal, and σ_s ² is the level of the signal, or its value in square mean modulus, determinable for example by means of sampling estimation.

The method according to the invention can comprise, furthermore, a step of definition a nominal value of complex response 26 of the composed spatial- temporal radar interferometer to the radio echo coming from a single scatterer, point-like (or distributed with null extension of the tomographical reflectivity profile in normal height), at a specific height and having a specific line-of-sight displacement velocity, in absence of noise and temporal decorrelation. This complex response has the purpose of describing the equation between the generic component of the radio echo from distributed and/or multiple scatterers and the corresponding component in the multi-pass multibaseline data, preferably necessary in the successive two-dimensional separation step when this is obtained by a method that uses said equation. This step of definition a value of complex response 26 can be carried out even with reference to interpolated baselines and acquisition time, for describing said equation referred to the interpolated data, preferably necessary in the preprocessing step when this comprises an interpolation step, and in the successive two- dimensional separation step, when this is obtained by a method that uses said equation referred to the interpolated data.

Such complex response definition 26 can be obtained by computation of the values of amplitude and phase of the complex valued pixels corresponding in the various focalized images, originated nominally by a single radio echo component, vs. height and displacement velocity of the relative scatterer in a two-dimensional domain. It can be structured in a vector called "steering vector" of complex elements, corresponding to the channel or to various receiving channels for each pass of the multiple passes. As particular case, in case of far field and linear geometry of the composed spatial-temporal interferometer, to which the data can be given by the deramping step, this complex response comprises the multi-pass multibaseline complex data, corresponding nominally to a specific spatial frequency and to a specific linear distribution of an interferometric phase temporal component responsive to the temporal delay between said acquisitions, which corresponds to a specific "temporal frequency", i.e. to a specific two-dimensional spatial-temporal harmonic. This equation between components of the data and parameters domain corresponds to a Fourier equation between data and parameters domain (equivalents, as a convention, the Fourier equation can be considered between the parameters domain and the data, without affecting the nature of the separation techniques owing to the duality between the Fourier transform and the reverse Fourier transform).

The complex response is calculated by the geometric configuration of the multiple baseline, relative to the cell in range-azimuth of interest and at a height of reference for this cell, and by the time required for the multi-pass acquisition, for a frequency of the so-called radar carrier, and for nominal radio propagation velocity and then radio wavelength, all known. Said geometric configuration and/or acquisition time can be the actual one, known a priori or detected in the calibration step, or the ideal one, which the data can be referred to in the deramping and/or interpolation steps.

In case the step of deramping has been carried out, the complex response can be calculated as described below. b{i_{p l} i_c) is defined as the length of the base orthogonal between the i_c ^th receiving channel for the i_p ^"th pass and a reference channel, for example the first of the first pass, such that b[i, l) = 0 , and t(i_p) is the temporal delay between the i_p ^th pass and a reference pass, for example the first for which t(i) = o . Such lengths of orthogonal bases are here defined in a so-called bistatic geometry with two paths or between bistatic equivalent channels, thus being equal to a half of the geometric actual lengths if one or both channels are not used in the generation of the pulses from whose echoes returning to the channels the relative images are formed.

The spatial and temporal frequencies can be expressed respectively as f_s = 2q_N J Ud_x) and f_τ = 2v_SR I λ , where q_N is the normal height referred to the height used as reference in the deramping step, λ is the wavelength, d_R is the slant range, and v_SR is the line-of-sight displacement velocity.

By using a complex vector with the same structure of the data vector, the complex response can be calculated as

where j is the imaginary unit.

Advantageously the computation of the complex response, preferably necessary in the subsequent multimensional separation step and/or in the preprocessing step when this comprises an interpolation step, can be carried out also in a step of definition a nominal value of complex response 26b including kinematic parameters and/or parameters defining temporal evolution, such as the line-of-sight acceleration a_SR , and/or the line-of-sight acceleration derivative c_SR , and/or line-of-sight step displacement δ_τ and/or the step amplitude relative variation 4 at the temporal delay L₁ from the reference pass, said parameters with the height and displacement velocity changing in an actual multidimensional domain. Using again a complex vector with the same structure of the data vector, the element of the complex response a(f_s, £-, O₁, γ_τ, δ_τ, £., t,,) corresponding to the pixel relative to the i/^th pass and to the i_c ^th channel can be calculated for example as

where α_τ is the so-called chirp rate, γ_τ the rate of change of the chirp rate, u(-) the step function, N_a{ξ_τ, L₁) =\ \ a(f_s, f_τ, α_τ, γ_τ, δ_τ, ξ_τ, t.) | | , with | | ^• | | the operator norma, considering the scaling functions α_τ = 2a_SR / λ , Y_T - ^2C _SR I ^λ • and ^r = ^2d _sR I ^λ ■ The harmonic temporal component of the complex response is thus generalizable to a signal having polynomial phase or more in general with non-linear distribution of temporal phase component and possible variation of amplitude. Advantageously, said step of definition the complex response 26 and/or

26b can include computing an effect of refraction of the radio propagation in volumetric scatterers, so-called dense, such as layers of ice, or the folding of the paths and the variation of displacement velocity of propagation of the radio echo in the dense medium, in particular, to the interface with the propagation space so-called free or the atmosphere. Such calculations are made according to known functions.

In case of preprocessing comprising an interpolation step, the complex response relative to the baselines and interpolated acquisition times a_τ(f_s# f_τ) can be calculated as described below. By defining Jb₁(I_p, i_c) the lengths of the orthogonal interpolated bases, and Jz₁(I_p) the temporal interpolated ranges, referred like as described for b{i_pf i_c) and t(i_p) but relatively to the baselines and/or acquisition times and/or number of phases n_PI and/or number of channels for each interpolatedpass π_CI(i_p) .

By using a complex vector with the same structure of the interpolated data vector, said complex response a^fj., f_τ) can be calculated as:

J2πf_sb_τ (l, 2) [I) )

j2πf_sb_I {n_PI , 1) + j2πf_Tt{n_PI ) j2πf_sb_I (n_{P 1} , 2) + j2πf_τt(n_{P ι} )

j2π£_sb_τ (n_Pι , n_CI {n_PI ) )+ j2πf_Tt(n_pl )

In case of two-dimensional separation based on a correlation matrix calculated after interpolation using a single look, with interpolated data relative to uniform baselines and interpolated acquisition time , the complex response a_IB(f_s, f_τ) that can be used at the separation can be obtained as a partition of said a_x(f_s, £.) with structure similar to 3L₁If₈, f_τ), relative to n_PIB successive interpolated temporal intervals and n_CIB lengths of the successive orthogonal interpolated bases.

An example of definition of the transformation matrix M₁ for two- dimensional interpolation of the data in the preprocessing step is described below. By defining s_fsιtt the support in the domain of the spatial and temporal frequencies corresponding to the support in the domain of height/displacement velocity where the components of signal are expected, M₁ can be calculated as M₁ = arg min_M ff | | A₁^f₃, f_τ) - Ma(f_s, f_τ) | |² df_sdf_τ . The least squares solution of this problem can be obtained by known methods, which can include a stabilization, after the discretization of the integral.

The method according to the invention carries out then a two-dimensional separation step 23 of multiple radio echo components in the height/displacement velocity joint domain, by means of two-dimensional processing in the baseline/acquisition time joint domain of the multi-pass multibaseline complex data, even sparse data, preferably calibrated and/or preelaborated, using a single look or multiple looks for corresponding pixels in the various focalized complex images acquired, or the corresponding correlation matrix by them calculated, since the two-dimensional processing can be based on the complex response calculated in the relative phase of definition.

The method according to the invention can carry out also a multidimensional separation step 23b of said components in a joint domain comprising, in addition to the height and displacement velocity, also kinematic parameters and/or parameters defining temporal evolution of said components, by said two-dimensional processing in the baseline/acquisition time joint domain, using said or each look, or the corresponding said correlation matrix, since the two-dimensional processing can be based on said complex response.

Such two-dimensional separation step 23 has the object to obtain information on the components distribution of distributed and/or multiple scatterers in the height/displacement velocity joint domain for an actual range- azimuth resolution cell, and/or more in particular, to obtain the radio reflectivity distribution in the height/displacement velocity joint domain for said cell. Similarly, this multidimensional separation step 23b has the object to obtain said data, and/or more in particularsaid distribution, in a joint domain of parameters of which at least one spatial parameter and two or more kinematic parameters and/or parameters defining temporal evolution.

The two-dimensional separation step 23 can be obtained from a processing technique chosen among, or obtained from combinations of, spatial-temporal two-dimensional spectral analysis in the baseline/acquisition time joint domain, with sampling also sparse, or two-dimensional resonant or passband filtering in said domain, or forming a two-dimensional hybrid synthetic beam (so-called beamforming) in the joint domain of height/line-of- sight displacement velocity, or so-called 2D array processing for so-called DOA 2D, where one of the two spatial directions of arrival is formally replaced by said displacement velocity parameter, in suitable scales according to the complex response calculated in the relative phase of definition. Between the used techniques of processing there is also the reconstruction of images by means of variational methods.

