• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    The invention relates to a method (10) for processing a SAR-type radar image, comprising the steps of: - Charging (100) an initial radar image (ISAR) of SAR type, - Charging (110) an optical image initial (lopt) said initial images having been optionally pre-processed so that they have the same axis of acquisition, -Charger (120) additional graphic data (Data) associated with the initial optical image and relating to elements geographical features present in said initial optical image, -Treating (130) at least one of the two initial images so as to make them capable of being superimposed in an overlap area to obtain a superimposed radar image (ISAR / sup) and an optical image superimposed (Iopt / sup), -Determine (140) additional superimposed graphic data (Datasup) from the additional graphic data (Data) and the superimposed optical image (Iopt / sup), -Locate (150) elements the geographic features present in the superimposed SAR image (ISAR / Sup) at least from said additional superimposed graphical data (Datasup).
    公开号:FR3042283A1
    申请号:FR1502127
    申请日:2015-10-09
    公开日:2017-04-14
    发明作者:Gilles Guerrini；Bernard Joseph；Thierry Sfez
    申请人:Thales SA；
    IPC主号:
    专利说明:

    Method of processing a SAR-type radar image and associated target detection method
    FIELD OF THE INVENTION The invention relates to the general field of airborne or spaceborne synthetic aperture radars. The present invention more particularly relates to a method of processing a SAR-type radar image to locate geographic features present in the image and a method of target detection using the method of processing.
    STATE OF THE ART
    Synthetic Aperture Radar (or SAR) radars are widely used in the remote analysis of geographical areas, for example the detection of targets of these zones, from their location. radar reflectivity. A SAR image allows a vision typically at 30 km and in any weather. The analysis of the SAR image, for example the precise location of certain geographical elements and the detection of targets present in the image, is complicated by the existence of artifacts such as bright and / or salient points coming from example of reflective angles of roofs of buildings. In addition, it is difficult to delineate the contours of a building in SAR imagery since the response of each of its elements can be very different.
    There are now many algorithms and treatments dedicated to the detection and identification of fixed targets on very high resolution SAR radar imagery. The results obtained in open rural areas validate the robustness of these treatments from the moment when the type of target to be searched is known (tank, truck, plane, ...). On the other hand, in urban areas, the rate of false detections remains high because of the presence of artifacts and poses a problem for the operator in charge of the interpretation of the images for the detection of real targets, in particular when the targets searched are close to buildings.
    To improve the analysis of SAR images, work has been done on radar and optical data fusion. Optical imaging does not allow easy detection of targets but gives an image of urban areas to accurately delineate the geographic location of larger elements whose location does not change over time, called "invariant" Such as structures fixed to the ground, for example buildings.
    Thus, the complementary use of SAR and optical images makes it possible to obtain improved knowledge of the environment.
    A first method of fusion of SAR and optical data uses SAR images and optronic images (for example using an optoelectronic pod) decorrelated temporally. This has the advantage of doing post-mission or post-acquisition processing, but there is no spatial geometric transformation of the SAR image with respect to the optical image, which prevents the two images from being superimposed.
    Indeed, these two images are taken in two different planes and with different lines of sight and the type of projection of a 3D object on a plane is not the same in the case of a SAR image and a optical image. Take the example that the observed soil is a horizontal plane. In the case of an optical image, the 3D objects on the horizontal plane of the ground as well as the horizontal plane of the ground are projected on the plane orthogonal to the line of sight of the optical sensor; thus, in the case where the line of sight is orthogonal to the ground plane (in which case the optical sensor is vertical to the observed area), then the optical image obtained is identical to the plane of the ground on which the projections will be superimposed. on the horizontal plane of 3D objects.
    In the case of a SAR image, the 3D objects on the horizontal plane of the ground and the horizontal plane of the ground are projected on the plane containing the line of sight of the radar (where direction of the beam of the radar antenna) and the line orthogonal to the line of sight and belonging to the horizontal plane; thus, in the case where the radar is grazing relative to the observed surface, the radar image obtained is identical to the plane of the ground on which will be superimposed the projections on the horizontal plane of the 3D objects.
