Baseline-assisted scene partition SAR imaging method and system

Abstract

The invention relates to the technical field of radar imaging, and particularly discloses a baseline-assisted scene partition SAR imaging method and system, and the method comprises the steps: obtaining more than two SAR images of a monitoring region through setting more than one baseline; strong scattering points of each SAR image are determined; estimating the offset of a strong scattering point of the main image under each baseline; determining a plane area and a slope area according to the offset; respectively imaging the plane area and the slope area; and splicing the images of the plane area and the slope area. According to the principle that imaging projection coordinates of target points of a flat ground area and a slope area can deviate under different base lines due to different elevations, division of the flat ground area and the slope area of a monitoring scene is completed. And finishing imaging processing on the flat ground area by adopting a flat ground hypothesis principle. And for the slope region, an approximate optimal imaging surface is constructed by taking the image entropy as an evaluation index. And splicing the flat ground area image and the slope area image to obtain a high-resolution and low-error SAR image.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of radar imaging technology, and particularly relates to a baseline-assisted scene partitioning SAR imaging method and system. BACKGROUND

[0002] Synthetic Aperture Radar (SAR) is a radar imaging technology that synthesizes a large aperture by using the relative motion between the radar and the target to obtain high-resolution ground surface images or data. With the development of small unmanned aerial vehicle technology, airborne SAR systems have unique advantages in complex terrain scenes such as mountain terrain mapping, disaster monitoring and assessment. However, the anti-interference ability of small unmanned aerial platforms is limited, and the motion track deviates from the ideal motion track due to air flow disturbance, resulting in track errors. At the same time, in the monitoring area lacking accurate digital elevation model (DEM), due to unknown terrain elevation, there is an elevation error. Therefore, the motion track error of the radar load platform and the unknown elevation error are coupled to form a phase error, which seriously restricts the quality of small unmanned aerial SAR imaging.

[0003] In order to obtain high-quality SAR images, the currently commonly used algorithms can be divided into frequency domain algorithms and time domain algorithms. The frequency domain algorithms mainly include range-Doppler domain algorithm, Chirp-scaling algorithm and Omega-K algorithm, etc. These algorithms transform the echo signal into the azimuth frequency domain, and complete the motion error compensation in the azimuth frequency domain. However, most of the frequency domain algorithms are based on the assumption of azimuth invariance, and when the nonlinear track motion error is large, although the motion error compensation algorithm can obtain certain effect, the imaging effect in the high-resolution and high-precision scene is not ideal. The time domain imaging algorithm is an imaging algorithm that compensates the phase error in the time domain, which can perform imaging processing under any flight trajectory. Since it does not depend on the assumption of azimuth invariance and can avoid the complex range-azimuth coupling problem, the time domain imaging algorithm has great imaging performance advantage.

[0004] Although the frequency domain algorithm and the time domain algorithm can obtain high-quality imaging results in the case of satellite-borne or airborne SAR systems, due to the limited flight height of the unmanned aerial vehicle, the low-frequency track error caused by air flow disturbance will produce nonlinear coupling effect with the terrain elevation error. This phenomenon is particularly prominent in slope areas, when the slope angle exceeds 15°, the elevation estimation error will expand 3-5 times for every 1 meter of equivalent along-track position error, and the air-varying quadratic phase error generated thereby will significantly reduce the azimuth resolution. In contrast, flat area targets only show geometric positioning deviation and do not cause significant defocusing phenomenon.

[0005] For the UAV-borne SAR imaging technology, in recent years, various solutions have been proposed. However, for complex terrain scenes (in addition to slope areas, there are also flat areas), the existing methods still have the following limitations: (1) the joint parameterization model of height error and motion error has not been established; (2) the phase compensation mechanism lacks terrain gradient constraint, resulting in insufficient decoupling of slope area error. SUMMARY

[0006] The present application provides a baseline-assisted scene partitioning SAR imaging method and system, which solves the technical problem of UAV-borne SAR imaging in complex terrain measurement, which is limited by the coupled phase error introduced by motion error and height error.