Similarly, the multidimensional separation 23b can be obtained from said processing techniques, in particular, by making beam and/or array processing, with an output in a joint multidimensional domain comprising in addition to the height and displacement velocity also kinematic parameters and/or parameters defining temporal evolution.

Advantageously said technique chosen for the two-dimensional separation step 23 can be carried out by a method foofr two-dimensional analysis, preferably at high resolution and/or ambiguity suppression, selected from the group comprised of: one of the methods described below, or obtained from a combination thereof. Among the methods used at the separation step 23, there are the methods based on Two-dimensional Fourier transform, preferably called irregular in case of sparse sampling, or so-called two- dimensional beamforming. Then, there are the methods of spectral analysis or passband filtering or adaptive two-dimensional beamforming, which have the object of reducing ambiguities from sparse data and increasing the resolution by means of adaptive cancellation of the not examined components, exploiting the data in the calculated correlation matrix, preferably stabilized by means of diagonal loading for reducing loss of radiometric precision and/or problems of ill-conditioning. Said ambiguities can be reduced also using apodization methods. an example of two-dimensional spectral analysis spatial-temporal or two- dimensional filtering resonant or passband baseline/acquisition time joint domain or two-dimensional beamforming of data to multiple looks after deramping step is described below, said f(f_s, Z₁) a complex vector of filtering coefficient spatial-temporal with the same structure of the data vector, the radio reflectivity distribution in the spatial frequency-temporal frequency joint domain can be calculated as:

Rf₃₁ Z_r) f^*(f_s, f_r)R_yf(f_s, f_T), where | • | is the operator of modulus and R_y is the calculated correlation matrix by means of estimation campionaria. By using an adaptive method, and being a(f_s, f_τ) the complex response structured as the data vector, the vector of the filtering coefficients can be calculated as: f(f_{s /} Z, ) = R-_y ^λ _cdaL(f_s , Z₁) I (a"(f_s , Z₇)R;¹^ f_s , Z₁ ) ) , where there is a dependence, in addition to the complex response, also from the data same through the calculated correlation matrix, preferably with fixed diagonal or adaptive loading, R_ycd .

Advantageously, in case of adaptive loading, said adaptive loading can be carried out as R_{y cd} = R_y + Mf₈, Z₁, R_y)i, where the value of the loading λ(f_s, Z₁, R_y) is the solution of the inequality | | f(f_s, Z₁) | |²< f_va , obtainable by known methods, where f_va is a factor of constraint of adaptive positive loading, affecting the rate of adaptivity of the adaptive loading, which in this case depends also on the examined height and displacement velocity. In case of use of a single look after interpolation, the radio reflectivity distribution in the spatial frequency-temporal frequency joint domain can be calculated as r(f_s, f_τ) = f"(f_s, f_T)R_yf(f₃, t,), where f(f_s, f_τ) has the same structure of the partitions of the interpolated data vector, and R_y is the

5 correlation matrix calculated through a coherent average on partitions of the interpolated data vector, and using an adaptive method f(f_s, f_τ) can be calculated as: f(f_s, f_τ) = R^a₁₈(JP_s, Z₁) I (a^H _IB(f_s, f_r)R^~ _y ^l _cda_IS(f_s, Z₁) ) , where R_{y cd} is called calculated correlation matrix through a coherent average on partitions,o preferably with fixed or adaptive diagonal loading.

The radio reflectivity distribution r{f_s, Z₁) can be expressed in the joint domain of normal height/line-of-sight displacement velocity by the scaling functions g_N = λd_Rf_s I 2 and v_SR = λf_τ / 2 . 5 Advantageously, said techniques of making a synthetic beam, in particular, adaptive beamforming, can be used in an extended version with an output in a multidimensional domain for carrying out a multidimensional separation step 23b.

An example of multidimensional separation of multiple radio echo0 components in a joint domain comprising in addition to the height parameters and displacement velocity also kinematic parameters and/or parameters defining temporal evolution is described below. The radio reflectivity distribution in the joint domain comprising spatial frequency - temporal frequency - chirp rate - rate of change of the chirp rate - step phase variation -5 step amplitude relative variation - temporal delay between the step displacement or phase variation and its reference pass r{f_s, f_{τ l} a_τ, γ_τ, δ_τ, ξ_τ, t_τ) can be calculated as f"(f_sl f_τ, a_τ, r_τ, δ_{τ l} ξ_τ, ^)R _yf (Z₃, f_{τ l} a_τ, γ_τ, δ_τ, ξ_τ, t_τ) , where, when using an adaptive method, the vector of the filtering coefficients for beamforming £(f_s, f_τ, a_τ, γ_τ, δ_τ, ξ_τ, £.,) can be calculated as:

R;^x _cda(f_s, Z₁, a_τ, γ_τ, δ_τ, ξ_τ, t,) / (a^H(f_s, f_τ, a_τ, γ_τ, δ_τ, ξ_τ, ^R^aif_g, Z₁, α_τ, γ_τ, δ_τ, ξ_τ, t_.) ) and the element of the complex response a(f_s, f_τ, α_τ, γ_τ, δ_{τ l} ξ_τ, t_τ) corresponding to the pixel relative to the i_p ^"lh pass and to the i_c ^'m channel can be calculated as: 5 _ei2πε_sb^{i_p,i_c ⁽i_? ^{) )}+j2π£fi⁽i_p ⁾+jmx₁t^:'(i_p)+jπ3-^ιr,t*^{i_p ⁾ + i2π&_Iu⁽tti_p ⁾-t_r ⁾Q ₊ μ u(t(l ) - fc-) ) / N (£. t~)

The use of an adaptive method has the object in this case of reducing also the intrinsic ambiguities in this multidimensional separation that corresponds substantially to a decomposition of the signal on a supercomplete base. The separation in a joint domain comprising only some of said kinematic parameters and/or parameters defining temporal evolution can be obtained using similar expressions obtainable by setting the other of said other parameters equal to zero and/or to known values, the latter case in particular, for said elapsed temporal delay.

The radio reflectivity distribution obtained can be expressed in the joint domain responsive to the line-of-sight acceleration and/or the line-of-sight acceleration derivative and/or line-of-sight step displacement by the scaling functions a_SR = α_τλ / 2 , c_SR = γ_τλ / 2 and d_SR = δ_τλ [ 2.

Furthermore, among the methods used in the two-dimensional separation step 23 there are the methods of spectral analysis or 2D array processing based on models or on parametric methods, which have the object of reducing ambiguities and increasing the resolution and advantageously the precision by means of the so-called identification or "fitting" of the model, exploiting the a priori information on the expected kind and functional shape of the radio reflectivity distribution in the height/displacement velocity joint domain.

Said models can be specified for multiple point-like or distributed scatterers with null extension of the tomographical reflectivity profile in normal height, with distribution of line-of-sight displacement velocity for the radio echo component at a certain height of null extension, or radio echo component without temporal decorrelation, the corresponding parametric methods being the class based on splitting intoto subspaces of the correlation matrix, or on alternated iterated estimation and cancellation, and/or least squares amplitude fitting. Said models can, furthermore, be specified for surface distributed scatterers or volumetric or multiple distributed scatterers, and/or with said distribution of displacement velocity at a not null extension height, or radio echo component with temporal decorrelation, fixed (decoupled) or variable (coupled) with the height of the various radio echo components, the corresponding parametric methods being of a class based on identification of the model obtained by means of fitting the correlation matrix, or based on models of two-dimensional autoregressive type (AR 2D), or autoregressive moving average type (ARMA 2D).

Furthermore, the parametric methods of said classes can be used for the separation also when the models on which they are based are not specific but approximate the model for distributed and/or multiple scatterers, with said distribution of displacement velocity of null extension and/or not null extension, without or with temporal decorrelation. Two examples of two-dimensional spatial-temporal spectral analysis or 2D array processing, made through model-based methods, of the classes based on decomposition into subspaces of the correlation matrix and least squares fitting for amplitudes, are described below, in the same order. 5 By defining d^ the size of the calculated correlation matrix, n_R the order of the model, V_1^ the i/^th among the d_R - n_R autovectors corresponding to the d_Ry - n_R smaller eigenvalues of the calculated correlation matrix, w_Xv the i/^th among d_R - n_R autovectors, in particular of unitary value, the fitting rate of the model vs. the spatial frequencies and time can be calculated aso g_f(f_s, £_r) = i / ∑^^{y ~}"^R ^_v I a." (f_s, E₁)V_1^ I² . This fitting rate in the spatial frequency-temporal frequency joint domain can be expressed as a joint distribution of height and line-of-sight displacement velocity by said functions of scaling. From the position of the n_R dominant peaks of the rate of fitting in the spatial frequency-temporal frequency joint domain, or in the5 height/displacement velocity joint domain, the corresponding parameters for n_R multiple scatterers can be obtained.