    This has the consequence that it is not possible to easily correlate geographic data that is not viewed along the same line of sight. Both SAR and optronic images are acquired and displayed on two different visualizations. This technique has the advantage of being able to process the two images independently but do not provide assistance to the operator to correlate the information of the two images.
    An object of the present invention is to overcome the aforementioned drawbacks by proposing a method for processing the SAR image for locating geographical elements present with the aid of an "augmented" optical image by additional graphic data. This location makes it easier to detect targets in the SAR image.
    DESCRIPTION OF THE INVENTION
    The subject of the present invention is a method for processing a SAR-type radar image, comprising the steps of: Charging an initial radar image of the SAR type of a first zone, Charging an initial optical image of a second zone zone at least partially covering the first zone, said initial images having been optionally pre-processed so that they have the same axis of acquisition, -charging additional graphic data associated with the initial optical image and relating to geographical elements present in said initial optical image, -Treating at least one of the two initial images so as to make them capable of being superimposed in an overlap zone between the first and the second zone, to obtain a superimposed radar image (Isar / sup) and an optical image superimposed (lopt / sup). -Determine additional graphic data superimposed (Datasup) from the additional graphic data (Data) and the superimposed optical image (lopt / sup), -Locate geographical elements present in the SAR image superimposed at least on said data additional graphics superimposed.
    Advantageously, the processing step is performed on the optical image only, the initial radar image then being equal to the superimposed radar image.
    Advantageously, the initial radar image is ortho-rectified.
    Advantageously, the initial optical image is an ortho-rectified satellite image.
    Preferentially, the geographical elements comprise frontiers and / or railway networks and / or hydrographic networks and / or elevation contours and / or coastlines, and / or approximate contours of structures fixed to the ground and / or airports and / or inhabited places and / or geographical name indexes.
    Preferably, the additional graphic data are displayed in the optical images in the form of lines and / or lines and / or outlines and / or geometric shapes and / or characters. Advantageously, the additional graphic data are cartographic vector information of Vmap type.
    According to one embodiment, the processing step comprises a step of determining at least one set of registration parameters comprising a homothetic factor, an angle of rotation, and a translation distance.
    Preferably, the step of determining the superimposed additional graphic data comprises a step of applying said set of adjustment parameters to the additional graphic data to obtain the additional superimposed graphic data. Advantageously, the step of determining the resetting parameters is carried out from the Fourrier-Mellin transform.
    According to one embodiment, the location step comprises a step of merging the additional graphic data superposed in the superimposed radar image.
    The present invention also relates to a target detection method comprising a method of treatment according to the invention, wherein the geographical elements comprise structures fixed to the ground, the additional graphic data comprising approximate geometrical shapes of said structures fixed to the ground, and comprising furthermore, a step of determining potential target positions from the superimposed SAR image, and a step of identifying possible false alarms by comparing the position of potential targets identified in the superimposed radar image with geometric shapes approached ground-based structures located at the locating step.
    According to one embodiment: * the potential target is considered as a false alarm when its position is located inside a structure fixed on the ground, * the potential target is considered as a false alarm when its position is located at a distance (d) of a structure fixed to the ground less than a predetermined minimum distance, * the potential target is considered as a real target when its position is located at a distance from a structure fixed to the ground greater than or equal to the minimum distance predetermined. Other features, objects and advantages of the present invention will appear on reading the detailed description which follows and with reference to the appended drawings given by way of non-limiting examples and in which:
    FIG. 1 illustrates the steps of the method 10 for processing a SAR type radar image according to the invention.
    Figure 2 shows an example of a charged Isar radar image, converted to black and white.
    Figure 3 illustrates an initial optical image corresponding to an ortho-rectified satellite image.
    FIG. 4 illustrates the satellite optical image of FIG. 3 enriched with additional graphic data in the form of polygonal geometric shapes.