[0007] To solve the above technical problems, the present application provides a baseline-assisted scene partitioning SAR imaging method, comprising the steps of:

[0008] Receiving echo signals and obtaining 2 or more SAR images of the monitoring area under 2 or more baselines;

[0009] Determining the strong scattering points of each SAR image based on the amplitude of the pixel points of each SAR image;

[0010] Selecting one SAR image as the main image and the remaining SAR images as auxiliary images, and estimating the offset of the strong scattering points of the main image in the auxiliary images under each baseline;

[0011] Defining the strong scattering points whose offset in the auxiliary images of the main image under each baseline is greater than a preset offset as being located in the flat area, and defining the other strong scattering points as being located in the slope area;

[0012] Imaging the flat area and the slope area respectively to obtain the flat area image and the slope area image;

[0013] Splicing the flat area image and the slope area image to obtain the SAR imaging of the entire monitoring area.

[0014] Further, the offset of the strong scattering points of the main image in the auxiliary images under each baseline is estimated, specifically:

[0015] A matching window with a side length of N is constructed with the coordinate of a strong scattering point of the main image as the center, and a search window with a side length of M is constructed with the coordinate of the strong scattering point in the auxiliary image as the center;

[0016] The correlation coefficient between the matching window and the block with the size of the matching window in the search window is calculated pixel by pixel, and the strong scattering point coordinate with the maximum correlation coefficient is found;

[0017] The offset between the matching window and the search window is calculated according to the strong scattering point coordinate with the maximum correlation coefficient;

[0018] According to the above steps, the offset of each strong scattering point in the main image on different auxiliary images is obtained.

[0019] Further, the offset between the matching window and the search window is calculated according to the coordinates of the strong scattering point with the largest correlation coefficient, specifically: subtracting the coordinates of the strong scattering point in the matching window from the coordinates of the strong scattering point with the largest correlation coefficient to obtain the offset between the matching window and the search window.

[0020] Further, the correlation coefficient τ between the matching window and the block with the matching window size in the search window is defined as Wherein, I m represents the matching window, I s represents the search window, cov(I m ,I s ) represents the covariance between I m and I s , and var() represents the variance.

[0021] Further, the strong scattering point of each SAR image is determined based on the amplitude of the pixel points of each SAR image, specifically:

[0022] The pixel points with an amplitude greater than the average amplitude in the SAR image are taken as the strong scattering points.

[0023] Further, the preset offset is defined as 1 / 4 pixel unit.

[0024] Further, the imaging of the slope region comprises the steps of:

[0025] Determining the direction of the slope surface;

[0026] Establishing a model of the elevation and ground distance of the target point on the slope surface based on the determined direction of the slope surface;

[0027] Using image entropy or contrast to fit the model of the elevation and ground distance of the target point on the slope surface to obtain the optimal imaging surface.

[0028] Further, the determination of the direction of the slope surface specifically comprises the steps of:

[0029] Calculating the distance difference Δd of the projected coordinates on the imaging surface under the baseline condition;

[0030] Taking the derivative of Δd with respect to the elevation z p ;

[0031] Determining the direction of the slope surface according to the value of the derivative.

[0032] Further, when the flat region image and the slope region image are spliced, weighted averaging is performed on the overlapping region.

[0033] The application further provides a baseline-assisted scene partition SAR imaging system, which is characterized by comprising a coarse imaging module, a strong scattering point screening module, a strong scattering point offset estimation module, a terrain partition module, a partition fine imaging module and an image splicing module.

[0034] The coarse imaging module is configured to receive echo signals and obtain two or more SAR images of a monitoring area under one or more baselines.

[0035] The strong scattering point screening module is configured to determine strong scattering points of each SAR image based on pixel point amplitudes of each SAR image.

[0036] The strong scattering point offset estimation module is configured to select one SAR image as a main image and the remaining SAR images as auxiliary images, and estimate offsets of the strong scattering points of the main image in the auxiliary images under the baselines.

[0037] The terrain partition module is configured to define strong scattering points with offsets of the strong scattering points of the main image in the auxiliary images under the baselines greater than a preset offset as being located in a flat area, and define other strong scattering points as being located in a slope area.

[0038] The partition fine imaging module is configured to perform imaging on the flat area and the slope area respectively to obtain a flat area image and a slope area image.

[0039] The image splicing module is configured to splice the flat area image and the slope area image to obtain a SAR image of the entire monitoring area.