By defining f_SiR and f_TiR ^tne spatial and temporal frequencies corresponding to the height and displacement velocity of the i_R ^"th multiple scatterer, extracted in a post-processing step or obtained directly by the0 chosen model, the complex amplitudes of the n_R multiple scatterers for the i_L ^"th look can be calculated as: α_tf_t) _< - ^• ' ά_nR{i_L) = arg min^, _{, ,}^, | | γ(i_L) - ∑"^* ₌₁ «_lR(i_L)a(f_SlB , f_TlR ) | |² .

The solution of this problem at the least squares can be obtained by known methods, preferably including a stabilization. The radio reflectivity of the n_R multiple scatterers can be obtained by averaging on the n_L looks the square modules of the corresponding complex amplitudes.

The number of multiple scatterers, or more in general the order of such models, can be chosen as known a priori, or can be obtained during the preprocessing, or during the separation when identifying the model or jointly to the calculation of its parameters, advantageously according to the known methods.

An example of extracting the number of multiple scatterers in the two- dimensional separation step during the identification of the model, based on least squares amplitude fitting, is described below. By starting from an order n_R hypothetical^ initialized at 1, the radio reflectivity is calculated of the hypothesized π_κ multiple scatterers by said least squares identification. The spatial and temporal frequencies of the hypothesized n_R multiple scatterers used in said identification are extracted in a post-processing step or obtained directly during the separation by means of model-based methods, for example of a class based on subspace decomposition, correspondingly to said hypothized n_R order. According to a thresholding test of the calculated reflectivity, said order n_R hypothesized at 1 is increased and the described identification procedure is repeated, or is declared as number of multiple scatterers once extracted said order n_R which is hypothesized decreased of 1.

Advantageously, said test can be replaced by or associated with a thresholding test of the error in square modulus of fitting by the least squares for the amplitudes | | y(i_L) - Y^%_ ά_x (i_L)a(fL _/ f_τi ) I f . advantageously divided by

I I y(ij I |² , mediated on the n_L look.

In case of said two-dimensional spectral analysis or said 2D array processing made through model-based methods specified for volumetric scatterer or distributed multiple scatterers, of the class based on said identification by means of fitting the correlation matrix, the equation between the parameters of the model and the corresponding nominal correlation matrix of the multi-pass multibaseline data, preferably calibrated and/or preelaborated, preferably used for fitting with the calculated correlation matrix, can be obtained analytically and/or numerically by inverse two-dimensional Fourier transform of the radio reflectivity distribution in the spatial frequency- temporal frequency joint domain or height-displacement velocity joint domain, with suitable scalings, corresponding to said parameters, preferably exploiting the complex response calculated in the relative phase of definition, this inverse transform providing the function of spatial-temporal two-dimensional correlation , of whose samples said nominal correlation matrix is constituded.

An example of method for the class based on identification by means of fitting the correlation matrix, is described below. By defining p_sra a vector in which the parameters can be structured of the specified model of radio reflectivity distribution in the spatial frequency-temporal frequency joint domain, for example parameters of spatial frequency, extension or band of spatial frequency, temporal frequency, extension or band of temporal frequency or correlation time, variation of displacement velocity in height, radio reflectivity, attenuation of reflectivity in height, characterizing the or each scatterer for which the model is specified. This vector of parameters can be calculated as p_STR = arg min_Psτ8 | | w^{1 / 2}(R_y - R_y(p_sra) )W^{1 / 2} | |² , where R_y(p_srR) is the nominal correlation matrix corresponding to said model of reflectivity distribution xif_s, f_τ, P_SΓR) , w is a matrix of weights, w^{1 / 2} is its Hermitian root square, and advantageously w = R^ . With reference to the structure of the data vector, after defining i_pr the index of the pass and i_Cr the index of the channel relative to the r^"th pixel in the element of the data vector, and i_Pc and i_Cc the analogue indexes relative to the c^'th pixel or element, the element of the nominal correlation matrix at the row r and column c [R_y(p_sra) ]_rtC can be expressed as

IVPSTJ U _τ {-} is the opposite two-dimensional Fourier operator calculated for values β and T in the inverse transform domain. Yet, among the methods used in the two-dimensional separation step 23, there are the methods of Fourier or two-dimensional beamforming followed by deconvolution two-dimensional methods, applied to the joint domain of the spacial-temporal frequencies or of height/displacement velocity, which have the object of reducing ambiguities by partially compensating the so-called convolutive distortion effect that is caused by the not ideal shape of the hybrid two-dimensional synthetic beam in the height/displacement velocity joint domain, or by the presence of lateral abnormal lobes in the PSF in said joint domain, owing to sparse sampling, which distortion affects the reconstruction of the actual radio reflectivity distribution in the height/displacement velocity joint domain. Said deconvolution methods can be maximum entropy deconvolution methods. Furthermore, among the methods used, image reconstruction methods are used by means of variational methods, in particular image reconstruction methods with maximum entropy.

Similarly to the above described examples, for a beamforming method in particular, adaptive, the technique for a multidimensional separation 23b can be made by one or more of said methods, in particular, methods of a class based on said subspace decomposition, and/or on alternate iterated estimation and cancellation, and/or on identification of the model by means of fitting the correlation matrix, with an output in a joint domain comprising in addition to the height and displacement velocity also kinematic parameters and/or parameters defining temporal evolution, and/or least squares identification for the amplitudes with input from said joint domain, and/or multidimensional deconvolution after beamforming with input and output from and into said joint domain. In order to minimize parasitic spatial decorrelation effects by migrations in range, the multidimensional, in particular two-dimensional, separation step can be obtained also by means of three-dimensional processing in step 23b, in a joint domain comprising in addition to the baselines and to the acquisition time also the range, of said data, which are preferably calibrated and/or preelaborated. In said processing, said data are previously subject to Fourier inverse transform in range, being thus refocalized in range jointly to said separation by said processing techniques (and methods) in three-dimensional version, in particular, three-dimensional spectral analysis, filtering resonant or three-dimensional passband, forming a beam with three-dimensional input, array processing with 3D input (in particular, obtained by methods of Fourier followed by deconvolution or apodization methods). In case of a two- dimensional separation, the output of the three-dimensional processing techniques is in the joint domain of height/displacement velocity jointly to the domain in range.

The application of the method according to the invention can bring to obtain, as indicated by reference 24, an image of components distribution of distributed and/or multiple scatterers and/or of reconstructed radio reflectivity distribution in the two-dimensional joint domain height/displacement velocity for a range-azimuth resolution cell of interest, or a differential tomographic image.

The application of the method according to the invention can bring furthermore, to obtain, as indicated by reference 24b, said distribution images and/or reflectivity distribution images in a multidimensional joint domain comprising also kinematic parameters and/or parameters defining temporal evolution, or a generalized differential tomographic image, preferably called chirp differential tomographic image and/or step in case of line-of-sight acceleration parameters and/or variation rate of line-of-sight acceleration and/or line-of-sight step displacement and/or step phase variation and/or step amplitude relative variation and/or temporal delay between the variations and the reference pass.

In case of three-dimensional processing, said image of distribution and/or of reflectivity distribution in the two-dimensional and/or multidimensional domain is obtained for each range of interest.

The reflectivity, height, displacement velocity, and kinematic parameters (and/or parameters defining temporal evolution) can be expressed also in equivalent scales, such as amplitude or intensity or so-called radar equivalent section that can be normalized with respect to the components of the scatterers, normal or vertical height, displacement velocity and kinematic parameters relative to displacements according to radial or vertical directions or in horizontal direction, respectively, and the distribution can be displayed in various equivalent graphic forms. In case of two-dimensional 23 or multidimensional 23b separation carried out with method of analysis based on models, said image of components distribution and/or of reflectivity distribution or differential tomographic image or differential tomographic generalized image is obtained from the chosen model or evaluated by parameters calculated in the fitting.

An example of differential tomographic image 70 is depicted in figure 5, which shows the radio reflectivity 71 in arbitrary unit and scales in deciBel (dB), reconstructed vs. height 72 and displacement velocity 73 in a predetermined domain, expressed in so-called Rayleigh resolution units in height and in so- called Fourier resolution in displacement velocity, i.e normalized, with the height referred to the height of reference used for the deramping step. The resolution cell in range-azimuth for which this differential tomographic image 70 is obtained contains three layover isolated scatterers (or perspectively overlapped) at normalized heights equal to 0, 1.5 and 3, with normalized line-of- sight displacement velocity equal to 0, -1 (where the negative singns indicates a centrifugal movement ) and 0 respectively, and reflectivity referred to the level of thermal noise, or signal-noise ratio, equal to 15, 12 and 9 dB respectively. Such scatterers are of distributed type with null extension of the tomographical reflectivity profile in normal height of each, and each scatterer is not affected separately by temporal decorrelation.