    FIG. 5 illustrates the processing of an initial image to be recalibrated illustrated in FIG. 5b, to make it superimposable to a reference image illustrated in FIG. 5a. FIG. 5c illustrates the scaled image to be scaled by applying a scaling factor h and the image 5d the rotated image to be rotated so that the orientation is identical to that of the image of reference.
    FIG. 6 illustrates the optical image corresponding to the optical image of FIG. 3 having undergone the transformation to make it superimposable to the SAR image of FIG. 2.
    FIG. 7 illustrates the result of the transformation performed on the additional graphic data to render them superimposable on the reference SAR image, in the form of a binary mask, which corresponds to the buildings present in the optical image of FIG. .
    FIG. 8 illustrates the enhanced superimposed optical image of the additional superimposed graphic data.
    FIG. 9 schematizes an example of an algorithm performing a Fourier Mellin transformation.
    Figure 10 schematizes a target detection method according to the invention comprising a method of treatment according to the invention.
    FIG. 11 illustrates the comparison between the position of potential targets identified in the superimposed radar image with the approximate geometric shapes of ground-based structures located at the location step.
    FIG. 12 illustrates a comparison stage variant of the target detection method according to the invention.
    Figures 13 to 15 illustrate an example of implementation of this detection method for the case in which the SAR radar image is displayed next to the image representing the contours of the structures fixed to the ground.
    Figure 13a illustrates the Isar image in which a bright spot, corresponding to a potential target such as a tank, is detected. FIG. 13b illustrates superimposed data corresponding to the precise position of the building contour generated according to the treatment method according to the invention, for the case in which the distance from the potential target to the building is too small. This is an artifact, and therefore the potential target is classified as a false alarm.
    In the case of FIG. 14, FIG. 14a illustrates the detected target and FIG. 14b illustrates the building in the case where the target is sufficiently far away from a building. The detected target is classified as a real target. In the case of Figure 15, the target detected at 15a is removed from any building (no building on 15b) and is therefore classified as a real target.
    DETAILED DESCRIPTION OF THE INVENTION
    FIG. 1 illustrates the steps of the method 10 for processing a SAR type radar image according to the invention. The method 10 comprises a step 100 of loading an initial radar image Isar of a first zone, typically high resolution, that is to say here whose resolution is better than 5m. We seek to treat the Isar SAR image in order to better identify, locate, delimit in this image geographical elements such as structures fixed on the ground (for example buildings), roads, airports, air bases, railroad tracks, railway and railway stations, inhabited places ...
    Figure 2 shows an example of a charged ISar radar image, converted to black and white. An Isar image from a radar typically has gray levels, not shown. This image is typically geo-referenced, that is, the longitude and latitude coordinates of points of this image are known.
    In the example of Figure 2 the initial SAR image is an image from the radar that has been pretreated to be made ortho-rectified. Orthorectification is a geometric image correction that aims to obtain an image identical to the ground plane on which are superimposed projections on the horizontal plane of 3D objects. For an optical image, this corresponds to a transformed line of sight to be orthogonal to the ground in the case where it is horizontal (image acquired from the vertical, in remote sensing, it is called "nadir"), and for a SAR image the ortho rectification corresponds to a line of sight transformed into a line of sight of the grazing radar. In another example, the SAR image is the image directly from a radar according to the state of the art.
    The method 10 according to the invention also comprises a step 110 of loading an initial optical image lopt of a second zone at least partially covering the first zone.
    We then look at the overlap zone of the two zones, for which we have the two images, SAR and optical. The recovery is for example determined because the two images, SAR and optical, are geo-referenced, from the outset or following a treatment.
    According to a first embodiment, the initial optical image is, for example, an ortho-rectified satellite image as illustrated in FIG.
    According to a second embodiment, the initial optical image is obtained from a camera embedded on an aircraft. The optical image is typically obtained in the visible spectrum or in the infra red spectrum.
    When both SAR and optical images are ortho-rectified, they have the same line of sight, corresponding to the vertical, and no prior processing is necessary.
    When at the time of acquisition of the SAR and optical images do not have the same acquisition axis, pretreatment is performed on these images to generate initial images so that they have the same axis of acquisition. This is the case for example when the optical image comes from an optronic image obtained by a camera on board an aircraft and the SAR radar image comes directly from the radar.