[0040] The baseline-assisted scene partition SAR imaging method and system provided by the application obtain two or more SAR images of a monitoring area under one or more baselines, determine strong scattering points of each SAR image, estimate offsets of the strong scattering points of a main image under the baselines, determine a flat area and a slope area according to the offsets, perform imaging on the flat area and the slope area respectively, and splice the flat area image and the slope area image. According to the principle that imaging projection coordinates under different baselines will be offset due to different elevations of target points in the flat area and the slope area, the application completes partition of the flat area and the slope area in the monitoring scene. Imaging processing of the flat area can be completed by using the flat assumption principle. For the slope area, an approximate optimal imaging surface is constructed by taking image entropy as an evaluation index. The flat area image and the slope area image are spliced to obtain a SAR image with high resolution and low error. Experimental results verify the effectiveness of the proposed method and system. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is a UAV-borne SAR imaging geometry diagram provided by an embodiment of the application;

[0042] Figure 2 is a complex scene area target SAR imaging schematic diagram provided by an embodiment of the application;

[0043] Figure 3 is a complex scene area 3D schematic diagram provided by an embodiment of the application;

[0044] Figure 4 is a projection diagram of a target point under different baselines provided by an embodiment of the application;

[0045] Figure 5 is a first point target schematic diagram provided by an embodiment of the application;

[0046] Figure 6 is a target point coordinate diagram provided by an embodiment of the application; Figure 5

[0047] Figure 7 is a target point offset gradient value schematic diagram provided by an embodiment of the application; Figure 5

[0048] Figure 8 is a second point target schematic diagram provided by an embodiment of the application;

[0049] Figure 9 is a target point coordinate diagram provided by an embodiment of the application; Figure 8

[0050] Figure 10 is a target point offset gradient value schematic diagram provided by an embodiment of the application; Figure 8

[0051] Figure 11 is a baseline-assisted scene partitioning SAR imaging method flowchart provided by an embodiment of the application;

[0052] Figure 12 is an offset estimation schematic diagram provided by an embodiment of the application;

[0053] Figure 13 is a schematic diagram provided by an embodiment of the application;

[0054] Figure 14 is a slope angle schematic diagram provided by an embodiment of the application;

[0055] Figure 15 is a radar track diagram provided by an embodiment of the application;

[0056] Figure 16 is a point target schematic diagram in an experiment provided by an embodiment of the application;

[0057] Figure 17 is a different method imaging result diagram provided by an embodiment of the application;​​​​​

[0058] Figure 18 is the distance slice and azimuth slice diagram of imaging results of different methods provided by the embodiment of the present application;

[0059] Figure 19 is the display diagram of the surface domain structure provided by the embodiment of the present application;

[0060] Figure 20 is the surface domain imaging result diagram provided by the embodiment of the present application. DETAILED DESCRIPTION

[0061] The embodiments of the present application are specifically described below with reference to the drawings. The embodiments are given only for the purpose of illustration and cannot be understood as limiting the present application. The accompanying drawings are only for reference and illustration and do not constitute a limitation on the scope of patent protection of the present application, because many changes can be made to the present application without departing from the spirit and scope of the present application.

[0062] Figure 1 is the imaging geometry diagram of the unmanned aerial SAR. As shown in Figure 1 , according to the positional relationship between the flight track of the unmanned aerial vehicle and the monitored scene, an X-Y-Z three-dimensional coordinate system is constructed. The motion speed of the unmanned aerial vehicle (velocity v) is the Y axis; the vertical motion track and the direction along the scene are the positive direction of the X axis; the plane constructed in the vertical direction of the X axis and the Y axis and pointing to the direction of the sky is the positive direction of the Z axis; the green dotted line is the ideal motion track of the unmanned aerial vehicle, and the red curved line is the real motion track of the unmanned aerial vehicle; it is assumed that any target point P in the monitored scene has an oblique distance R p from the radar antenna.

[0063] For any target point P, the coordinate value can be represented as (x p ,y p ,h p ), the instantaneous coordinates of the phase center of the radar antenna are (X(t a ),vt a ,H(t a )), and the instantaneous oblique distance between the phase center of the radar antenna and the target point can be represented as:

[0064]

[0065] where t a represents the azimuth time, H represents the radar height, Δx(t a ) represents the motion error in the X direction, and Δz(t a ) represents the motion error in the Z direction.