The tomographic-differential image 70 is obtained in conditions of calibration comprising deramping with two-dimensional separation through a resonating two-dimensional filtering step with adaptive method, with diagonal loading of the correlation matrix calculated with fixed loading factor equal to 10, computing multi-pass multibaseline data with one receiving channel for each pass with the configuration of acquisition shwn in figure 3a and sixteen looks, simulated by a computer. In the differential tomographic image 70 three dominant reflectivity peaks 74, 75, 76 are visible, with heights and positions in the height/displacement velocity joint domain corresponding with good approximation to the actual values of reflectivity, height and displacement velocity of the three layover scatterers. Furthermore, the maximum ambiguities level 77 is satisfactorily low, 11 dB less than the reflectivity of the weakest scatterer, and the resolution both in height and in displacement velocity is better than that of Rayleigh and of Fourier, respectively. Another example of differential tomographic image is the differential tomographic image 80 depicted in figure 6 that is shown with respect at a same height of figure 5. The resolution cell in range-azimuth for which this differential tomographic image is obtained contains a volumetric scatterer extending between the normalized position -3.5 and 3.5, with normalized line-of- sight displacement velocity increasing in a negative direction with the height from 0 to -3.5 according to a root square profile, overall ratio signal-noise at 16 dB, with reflectivity decreasing linearly with the height of 2 dB for unit for normalized height, up to the component having zero displacement velocity that has the highest reflectivity, with a ratio signal-noise of 6 dB. The various components of the volumetric scatterer are not affected separately by temporal decorrelation.

The tomographic-differential image is obtained in conditions like those defined in figure 5, except from the configuration of acquisition that is that with three receiving channels for each pass shown in figure 3b. In the differential tomographic image 80 the following are shown: reflectivity peak value 81 of the static component with minimum height and the course of reflectivity 82 increasing along the profile of displacement velocity which decreases with the height, corresponding with good approximation to the effective profile. Said differential tomographic reconstructed image 24, in particular, unless obtained from two-dimensional separation carried out with a method of analysis based on models, can then be computed by a possible postprocessing step 25, shown in figure 4, in the height-displacement velocity domain, for extracting additional data on the distributed and/or multiple scatterers from said image.

Such post-processing step 25 can comprise an extraction step 27 of the multiple parameters of height and/or displacement velocity and/or radio reflectivity of multiple scatterers, possibily even distributed, in layover, preferably made by extraction of the position and/or height of dominant peaks in the differential tomographic image. The multiple parameters of height and/or displacement velocity can be extracted also by computation of the centroids of the radio reflectivity distribution in the joint domain of height/displacement velocity restricted to neighbourhoods of said peaks, and the multiple parameters of radio reflectivity can be extracted also by integration of the radio reflectivity distribution in the joint domain of height/displacement velocity restricted to said neighbourhoods, preferably normalized to a coefficient, preferably dimensional, that may be equal to the product of the Rayleigh and Fourier resolutions.

The post-processing step 25 can comprise furthermore, an extraction step of the number of multiple, possibily distributed, scatterers, in layover, preferably made by a thresholding test of the extracted eflectivity parameters.

Furthermore, the post-processing step 25 can comprise also an extraction of a so-called marginal radio reflectivity distribution in the only height domain from the differential tomographic image, in order to obtain a tomographic reflectivity profile in height that is robust with respect to a temporal signal decorrelation 28, or in order to reduce virtually phenomena of temporal coherence loss. Such marginal radio reflectivity distribution extraction can be made by integration of the reconstructed differential tomographic image in the height/displacement velocity joint domain along the domain of the displacement velocity. For reducing the cumulated effect of the so-called background of the differential tomographic image, the integration along the displacement velocity domain can be carried out preferably after the phase of so-called settingof values of reflectivity distribution less than a predetermined threshold, advantageously at zero, and/or with restriction of the interval of integration to a fixed interval of displacement velocity, or of predetermined extension centred about the position of the maximum or of the monodimensional centroid of the reflectivity distribution restricted to each of the various heights comprised in the domain of interest for which the marginal radio reflectivity distribution is extracted, or working on a reflectivity distribution, called fixing or windowing, respectively.

With reference to the normal heights and to the line-of-sight displacement velocity, in case of windowing, for example on a interval of line-of-sight displacement velocity fixed between -v_{SR max} to v_{SB max} , the marginal radio reflectivity distribution f(g_w) can be calculated as: r(q_N) = υ^'1 J^ r(q_N, v_SR)dv_SR _ where the integral can be evaluated numerically and υ is preferably a dimensional coefficient that can be set equal to the Fourier resolution in line-of- sight displacement velocity.

Said extraction of marginal radio reflectivity distribution can be carried out also by extraction of a monodimensional maximum of the differential tomographic image restricted to each of the various positions set in the domain of interest for which the marginal radio reflectivity distribution is extracted. Said robust tomographic reflectivity profile in height can be displayed in various scales and equivalent graphic forms.

Yet, the post-processing step can comprise also an extraction step of multiple parameters 27 of height and/or radio reflectivity of multiple, possibily distributed, scatterers, in layover from said robust tomographic reflectivity profile in height, to prepare said parameters that are robust vs. a temporal signal decorrelation, or for virtually reducing phenomena of temporal coherence loss. This step can be carried out by extraction of the position and/or height of dominant peaks in the robust tomographic reflectivity profile in height, the multiple height parameters can be extracted also by computation of the centroids of the marginal radio reflectivity distribution restricted to neighbourhoods of said peaks, and the multiple parameters of radio reflectivity can be extracted also by integration of the marginal radio reflectivity distribution restricted to said neighbourhoods.

In case of extracting multiple parameters of height and/or radio reflectivity of multiple scatterers, possibily even distributed, in layover from the differential tomographic image, said parameters can be obtained in a way that is already inherently robust with respect to a temporal signal decorrelation, since the two- dimensional separation has discarded undesired temporal effects in the extraction of information in the height domain.

Said extraction of the multiple parameters of height and/or displacement velocity and/or radio reflectivity of multiple scatterers, even distributed, from the differential tomographic image, and of extracting multiple parameters of height and/or radio reflectivity of multiple scatterers, possibily even distributed, from the robust tomographic reflectivity profile in height, can be carried out as particular case also for the case of a single scatterer, either a point-like or a surface or a volumetric scatterer, in particular for extraction of a single height parameter, with the object to obtain a robust DEM having a temporal signal decorrelation 29, or for virtually reducing phenomena of temporal coherence loss.

Such extraction of a single height parameter to obtain a robust DEM can be carried out like the case of extracting multiple height parameters, starting both from the differential tomographic image and from the robust tomographic reflectivity profile in height, with reference to the only highest peak value and advantageously even at the corresponding centroid relative to the corresponding neighborhood. The post-processing step 25 can comprise, furthermore, also an extraction step from the differential tomographic image of measurements relative to the temporal coherence of the various components of volumetric scatterers or of multiple scatterers, preferably distributed, in layover, to obtain a profile of information of temporal coherence vs. height 31, called tomographical temporal coherence profile in height, or multiple parameters of information of temporal coherence 31 of the extracted multiple layover scatterers, called temporal coherence multiple parameters, thus providing more complete information with respect to the known temporal coherence measurements relative to the only whole of the volumetric scatterer or of the multiple scatterers.

Such extraction of measurements relative to the temporal coherence 31 of the various components of volumetric scatterers can be made by computation of an measurement of extension, called of displacement velocity "band" By , of the radio reflectivity distribution in the joint domain of height/line- of-sight displacement velocity restricted to each of the various heights set in the domain of interest for which the measurements relative to the temporal coherence of the various components are extracted. This displacement velocity band can be defined according to known conventions on bands, for example the so-called -3 dB band. From the displacement velocity band, calculated for various position set in the domain of interest, then the corresponding band of temporal frequencies can be determined as and from this a measurement of the so-called correlation time τ_c, according to functions taken from known conventions, for example τ_c = l / (2πB_T) .

Such correlation time can also be converted into a measurement of temporal coherence referred to a fixed temporal delay, using a known model of temporal decorrelation course that is chosen a priori.

The extraction of measurements relative to the temporal coherence of the various components of volumetric scatterers, both vs. correlation time and temporal coherence referred to a fixed temporal delay, can be carried out also by extraction of an actual temporal decorrelation course of the various components of volumetric scatterers at the various heights set in the domain of interest. This course can be calculated by Fourier reverse transform of the radio reflectivity distribution in the height/displacement velocity joint domain restricted to each of the various heights set in the domain of interest and with the displacement velocity expressed in the temporal frequency scale. Such displacement velocity band or a temporal decorrelation course actual can be calculated preferably acting on reflectivity distribution with fixing and/or with windowing. Said tomographical temporal coherence profile in height can be displayed in various scales and equivalent graphic forms.