    In a step 120 additional data data data associated with the initial optical image and relating to geographical elements present in the lopt image are loaded, these data making it possible to improve the knowledge of the environment. These graphic data are typically virtual data displayed in the image in the form of lines, lines, contours, geometric shapes, areas of different colors, characters ... which are superimposed on the initial optical image, and specify / represent geographical features such as: ground-based structures such as buildings, roads, railway networks, river systems, coasts, populated places, boundaries, elevation contours, geographical names index, airports.
    These additional graphic data are for example provided (or sold) by public bodies (for example the National Geographical Institute) or private (Google, etc.), in the form of so-called "vector" data associated with a typical satellite optical image. . An example of data of this type are cartographic vector information, for example of Vmap type for Vector Map, also called Vector Smart Map. These data exist in different levels of detail (VmapO to Vmap2) allowing to access information about the environment with great precision (typically a few meters up to 1 meter.) We speak of vector data because a surface consisting of a large number of pixels can be reduced to the knowledge of the positions of a limited number of characteristic points.
    FIG. 4 illustrates the satellite optical image of FIG. 3 enriched with additional graphic data in the form of geometric shapes of polygon type 41, 42, 43, representing buildings.
    When the optical image is obtained from an onboard camera having a first line of sight determined, a variant is to generate the initial optical image by "ortho-rectifying" the image from the onboard camera. The additional graphic data are obtained by superimposing the obtained ortho-rectified initial optical image on a satellite optical image which is itself ortho-rectified for which additional graphic data are available.
    The method 10 according to the invention further comprises a step 130 of treating at least one of the two initial images, Isar and / or lopt, so as to make them capable of being superimposed in a recovery zone between the first and the second zone, to obtain a superimposed radar image Isar / sup and a superimposed optical image lopt / sup- These two superimposed images are therefore images representing the same scene at the same scale and in the same orientation. A pixel located in one of the images corresponds to the same geographical point as the pixel of the same coordinates identified in the other image. This processing step is also called resetting.
    According to a preferred variant, the processing step is performed on the optical image lopt only, the initial radar image Isar then being equal to the superimposed radar image ISar / suP-
    To use additional graphic data associated with satellite images (google / Vmap type), a preferred variant is that the common line of sight is perpendicular to the ground.
    Starting from two images presenting the same axis of acquisition (same line of sight), the treatment consists in operating on at least one image a similarity, ie a homothety (factor of homothety h), a rotation (angle of rotation a) and a translation (translation distance d), as illustrated in FIGS. 5a to 5d with a standard image. Thus, the processing step comprises a step of determining at least one set Prec of readjustment parameters including the homothety factor h (which may be identical for the whole image or local, in case of non-uniform deformation of the optical image), the angle of rotation a and a translation distance d. When the processing is performed on a single image, the other being the reference image, a single set of parameters is determined.
    In the following, the variant in which the SAR ISar image is the reference image, the processing being performed on the optical image lopt only, is described. The processing consists in making homothety / rotation / translation on the optical image, so as to register it on the SAR image. But this preferred variant is not limiting, the processing can be performed on the SAR image only or on both images.
    FIG. 5a illustrates the reference image, FIG. 5b the image to be recalibrated, FIG. 5c the image to be rescaled scaled by applying a homothety factor h and the image 5d the image to be readjusted. rotated so that the orientation is identical to that of the reference image. From the image 5d, the translation distance is determined to obtain an image that can be superimposed on the reference image.
    FIG. 6 illustrates the optical image lopt / suP corresponding to the optical image of FIG. 3 having undergone the transformation to render it superimposable to the SAR image Isar of FIG. 2.
    The processing method 10 according to the invention also comprises a step 140 of determining additional graphic data superimposed Datasup from the superimposed optical image lopt / SuP · The transformation performed on the optical image is performed on the additional graphic data to make them superimposable to the reference SAR image. The result of the transformation is Datasup, as illustrated in FIG. 7 for example in the form of a bit mask, which corresponds to the data buildings present in the optical image of FIG. 4 and having undergone the same transformation as FIG. optical image to be superimposable to the SAR image.