The extraction of measurements relative to the temporal coherence 31 of the various components of multiple layover scatterers can be made by means of procedures similar to those above described, applied to single values of height corresponding to the extracted multiple height parameters of the multiple scatterers, or to the values of height comprised in their neigbourhoods, with a step of averaging of the various bands, or correlation times, or temporal coherences, or distributions of radio reflectivity in the height/displacement velocity joint domain ristrected to the various height set in said neighbourhoods, said distributions being preferably equalized to identical integral value, or temporal decorrelation effective trands, calculated for each neighbourhood.

The post-processing step 25 can also comprise an extraction step from the differential tomographic image of displacement velocity measurements of the various components of the volumetric scatterers, to obtain a profile of displacement velocity vs. height 32, called tomographical displacement velocity profile in height.

Such extraction of displacement velocity measurements of various components of volumetric scatterers can be carried out by acting on the reflectivity distribution in the height/displacement velocity joint domain, possibly by fixing and/or windowing, by extraction of the position of the dominant monodimensional peak value of said reflectivity distribution restricted to each of the various positions set in the domain of interest for which the measurement displacement velocity is extracted, or computation of the centroid of said reflectivity restricted distribution, possibily even after rextriction in a neighborhood of said peak. Said tomographical displacement velocity profile in height can be displayed in various scales and equivalent graphic forms.

Finally, the post-processing step 25 can comprise, in case of use of multiple looks to each of which, or to subsets of which, the two-dimensional separation has been applied separately, a step of averaging of various results from this obtained or various results obtained starting from these by separately applying to each other post-processing steps, in order to minimize random speckle scintillation effects.

The application of the method according to the invention can attain, for an actual range-azimuth resolution cell, multiple parameters of height and/or displacement velocity and/or radio reflectivity, and/or of the number of multiple layover scatterers 27, said multiple parameters of height and/or radio reflectivity being robust, or as particular case of a single height parameter of a robust DEM 29, and/or of a robust tomographic reflectivity profile in height 28, and/or of a tomographical temporal coherence profile in height of volumetric scatterers or of temporal coherence multiple parameters of multiple layover scatterers 31 , and/or of a tomographical displacement velocity profile in height 32. The reflectivity, height, displacement velocity, and information of temporal coherence can be expressed also in the scales and forms equivalent already described.

In case of multidimensional separation 23b in a joint domain comprising in addition to the height parameters and displacement velocity also other kinematic parameters and/or parameters defining temporal evolution, the postprocessing step 25b can comprise furthermore, the steps described below.

The post-processing step 25b can comprise an extraction step of a marginal radio reflectivity distribution in the only domain of height/displacement velocity from the generalized differential tomographic image, to obtain a differential tomographic image 28b. This differential tomographic image can then be computed in one or more post-processing steps 25. This marginal radio reflectivity distribution extraction can be attained by integration of the generalized differential tomographic image in the domain of said kinematic parameters and/or parameters defining temporal evolution, preferably after operations of fixing or windowing, or by extraction of the monodimensional or multidimensional peak of the generalized differential tomographic image restricted to each different couples of values of height and displacement velocity set in the domain of interest. In said extraction step the values of reflectivity can be cumulated after being remapped in a domain where the displacement velocity is not referred to the reference pass, for example it is referred to the given temporal delay elapsed starting from its reference pass or it is the mean displacement velocity, said displacement velocity being deducted from known kinematical functions.

The post-processing step 25b can comprise furthermore, an extraction step of the multiple parameters of height and/or displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution and/or radio reflectivity 27b, preferably made by extraction of the position and/or height of dominant peaks in the generalized differential tomographic image. The multiple parameters of height and/or displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution can be extracted also by computation of the centroids of the radio reflectivity distribution in the multidimensional domain restricted to neighbourhoods of said peaks, and the multiple reflectivity parameters can be extracted also by integration of the radio reflectivity distribution in the multidimensional domain restricted to said neighbourhoods. The post-processing step 25b can comprise also an extraction step of the number of scatterers that can be carried out by a thresholding test of the reflectivity parameters extracted from the generalized differential tomographic image 24b. The post-processing step 25b can also comprise a step of extraction from the generalized differential tomographic image of measurements of displacement velocity and/or of other kinematic parameters and/or parameters defining temporal evolution of the various components of volumetric scatterers, to obtain one or more profiles of kinematic parameters and/or parameters defining temporal evolution vs. the height 32b, called profiles of kinematic parameters and/or parameters of tomographic temporal evolution in height, in particular a profile of line-of-sight acceleration and/or of line-of-sight step displacement and/or of temporal delay between the step displacement or phase variation and its tomographical reference pass in height. This extraction of measurements of kinematic parameters and/or parameters defining temporal evolution of the various components of volumetric scatterers can be carried out through the reflectivity distribution in the multidimensional domain, possibly by fixing and/or windowing, by extraction of the position of the dominant monodimensional or multidimensional peak value of said reflectivity distribution restricted to each of the various positions set in the domain of interest for which said measurement/s is/are extracted, or computation of the centroid of said reflectivity restricted distribution, possibily even after rextriction in a neighborhood of said peak. Said profiles can be displayed in various scales and equivalent graphic forms.

In said extracting step, the kinematic parameters, such as displacement velocity and/or line-of-sight acceleration, can be expressed not referred to the reference pass, using known kinematical functions.

Finally, the post-processing step 25b can comprise, in case of a multidimensional separation applied separately to multiple looks or to subsets thereof, a step of averaging various results from it obtained or various results obtained starting from them by applying to each separately other postprocessing steps.

The application of the method according to the invention can thus lead also to attain, for an actual range-azimuth resolution cell, multiple kinematic parameters (in addition to the parameters of displacement velocity) and/or parameters defining temporal evolution of multiple scatterers 27b, and/or of one or more profiles of kinematic parameters (in addition to the displacement velocity parameter) and/or parameters of tomographic temporal evolution in height 32b.

In case of two-dimensional 23 or multidimensional 23b separation obtained with a method of analysis based on models, the attainment of said parameters and/or number of scatterers 27 or 27b, or robust parameter 29 and/or robust tomographic reflectivity profile in height 28, and/or tomographical temporal coherence profile in height and/or parameters defining temporal coherence 31, and/or one or more profiles of displacement velocity, and/or other kinematic parameters and/or parameters of tomographic temporal evolution in height 32 or 32b, is achieved also directly by the chosen model. In particular, the robust tomographic reflectivity profile in height 28 can be obtained, in case of model suitable for surface or volumetric distributed scatterers, preferably multiple, with distribution of displacement velocity for the radio echo component at a certain not null extension height, or radio echo component with temporal decorrelation, by evaluation of various heights set in the domain of interest of the analytic expression of the integral along the domain of the displacement velocity, preferably with windowing, of the radio reflectivity distribution in the height/displacement velocity joint domain corresponding to the model, for parameters calculated in the fitting. In case of use of multiple looks separately, various results of identification or various multiple or single parameters and/or profiles are averaged out.

An example of robust radio tomographical reflectivity profile in height 93 is shown in the diagram 90 of figure 7, which shows the radio reflectivity 92 responsive to height 91. The graphical shows the radio reflectivity normalized at a maximum value vs. height in a predetermined domain, expressed in Rayleigh resolution units, or normalized, referred to the height of reference used for the deramping step. The resolution cell in range-azimuth for which is obtained this robust radio tomographical reflectivity profile in height is obtained contains a distributed surface scatterer, having zero height and zero average displacement velocity, consisting of some equivalent scatterers with different displacement velocity and then affected by temporal decorrelation owing to change of the inner position of the scatterers. This distributed scatterer has the extension of the tomographical reflectivity profile in height equal to 0.3, in height normalized units, with normalized line-of-sight displacement velocity of the equivalent scatterers that make them up proportionally to their height, in the range from -0.5 to 0.5, and overall ratio signal-noise of 15 dB.

The robust radio tomographical reflectivity profile in height is obtained in conditions of calibration comprising deramping with two-dimensional separation by a resonating two-dimensional filtering with adaptive method, with diagonal loading of the correlation matrix calculated with fixed loading factor equal to 10, and integration of the differential tomographic image along the displacement velocity domain with windowing on a range from -2 to 2, in normalized line-of-sight displacement velocity units, computing multi-pass multibaseline data with one receiving channel for each pass with the configuration of acquisition of figure 3a and sixteen looks, simulated by a computer.

Figure 7 shows also, by comparison, the radio tomographical reflectivity profile in height 94 obtained with classic Fourier monodimensional processing. In the robust radio tomographical reflectivity profile in height 93 the dominant peak value 95 is evident, with a position corresponding with good approximation to the height actual value of the distributed scatterer affected by temporal decorrelation, or producing a single height parameter of a robust DEM. The dominant peak value 96 in the radio tomographical reflectivity profile in height obtained with classic processing provides instead a value of height with significant loss of precision. Furthermore, the latter profile is affected by a significant defocalization caused by temporal decorrelation, resulting in a maximum ambiguity level 97 that is abnormally high, larger than that which is already intrinsic in the use of not uniformly spaced baselines and classic processing, whereas the maximum ambiguities level in the robust profile is satisfactorily low.

The foregoing description of a specific embodiment will so fully reveal the invention according to the conceptual point of view, so that others, by applying current knowledge, will be able to modify and/or adapt for various applications such an embodiment without further research and without parting from the invention, and it is therefore to be understood that such adaptations and modifications will have to be considered as equivalent to the specific embodiment. The means and the materials to realise the different functions described herein could have a different nature without, for this reason, departing from the field of the invention. It is to be understood that the phraseology or terminology employed herein is for the purpose of description and not of limitation.

Claims

1. A method for processing imaging radar data, said radar data being relative to distributed and/or multiple layover scatterers contained in one or more preselected radar resolution cells, said radar data being acquired by a radar system with a multi-pass multibaseline acquisition so that input radar data are obtained of multi-pass multibaseline type, said input radar data consists of at least one set of complex valued pixels, said or each set of pixels corresponding substantially to a respective resolution cell in a plurality of complex focalized radar images, said radar system having at least one receiving channel, and each radar image being formed through said or each receiving channel of said radar system for each pass, wherein said processing comprises a multidimensional separation step, in particular two-dimensional, of focalized multiple radio echo components, in a joint hybrid output domain of parameters, wherein said parameters comprise at least one spatial parameter and a kinematic parameter and/or a parameter defining a temporal evolution, wherein said joint output domain comprises the height/displacement velocity joint domain, said focalized multiple radio echo components originating from said distributed and/or multiple layover scatterers in said or each cell, said separation step being carried out by an at least two-dimensional processing step of said input radar data, or of radar data obtained from said input radar data, in a processing joint domain comprising the baseline/acquisition time joint domain, said separation step using functions, which are responsive to said parameters, describing components of said input radar data, or describing data obtained from said input radar data, in said processing joint domain, and/or using functions, which are responsive to said parameters, or describing correlations or correlation components of said input radar data, or describing data obtained from said input radar data, in said processing joint domain, said separation step comprising an inference of data, in said joint hybrid output domain and in particular in the height/displacement velocity joint domain, on a distribution with respect to said parameters, in particular height and displacement velocity, of said distributed and/or multiple scatterers and/or of components of said scatterers and/or of said components of focalized radio echo.

2. A method for processing imaging radar data, according to claim 1, wherein said height of said hybrid joint domain of parameters, which comprise at least one spatial parameter and a kinematic parameter and/or a parameter defining a temporal evolution, is selected from the group comprised of:

- normal height, or height measured in a direction orthogonal to a nominal line-of-sight, or even to an actual line-of-sight; - vertical height.

3. A method for processing imaging radar data, according to claim 1, wherein said displacement velocity of said hybrid joint domain is selected from the group comprised of:

- line-of-sight displacement velocity, or in a direction of a nominal or actual line-of-sight;

- vertical displacement velocity;

- displacement velocity in horizontal range.

4. A method for processing imaging radar data, according to claim 1, wherein said kinematic parameters of said hybrid joint domain are relative to displacements according to directions selected from the group comprised of:

- line-of-sight;

- vertical;

- horizontal.

5. A method for processing imaging radar data, according to claim 1, wherein a phase is provided of definition of a nominal value of a complex response, of the multi-pass multibaseline radar acquisition system, to the radio echo coming from a single point-like scatterer at a specific height and having a specific line-of-sight displacement velocity, in absence of noise and temporal decorrelation.

6. A method for processing imaging radar data, according to claim 5, wherein said complex response consists of complexjiominal values of corresponding pixels in said focalized images formed by various channels in various passes, said pixel values originating from a single radio echo component, vs. height and displacement velocity of said single point-like scatterer in a preselected two-dimensional domain.

7. A method for processing imaging radar data, according to claim 6, wherein each of said complex nominal values can be defined in amplitude and phase, and in particular said amplitude being unitary, said phase being obtained by evaluating changes of radio echo phase shift in its path from said point-like scatterer towards a corresponding receiving channel with respect to a radio echo phase shift in its path between the same scatterer and a predetermined reference receiving channel of a predetermined reference pass, each of said phase shifts being obtained by evaluating length of said path accounting for its variation with respect to said reference pass and for wavelength.

8. A method for processing imaging radar data, according to claim 7, wherein the length of said path is obtained by evaluating a relative geometry between channel and scatterer, and said variation with respect to said reference pass is determined by the line-of-sight displacement velocity and by the temporal delay after said reference pass.

9. A method for processing imaging radar data, according to claim 8, wherein said relative geometry is determined by the geometric configuration and by the location of said multiple baseline with respect to said range-azimuth resolution cell and by the height of the scatterer.

10. A method for processing imaging radar data, according to claim 5, where said complex response is computed including also other kinematic parameters and/or parameters defining temporal evolution.

11. A method for processing imaging radar data, according to claim 10, wherein said parameters can be line-of-sight acceleration and/or the line- of-sight acceleration derivative and/or line-of-sight step displacement or step phase variation and/or the temporal delay between the step displacement or phase variation and the reference pass.

12. A method for processing imaging radar data, according to claim 5, wherein said values of the complex response are structured in a "steering vector".

13. A method for processing imaging radar data, according to claim 5, where the multi-pass multibaseline acquisition forms a composed spatial- temporal interferometer.

14. A method for processing imaging radar data, according to claim 13, wherein in conditions equivalent to the far field and linear geometry of the composed spatial-temporal interferometer, said complex response consists of complex data in the baseline/acquisition time domain corresponding nominally to a specific spatial frequency, and to a specific linear distribution of an interferometric phase temporal component responsive to the temporal delay between said acquisitions, corresponding to a "temporal frequency", forming a specific two- dimensional spatial-temporal harmonic.

15. A method for processing imaging radar data, according to claim 13, where in conditions equivalent to the far field and linear geometry of the composed spatial-temporal interferometer, said complex response consists of complex data in the baseline/acquisition time domain corresponding nominally to a specific spatial frequency, and to a specific non-linear distribution of an interferometric phase temporal component responsive to the temporal delay between said acquisitions.

16. A method for processing imaging radar data, according to claim 13, where in conditions equivalent to the far field and linear geometry of the composed spatial-temporal interferometer, said complex response corresponds to a specific spatial frequency, and to a specific temporal frequency and/or chirp rate and/or rate of change of the chirp rate and/or step phase variation and/or temporal delay between the step displacement or phase variation and the reference pass.

17. A method for processing imaging radar data, according to claim 6, wherein said step of definition the complex response includes computing an effect of refraction of the radio propagation in volumetric scatterers, so-called dense, such as glaciers.

18. A method for processing imaging radar data, according to claim 1, wherein said two-dimensional separation step in the joint output domain of height/displacement velocity is obtained with a two-dimensional processing technique in the baseline/acquisition time processing joint domain, chosen among, or obtained from combinations of:

- two-dimensional spacial-temporal spectral analysis of said complex data in the baseline/acquisition time processing joint domain, even sparse data;

- resonant or passband two-dimensional filtering of said data in said processing joint domain;

- forming a "hybrid" two-dimensional synthetic beam in the height/displacement velocity joint output domain;

- signal processing of two-dimensional array sensors, so-called 2D "array processing", for sensing arrival directions of waves propagating in a three-dimensional space, so-called 2D direction of arrival estimation (DOA), and in particular for sensing also their amplitude, where one of the two spatial directions is formally replaced by said displacement velocity parameter;

- reconstruction of images by means of variational methods.

19. A method for processing imaging radar data, according to claim 18, wherein said multidimensional separation step in the joint output domain of height/displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution, is obtained with a two- dimensional processing technique in the baseline/acquisition time processing joint domain, chosen among, or obtained from combinations of:

- forming a "hybrid" multidimensional synthetic beam in the joint output domain of height/displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution;

- signal processing of two-dimensional array sensors or array processing, for sensing multiple parameters of waves.

20. A method for processing imaging radar data, according to claim 18, wherein said technique chosen for the two-dimensional separation step is obtained by a method for two-dimensional analysis with high resolution and/or ambiguity suppression, selected from the group comprised of or obtained from a combination of:

- adaptive methods; - apodization methods;

- model-based methods or parametric methods;

- maximum entropy methods.