    According to one embodiment, when the processing step 130 comprises the step of determining the set Prec (h, a, d), the step of determining Datasup comprises a step of applying the parameters (h, a, d) to the additional data to obtain Datasup as illustrated in Figure 12. Thus, the registration performed on the optical image makes it possible to determine the adjustment parameters to be applied to the additional graphic data. This saves computing time since to process the data Data is reused parameters determined for the optical image.
    Finally, a last step 150 consists of locating geographical elements present in the superimposed SAR image Isar / suP, equal to Isar in the preferred variant, at least from said additional data data superimposed Datasup.
    According to a first variant, the superimposed SAR image (for example the image of FIG. 2) and the superimposed optical image enriched additional superimposed graphic data (that is to say that the graphic information is integrated in the optical image as shown in Figure 8) are displayed side by side and an operator visually realizes the location in the SAR image by comparing the two images.
    According to a second variant, only the additional superimposed graphic data (for example illustrated in FIG. 7) are displayed side by side with the superimposed SAR image.
    According to a third variant, the location step comprises a step of merging the additional DataSUp superimposed graphic data in the superimposed radar image Isar / sup. Preferably, the merger is performed by integrating the Datasup data into the superimposed radar image. the operator visually locating the corresponding geographical elements directly in the superimposed radar image.
    According to a fourth preferred variant, it is an automatic image processing that uses the merged data to identify false alarms, as described below.
    According to a preferred embodiment, the processing / resetting step is carried out from the Fourier-Mellin or TFM transform, which allows the iso-referencing condition and makes it possible to estimate the similarity type transformation between two images. The use of the TFM is an approach of automatic registration multi-sensor radar / optics, which starting only input image data, without prior knowledge of the image sensor parameters, manages to find a geometric transformation to recalibrate the images. This approach thus makes it possible to perform the registration while the photographic quantities are not all accessible.
    An example of an algorithm performing a Fourier Mellin transformation is illustrated in FIG. 9. The reference image is the SAR image and the optical image is recaled on the SAR image. Obtaining the adjustment parameters is obtained by performing treatments on the two images. The first step of the algorithm is to represent the images in the spectral domain. To make the Fourier Discrete Fourier Transform (DFT) process faster, we use the fast DFT algorithm called Fast Fourier Transform (FFT). The output images of the FFT algorithm are complex: we extract the modules to work with amplitude spectra that are invariant by translation. The application of a "high-pass" filtering makes it possible to eliminate the spectral information on the low frequencies. The abrupt changes in the image, represented by the high frequencies in the spectral domain, are the most significant for performing the registration.
    Once the transformations of homothety and rotation carried out on the image to be recalibrated, one obtains two images 5a and 5d which have same factor of scale and same orientation. It is necessary to determine the displacement 2D (translation) between these two images to finish the process of registration. As before, the translation parameters can be calculated by the phase correlation method and the final registration is performed.
    Thus, the method according to the invention makes it possible to obtain, with a limited calculation time, additional graphic data that are directly superimposable and comparable to the SAR image, by using the registration performed between the optical image to which these graphic data are associated and the SAR image. These additional data allow an improved, more precise knowledge of the geographical elements of the environment in the SAR image (precise delimitation of buildings, location of railways, roads, various infrastructures, etc.). The invention also relates to a target detection method 20 comprising a processing method 10 according to the invention illustrated in FIG. 10. Typically the geographical elements comprise structures fixed to the ground, such as buildings, the additional graphic data comprising approximate geometrical shapes. of these structures fixed on the ground.
    This gives the precise contours of the buildings, for which artefacts in the radar image can be present in the form of highlighting at angles reflecting roofs of buildings, these contours being directly comparable / superimposable to the SAR image.
    The method of detecting targets comprises a step 155 of detecting the presence of potential targets and determining their position by image processing algorithms operated on the SAR image or the superimposed Isar / sup- image. preferred, an operator chooses the type of target (or objects of interest) to be searched for, for example an assault tank or an aircraft on the ground, and the image processing algorithm performs previously, in parallel or as a result processing steps 10, an automatic search for potential targets of the chosen type and their position in the image Isar / sup · The automatic detection of a potential target corresponds to an alarm.
    The target detection method 20 further comprises a step 160 of identifying possible false alarms among alarms generated by the image processing algorithms for the automatic detection of objects of interest (potential targets), by comparing a position of identified potential targets C1, C2, C3, C4 ... in the superimposed Isar / sup radar image with the approximate geometrical shapes of structures fixed to the ground S1, S2, S3 ... located at the location step as shown in Figure 11.
    In fact, SAR imagery can detect small "known" objects (for example trucks or tanks), while additional graphic data associated with optical imaging gives the location of buildings, which allows improve the detection of targets near buildings, by discriminating real targets and artifacts in the SAR image. The advantage of the target detection method 20 is thus to exploit the advantages of SAR imaging to detect small targets and to eliminate false detections from the additional graphic data available associated with the optronic image, in particular removing remaining ambiguities on SAR imagery.
    According to one embodiment, the comparison is carried out on particular areas of the SAR image selected by the operator and zoomed, because presenting an ambiguity.
    According to a variant illustrated in FIG. 12, the comparison step is carried out as follows: the potential target is considered to be a false alarm when its position is located inside a structure fixed to the ground (for example the target C1 of 11), the potential target is considered to be a false alarm when its position is located at a distance (d) of a structure fixed to the ground less than a predetermined minimum distance (Dmin) (for example the target C4 of the 11), the potential target is considered to be a real target when its position is located at a distance (d) from a structure fixed to the ground greater than or equal to the predetermined minimum distance (Dmin) (for example the target C3 of Figure 11).
    Thus if a potential target detected is located at a distance greater than a Dmin of any real estate infrastructure making unambiguous a detection, then this target is considered real; this threshold Dmin, due to the distortions existing on the SAR image and the resulting imperfect local registration, is fixed arbitrarily, according to the average size of the target.
    If a potential target detected is located at a distance close to a real estate infrastructure making ambiguous a detection, then the removal of the ambiguity is for each potential target detected, the comparison with the geographic location for example VMAP superimposed to determine the distance separating it from a real estate infrastructure.
    If a potential target detected is located at the known location of a building then the detection is considered to be erroneous.
    An exemplary implementation of this detection method is illustrated in FIGS. 13 to 15, for the case in which the SAR radar image is displayed next to the image representing the contours of the structures fixed to the ground (for example a binary mask equal to VMAP real estate correspondence).
    Figure 13a illustrates the Isar image in which a bright spot, corresponding to a potential target such as a tank, is detected. By comparing the position of this bright point with the Datasup data illustrated in FIG. 13b corresponding to the precise position of the building contour generated according to the treatment method described above, it is deduced that the distance from the potential target to the building is too small and that it is an artifact, and therefore the potential target is classified as a false alarm.
    In the case of FIG. 14, the target detected at 14a is far enough away from a building displayed at 14b, and the detected target is classified as a real target.
    In the case of Figure 15 the target detected at 15a is away from any building (no building on 15c) and is therefore classified as a real target.
    Thus, thanks to the use of the VMAP data superimposed on the SAR image, certain automatic detections of objects of interest close to or superimposed on buildings can then be considered as false detections and thus be eliminated in this final phase of treatment. .
    权利要求:
    Claims (13)
    [1" id="c-fr-0001]
    A method (10) of processing a SAR-type radar image, comprising the steps of: -loading (100) an initial radar image (Isar) of the SAR type of a first area, -Charger (110) a an initial optical image (lopt) of a second zone at least partially covering the first zone, said initial images having, if necessary, been preprocessed so that they have the same axis of acquisition, -Charger (120) additional graphic data (Data) associated with the initial optical image and relating to geographical elements present in said initial optical image, -Treat (130) at least one of the two initial images so as to make them able to be superimposed in a recovery zone between the first and the second zone, to obtain a superimposed radar image (Isar / sup) and a superimposed optical image (lopt / sup), -Determine (140) additional superimposed graphic data (Datasup) to r additional data data (Data) and superimposed optical image (lopt / sup), -Locate (150) geographical elements present in the superimposed SAR image (Isar / suP) at least from said additional graphic data superimposed (Datasup).