21. A method for processing imaging radar data, according to claim 18, wherein said parametric methods are based on specific or approximate models of radio reflectivity distribution in the height/displacement velocity joint domain, or of spatial-temporal spectra, for surface and/or volumetric and/or multiple scatterers, with distribution of displacement velocity for the radio echo component at a certain height of null extension, and/or of a not null extension, fixed (decoupled) or variable (coupled) with the height of the various radio echo components, in particular, said specific or approximate models modelling radio echo components without and/or with temporal decorrelation.

22. A method for processing imaging radar data, according to claim 19, wherein said parametric methods are of a class, or are obtained from combinations of classes, based on a type selected among: - subspace decomposition of a correlation matrix of said data;

- iterating alternated estimation and cancellations;

- least squares amplitude fitting;

- autoregressive models or autoregressive moving average models;

- models identification correlation or covariance matrix fitting.

23. A method for processing imaging radar data, according to claim 1, wherein for said two-dimensional separation step of a two-dimensional analysis based on Fourier transform is used.

24. A method for processing imaging radar data, according to claim 18, wherein for said multidimensional separation step of a technique of two- dimensional analysis with high resolution and/or ambiguity suppression is used with an output in said joint hybrid output domain, selected from the group comprised of or obtained from a combination of: - adaptive methods;

- model-based methods or parametric methods;

- methods of beamforming followed by multidimensional deconvolution methods.

25. A method for processing imaging radar data, according to claim 19, wherein said methods are based on the knowledge of the complex response, or of the steering vector.

26. A method for processing imaging radar data, according to claim 1 , wherein said two-dimensional or multidimensional separation includes the extraction of a number of multiple layover scatterers, based on least squares amplitude fitting.

27. A method for processing imaging radar data, according to claim 1 , wherein said method for processing imaging radar data comprises a preliminary calibration step of said multi-pass multibaseline radar data.

28. A method for processing imaging radar data, according to claim 27, wherein said calibration step comprises at least one phase selected from the group comprised of:

- calibrating amplitude and/or phase of various receiving channels; - compensating the effects due to a possible non-linearity of the geometry of said composed spatial-temporal interferometer and at the curvature of the electromagnetic radio echo wave fronts, so-called deramping; - compensating parasitic phase shifts due to variation of the radio propagation velocity in the atmosphere and/or in the ionosphere during the multi-pass acquisition, so-called atmospheric compensation.

29. A method for processing imaging radar data, according to claim 1, wherein said method for processing imaging radar data comprises a preprocessing step of said multi-pass multibaseline radar data for bringing said data in conditions adapted to the successive two- dimensional or multidimensional separation, said preprocessing step comprising, in particular, at least one phase selected from the group comprised of:

- minimizing parasitic spatial decorrelation effects due to variation of geometric conditions for optimal coregistration of the images vs. an examined height;

- minimizing random scintillation effects of the radio echo, which are due in case of distributed scatterers or distributed multiple scatterers, at an interaction of the radio waves with a surface or volume microstructure of said scatterers, or the inner position of equivalent scatterers, so-called complex "speckle" phenomenon or "fading", that adds to additive noise so-called "thermal" noise; - minimizing ambiguities effects, in a successive two-dimensional or multidimensional separation, due to two-dimensional sparse non uniform sampling of the baseline-acquisition time domain carried out in said multi-pass multibaseline radar data acquisition step;

- stabilizing the correlation matrix of the multi-pass multibaseline complex data by means of so-called diagonal loading, in case of two- dimensional or multidimensional separation with adaptive method, based on said correlation matrix, and/or in case of a preprocessing step comprising an extraction step of the number of multiple scatterers by means of eigenvalue-based methods applied to said matrix; - extracting the number of multiple scatterers, for example in case of two-dimensional or multidimensional separation of multiple scatterers with a class method based on subspace decomposition, or more in general calculating a so-called data model order.

30. A method for processing imaging radar data, according to claim 29, wherein said step of minimizing parasitic spatial decorrelation effects is obtained carrying out a compensation of the so-called migration in range by means of a coregistration in variable range with the height of interest, said preprocessing step for minimizing said effects of spatial decorrelation with that of separation being a method for three- dimensional processing in the processing joint baseline-acquisition time - range domain.

31. A method for processing imaging radar data, according to claim 29, wherein said step of minimizing scintillations effects is carried out by using so-called multiple looks for each corresponding pixels in the various focalized complex images, or for groups of pixels adjacent to and comprising a pixel of interest or reduced resolution multiple focalized versions of the same.

32. A method for processing imaging radar data, according to claim 31, wherein said step of minimizing scintillations effects is obtained by calculating the correlation matrix of the multi-pass multibaseline complex data, through a coherent average on said multiple looks, on which a successive two-dimensional or multidimensional separation is based.

33. A method for processing imaging radar data, according to claim 1 , wherein said complex multi-pass multibaseline radar data are structured in a data vector with the same structure of said steering vector, or in more data vectors with the same structure of said steering vector in case of multiple looks.

34. A method for processing imaging radar data, according to claim 29, wherein said step of minimizing ambiguities effects is obtained by two- dimensional interpolation of the data with a priori information on the so- called support in the domain of height/displacement velocity and/or other kinematic parameters or parameters defining the temporal evolution, where the radio echo components are expected, or on their average statistic reflectivity distribution in the height-displacement velocity domain, or extension of methods of monodimensional interpolation with a priori information, to obtain interpolated multi-pass multibaseline data.

35. A method for processing imaging radar data, according to claim 34, wherein said two-dimensional interpolation can be obtained by a linear transformation of the data, defined for minimizing the square modulus of the deviation between the complex response relative to the interpolated baselines and/or acquisition times and the response obtained from said linear transformation applied to the complex response relative to the available baselines and acquisition time, in particular, said square modulus of the deviation being cumulated, or weighed and cumulated, for a grid of values of height and displacement velocity in said support.

36. A method for processing imaging radar data, according to claim 29, wherein said step of minimizing ambiguities effects is carried out by two- dimensiqnal windowing methods, in particular, said two-dimensional windowing methods being applied as desired at least to one of the following types of said data and matrix:

- complex multi-pass multibaseline radar data;

- interpolated data;

37. A method for processing imaging radar data, according to claim 29, wherein said preprocessing step comprises a phase of computation of the correlation matrix of the complex multi-pass multibaseline radar data through a coherent average on the baselines of identical length and/or on the temporal ranges between acquisitions of identical duration, or on partitions of the data, in particular, after deramping and two-dimensional interpolation, and advantageously two-dimensional windowing, in case of two-dimensional or multidimensional separation based on said correlation matrix and/or in case of preprocessing comprising an extraction step of the number of multiple scatterers by said eigenvalue based methods, and use of a so-called single look for keeping a full capacity of resolution in range-azimuth, in particular, said diagonal loading for stabilizing the correlation matrix being fixed or adaptive.

38. A method for processing imaging radar data, according to claim 37, wherein said step of extracting the number of multiple scatterers is obtained by eigenvalue-based methods applied to the correlation matrix, in particular a stabilized correlation matrix.

39. A method for processing imaging radar data, according to claim 5, wherein said complex response is calculated for ideal geometric configuration and/or acquisition times which the data can be brought to, in case a deramping step and/or interpolation is carried out.

40. A method for processing imaging radar data, according to claim 37, where in case of a separation based on a calculated correlation matrix after interpolation using a single look, the complex response that can be used at the separation is obtained as a partition of the complex response corresponding to said configuration and/or acquisition times of the interpolated data.

41. A method for processing imaging radar data, according to claim 27, wherein said two-dimensional or multidimensional separation is carried out on said calibrated and/or preelaborated multi-pass multibaseline radar data, in case a calibrating and/or preprocessing step has been carried out previously.

42. A method for processing imaging radar data, according to claim 27, wherein for minimizing parasitic spatial decorrelation effects by migrations in range, the separation step is obtained by a three- dimensional processing of said data, in a processing joint domain comprising in addition to the baselines and to the acquisition time also the range, wherein said data can be calibrated and/or preelaborated with preliminary inverse Fourier transform in range, and refocalization in range jointly to said separation by three-dimensional processing techniques, such that, in case of two-dimensional separation the output of the three- dimensional processing techniques is in the joint domain of height/displacement velocity jointly to the domain in range.

43. A method for processing imaging radar data, according to claim 1 , wherein said method for processing imaging radar data comprises at least one phase selected from the group comprised of:

- preparing an image of height and displacement velocity joint distribution and/or other kinematic parameters and/or parameters defining temporal evolution for said cell;

- preparing an image of reconstructed radio reflectivity distribution in a joint domain of height and displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution for said cell (hereinafter called "generalized differential tomographic image"), said reflectivity being also exprimible in an equivalent scale such as amplitude, intensity, radar cross section (RCS), normalized RCS.