    [2" id="c-fr-0002]
    2. The method of claim 1 wherein the processing step is performed on the optical image (lopt) only, the initial radar image (Isar) then being equal to the superimposed radar image (Isar / suP).
    [3" id="c-fr-0003]
    3. Method according to one of claims 1 or 2 wherein the initial radar image is ortho-rectified.
    [4" id="c-fr-0004]
    4. Method according to one of the preceding claims wherein the initial optical image is an ortho-rectified satellite image.
    [5" id="c-fr-0005]
    5. Method according to one of the preceding claims wherein the geographical elements comprise borders and / or railroad networks and / or hydrographic networks and / or elevation contours and / or ribs, and / or approximate contours of ground-based structures and / or airports and / or inhabited places and / or geographical name indexes.
    [6" id="c-fr-0006]
    6. Method according to one of the preceding claims wherein the additional graphic data are displayed in the optical images in the form of lines and / or lines and / or outlines and / or geometric shapes and / or characters.
    [7" id="c-fr-0007]
    The method of claim 6 wherein the additional graphics data is Vmap-type vector map information.
    [8" id="c-fr-0008]
    8. Method according to one of the preceding claims wherein the processing step comprises a step of determining at least one set (Prec) of registration parameters comprising a scaling factor (h), an angle of rotation ( a) and a translation distance (d).
    [9" id="c-fr-0009]
    The method of claim 8 wherein the step of determining the additional overlay graphic data includes a step of applying said set of registration parameters to the additional graphic data (Data) to obtain the additional overlay graphic data (DataSUp).
    [10" id="c-fr-0010]
    10. Method according to one of claims 8 or 9 wherein the step of determining the resetting parameters is effected from the Fourrier-Mellin transform.
    [11" id="c-fr-0011]
    11. Method according to one of claims 1 to 10 wherein the locating step comprises a step of merging additional superimposed graphic data (Datasup) in the superimposed radar image (IsAR / sup) -
    [12" id="c-fr-0012]
    A target detection method comprising a method of treatment according to one of claims 1 to 11, wherein the geographical elements comprise ground-fixed structures, the additional graphic data including approximate geometric shapes of said ground-mounted structures, and further comprising a step (155) of determining potential target positions from the superimposed SAR image, and a step (160) of identifying possible false alarms by comparing the position of potential targets identified in the superimposed radar image with approximate geometric shapes of ground-based structures located at the locating step.
    [13" id="c-fr-0013]
    The target detection method according to claim 12 wherein: * the potential target is considered a false alarm when its position is located within a ground-based structure, * the potential target is considered a false alarm when its position is located at a distance (d) from a structure fixed to the ground less than a predetermined minimum distance (Dmin), * the potential target is considered a real target when its position is located at a distance (d) a structure fixed to the ground greater than or equal to the predetermined minimum distance (Dmin).
    类似技术:
    公开号 | 公开日 | 专利标题
    US8594375B1|2013-11-26|Advanced cloud cover assessment
    US7116265B2|2006-10-03|Recognition algorithm for the unknown target rejection based on shape statistics obtained from orthogonal distance function
    WO1993021596A1|1993-10-28|Road image sequence analysis device and method for obstacle detection
    EP3359978B1|2020-01-08|Method for processing an sar image and associated target-detecting method
    EP2333481B1|2014-10-15|Optoelectronic system and method for creating three-dimensional identification images
    Tison et al.2004|Retrieval of building shapes from shadows in high resolution SAR interferometric images
    Lichti2005|Spectral filtering and classification of terrestrial laser scanner point clouds
    Shi et al.2011|Object-based change detection using georeferenced UAV images
    Guan et al.2019|A novel algorithm for estimating the relative rotation angle of solar azimuth through single-pixel rings from polar coordinate transformation for imaging polarization navigation sensors
    Wu et al.2014|A comparison of 4 shadow compensation techniques for land cover classification of shaded areas from high radiometric resolution aerial images
    EP0863488A1|1998-09-09|Method for detecting level contours in two stereoscopic images
    EP0588815B1|2000-06-07|Road image sequence analysis device and method for obstacle detection
    Tupin2010|Fusion of optical and SAR images
    Gamba et al.2011|Limits and challenges of optical high-resolution satellite remote sensing for urban applications
    Carrano2009|Ultra-scale vehicle tracking in low spatial resolution and low frame-rate overhead video
    Sportouche et al.2012|How to combine TerraSAR-X and Cosmo-SkyMed high-resolution images for a better scene understanding?