44. A method for processing imaging radar data, according to claim 1, comprising a further post-processing step in the joint domain of height/displacement velocity, or of spacial-temporal frequencies, for extracting additional data from said differential tomographic image, said post-processing step comprising in particular, at least one phase selected from the group comprised of:

- extracting multiple parameters of height and/or displacement velocity and/or radio reflectivity of multiple layover scatterers;

- extracting the number of multiple layover scatterers; - extracting a radio reflectivity distribution so-called marginal in the only height domain from the differential tomographic image, for obtaining a robust tomographic reflectivity profile in height with respect to a temporal signal decorrelation, or for virtually reducing phenomena of temporal coherence loss; - extracting multiple parameters of height and/or radio reflectivity of multiple layover scatterers from said robust tomographic reflectivity profile in height, or from the differential tomographic image, for obtaining said parameters that are robust vs. a temporal signal decorrelation, or for virtually reducing phenomena of temporal coherence loss; - extraction of a single height parameter of a single scatterer, either a point-like or a surface or a volumetric scatterer, from the differential tomographic image or from said robust tomographic reflectivity profile in height, to obtain a robust DEM having a temporal signal decorrelation, or for virtually reducing phenomena of temporal coherence loss; - extracting temporal coherence measurements of the various components of volumetric scatterers or of multiple layover scatterers, or preparing a profile of temporal coherence vs. height, (hereinafter called "tomographical temporal coherence profile in height"), and/or obtaining the temporal coherence multiple parameters of the extracted multiple layover scatterers;

- extracting displacement velocity measurements of various components of volumetric scatterers, or preparing a profile of displacement velocity vs. height, (hereinafter called "tomographical displacement velocity profile in height"); - minimizing random speckle scintillation effects by using multiple looks, to each of which, or to subsets of which, the two-dimensional separation has been applied separately, and by an incoherent average of the various differential tomographic images obtained from them, or by an average of various multiple parameters of height and/or displacement velocity and/or radio reflectivity, or various parameters of only height, and/or various robust tomographic reflectivity profiles in height, and/or various height tomographic coherence profiles and/or various temporal coherence multiple parameters and/or various tomographic displacement velocity profiles in height extracted therefrom.

45. A method for processing imaging radar data, according to claim 44, wherein said step of extracting multiple parameters of height and/or displacement velocity and/or radio reflectivity is obtained from dominant peaks of said differential tomographic image and/or from their neigbourhoods.

46. A method for processing imaging radar data, according to claim 44, wherein said step of extracting the number of multiple scatterers is obtained by a threshold test on said multiple reflectivity parameters.

47. A method for processing imaging radar data, according to claim 27, wherein said step of extracting a marginal radio reflectivity distribution is obtained by integration along the displacement velocity domain and/or by extraction of a monodimensional maximum along said domain of said differential tomographic reconstructed image in the height/displacement velocity joint domain, restricted to each of the various actual heights.

48. A method for processing imaging radar data, according to claim 44, wherein said step of extracting multiple parameters of height and/or radio reflectivity from said robust tomographic reflectivity profile in height is obtained from dominant peaks of said profile and/or from their neigbourhoods.

49. A method for processing imaging radar data, according to claim 44, wherein said step of extraction of a single height parameter from said robust tomographic reflectivity profile in height is obtained from the highest peak value of said profile and/or from a neigbourhood thereof.

50. A method for processing imaging radar data, according to claim 44, wherein said step of extracting temporal coherence measurements of the various scatterer components is obtained by extraction of a displacement velocity band and/or a band of temporal frequencies and/or so-called correlation time and/or coherence measurements and/or by extraction of the course of a temporal decorrelation through a Fourier reverse transform from said differential tomographic reconstructed image in the height/displacement velocity joint domain restricted to each of the various actual heights, and/or to said multiple height parameters, or to the heights comprised in their neighbourhoods with following average operations for each neighbourhood.

51. A method for processing imaging radar data, according to claim 27, wherein said step of extracting displacement velocity measurements of different scatterer components is obtained from the highest monodimensional peak value and/or from a monodimensional centroid of said differential tomographic reconstructed image in the height/displacement velocity joint domain restricted to each of the various actual heights.

52. A method for processing imaging radar data, according to claim 44, wherein said post-processing step, in case of multidimensional separation in a joint output domain of height/displacement velocity and/or of other kinematic parameters and/or defining the temporal evolution, can comprise a post-processing step in the joint domain of height/displacement velocity and/or of other said parameters, for extracting additional data from said generalized differential tomographic image.

53. A method for processing imaging radar data, according to claim 52, wherein said post-processing step in the joint domain of height/displacement velocity and/or of other kinematic parameters and/or parameters defining temporal evolution comprises at least one phase selected from the group comprised of: - extracting a marginal radio reflectivity distribution in the only domain of height/displacement velocity from the generalized differential tomographic image, for obtaining a differential tomographic image, wherein this differential tomographic image can then be computed in one or more post-processing steps in the joint domain of height/displacement velocity. - extracting multiple parameters of height and/or displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution and/or radio reflectivity of multiple layover scatterers.

- extracting the number of multiple layover scatterers.

54. A method for processing imaging radar data, according to claim 44, wherein said extraction of marginal radio reflectivity distribution in the only domain of height/displacement velocity is obtained by integration of the generalized differential tomographic image in the domain of said kinematic parameters and/or parameters defining temporal evolution, and/or by extraction of the monodimensional or multidimensional peak of the generalized differential tomographic image restricted to each different couples of actual values of height and displacement velocity.

55. A method for processing imaging radar data, according to claim 44, wherein said step of extracting multiple parameters of height and/or displacement velocity and/or other kinematic parameters and/or parameters defining temporal evolution and/or radio reflectivity is obtained from dominant peaks of said generalized differential tomographic image and/or from their neigbourhoods.

56. A method for processing imaging radar data, according to claim 44, wherein said step of extracting the number of multiple scatterers is obtained by a threshold test on said multiple reflectivity parameters extracted from the generalized differential tomographic image.

57. A method for processing imaging radar data, according to claim 44, wherein said step of extracting displacement velocity measurements and/or of other kinematic parameters and/or parameters defining temporal evolution of the various scatterer components is obtained from the highest monodimensional or multidimensional peak value and/or from a centroid of said generalized differential tomographic image restricted to each of the various actual heights.

58. A method for processing imaging radar data, according to claim 21 , comprises, in case of two-dimensional or multidimensional separation with parametric method or method based on models, a step of obtaining from the model at least one said parameter and/or profile for the actual range-azimuth resolution cell selected from the group comprised of:

- number of multiple layover scatterers;

- a single height parameter to obtain a robust DEM; - multiple parameters of robust height and/or radio reflectivity;

- a robust tomographic reflectivity profile in height;

- a tomographical temporal coherence profile in height and/or temporal coherence multiple parameters of the multiple layover scatterers;

- a tomographical displacement velocity profile in height and/or one or more profiles of other kinematic parameters and/or parameters defining temporal evolution.

59. A method for processing imaging radar data, according to claim 31 , where in case of use of multiple looks, to each of which, or to subsets of which, the two-dimensional or multidimensional separation with parametric method has been applied separately, various results of identification or various of said multiple or single parameters and/or profiles are averaged out.

60. A method for processing imaging radar data, according to claim 1, mounted on multi-pass multibaseline data of a synthetic aperture radar (SAR), wherein said multi-pass multibaseline data are acquired by a radar imaging system of a type selected from the group comprised of or obtained from a combination of:

- radar with a single receiving channel forming a single complex focalized image for each pass; - multichannel radar of multiantenna type, forming at least two focalized complex images for each pass;

61. A method for processing imaging radar data, according to claim 60, wherein said commutation phase of the transmitter uses a technique, so- called "ping-pong" in case of two antennas, capable of forming for each pass more images of the number of receiving channels so-called real, owing to the synthesis of receiving channels so-called bistatic equivalent additional channels.

62. A method for processing imaging radar data, according to claim 60, wherein said multichannel radar of multiantenna type is of a type selected from the group comprised of or obtained from combination of:

- multiantenna distributed radar, having antennas arranged separately on vehicles or different platforms, which travel along substantially parallel tracks in case of said vehicles or mobile platforms.

63. A method for processing imaging radar data, according to claim 1, wherein said radar imaging system is transported by one or more means selected from the group comprised of: - an aircraft or other avionic platform;

- a satellite or other spatial platform;

- a flight of aircrafts and/or other multiple avionic platforms;

- a flight of satellites so-called satellite "cluster" and/or other spatial platforms; - a ground based motorized rail vehicle;

- a plurality of ground based motorized rail vehicles.

64. A method for processing imaging radar data, according to claim 1, wherein said multi-pass multibaseline data are acquired by a radar imaging system of passive type with a single receiving channel or with a multichannel of multiantenna type, in particular, using radar pulses or in general radio signals transmitted by an external source, said passive radar imaging system being, in particular, based fixed on ground or based on a cluster of satellites, the source of the pulses or signals used by said passive radar imaging system being in movement in case of SAR static passive radar system.