    EP0810496B1|2001-07-04|Method and device for the identification and localisation of fixed objects along a path
    WO2015082293A1|2015-06-11|Identification of shapes by image correction
    Saidi et al.2016|A refined automatic co-registration method for high-resolution optical and sar images by maximizing mutual information
    Bulatov et al.2016|Segmentation methods for detection of stationary vehicles in combined elevation and optical data
    Lopez et al.2017|Automatic change detection using very high-resolution SAR images and prior knowledge about the scene
    EP2769360B1|2016-02-03|Method for locating objects by resolution in the three-dimensional space of the scene
    Leitloff et al.2007|Inferring traffic activity from optical satellite images
    WO2016132068A1|2016-08-25|Method for determining bathymetry from quasi-synchronous optical satellite images
    Kaynarca et al.2018|Detection of Urban Buildings by Using Multispectral Gokturk-2 and Sentinel 1A Synthetic Aperture Radar Images
    同族专利:
    公开号 | 公开日
    EP3359978A1|2018-08-15|
    EP3359978B1|2020-01-08|
    FR3042283B1|2019-07-19|
    WO2017060000A1|2017-04-13|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    EP2180337A1|2007-08-17|2010-04-28|PASCO Corporation|Terrestrial object information judging image producing method and program|
    US20120274505A1|2011-04-27|2012-11-01|Lockheed Martin Corporation|Automated registration of synthetic aperture radar imagery with high resolution digital elevation models|CN113096058A|2021-04-23|2021-07-09|哈尔滨工业大学|Spatial target multi-source data parametric simulation and MinCenterNet fusion detection method|
    CN108761444B|2018-05-24|2021-12-21|中国科学院电子学研究所|Method for calculating ground point height by combining satellite-borne SAR and optical image|
    CN113420838B|2021-08-20|2021-11-02|中国科学院空天信息创新研究院|SAR and optical image classification method based on multi-scale attention feature fusion|
    法律状态:
    2016-09-28| PLFP| Fee payment|Year of fee payment: 2 |
    2017-04-14| PLSC| Publication of the preliminary search report|Effective date: 20170414 |
    2017-09-29| PLFP| Fee payment|Year of fee payment: 3 |
    2018-09-28| PLFP| Fee payment|Year of fee payment: 4 |
    2019-09-27| PLFP| Fee payment|Year of fee payment: 5 |
    2020-10-13| PLFP| Fee payment|Year of fee payment: 6 |
    优先权:
    申请号 | 申请日 | 专利标题
    FR1502127A|FR3042283B1|2015-10-09|2015-10-09|METHOD OF PROCESSING RADAR IMAGE OF SAR TYPE AND METHOD OF DETECTING TARGET THEREOF|
    FR1502127|2015-10-09|FR1502127A| FR3042283B1|2015-10-09|2015-10-09|METHOD OF PROCESSING RADAR IMAGE OF SAR TYPE AND METHOD OF DETECTING TARGET THEREOF|
    EP16754305.7A| EP3359978B1|2015-10-09|2016-08-23|Method for processing an sar image and associated target-detecting method|
    PCT/EP2016/069874| WO2017060000A1|2015-10-09|2016-08-23|Method for processing an sar image and associated target-detecting method|
    [返回顶部]