[0066] Transforming equation (1), we have:

[0067]

[0068] where, the custom parameters

[0069] From equation (2), the instantaneous slant range of UAV SAR is composed of three parts, the first part is the ideal flight path, the second part is the instantaneous slant range with the height h p , the third part is the instantaneous slant range with the flight path error.

[0070] Now, the slant range of zero Doppler plane is analyzed, Figure 2 SAR imaging diagram of complex scene area target, where (a), (b) are respectively the SAR imaging diagram of flat area and slope area. From equation (2), when the detection area is flat, that is, the target point height h p = 0, the target point P coincides with the projection point P', as shown in (a) of Figure 2 , the instantaneous slant range between the radar antenna phase center and the target point can be expressed as:

[0071]

[0072] At this time, the UAV SAR system is equipped with a high-precision navigation system to record the radar motion trajectory, and based on the "flat assumption" theory, the accurate instantaneous slant range can be obtained, and the high-quality SAR image can be obtained by using the BP imaging algorithm.

[0073] However, when the target point height h p ≠ 0, the target point will be projected on the imaging plane, as shown in (b) of Figure 2 , the target point P will be projected to the imaging plane P', and its instantaneous slant range can be expressed as:

[0074]

[0075] From equation (4), when the target point height is not zero, the mathematical expression of its instantaneous slant range R(t a ) is related to the motion error and h p . In practice, since the height of the target point is unknown, if the traditional "flat assumption" theory is still used, residual error ΔR(t a ) will be generated, which can be expressed as:

[0076]

[0077] Transforming equation (5) into residual phase error, which can be expressed as:

[0078]

[0079] At this time, due to the existence of residual phase error, when imaging by using the traditional BP algorithm, the imaging result will appear serious azimuth defocus phenomenon, resulting in the quality of the imaging result is reduced.

[0080] In some application scenarios, such as Figure 3 As shown in the road traffic monitoring scene schematic diagram, there are flat areas and mountain areas in the whole monitoring scene. For the flat area, based on the flat assumption theory, the traditional BP algorithm can be used to obtain high-quality images. However, for the slope area, if it is still based on the flat assumption theory, the result obtained will be seriously defocused in the azimuth direction. Therefore, in order to obtain high-quality SAR images for this kind of scene, the first prerequisite is to determine the slope area and the flat area in the region.

[0081] For the complex monitoring scene where flat ground and slope exist at the same time, due to the limitation of the flight height of the unmanned aerial vehicle, the elevation change of the slope area cannot be ignored during imaging. In order to obtain high-quality SAR images of the whole scene, the first prerequisite is to divide the monitoring area, such as distinguishing the flat area and the slope area, and then respectively imaging processing the single sub-area to obtain the sub-image of the sub-area, and then splicing the sub-images to obtain the final SAR image.

[0082] Figure 4 The projection schematic diagram of the target point under different baselines, wherein the baseline refers to the distance difference between the two motion trajectories (flight path). As shown in Figure 4 For the target point P, the slant range difference Δr under the baseline can be expressed as:

[0083]

[0084] Wherein, α represents the baseline angle, and B represents the baseline length.

[0085] From formula (7), it can be seen that the slant range difference Δr is related to the radar height H, the target elevation h p , the baseline length B, the baseline angle α, and the ground distance x p For a certain imaging geometric scene, the slant range difference Δr is only related to the target elevation h p and the ground distance x p .

[0086] The derivative of Δr with respect to h p can be expressed as:

[0087]

[0088] Wherein, the parameter N is defined as:

[0089] N = 2 (H-h p ) B sin α + B 2 -2x p B cos α (9)

[0090] According to the actual situation, The value of x p and h p occupy the main part, so The sign needs to be discussed separately. Because the calculation is very complex, it is related to the specific value of each parameter, which is not discussed here.

[0091] Using the BP algorithm, the echo signals under different baselines are processed to obtain SAR images under different baselines. Using pixels with high SNR ratio, using correlation coefficient method, the coordinate offset of the same scattering point is estimated. When the offset changes little or does not change, it is considered that the position of the scattering point is in the flat area; When the offset changes, it is considered that the height of the scattering point changes, which is in the slope area; According to this principle, the flat area and the slope area can be distinguished;

[0092] Figure 5 The three-dimensional coordinate diagram of nine target points is shown in the figure, wherein the heights of target points 1, 4, 7, 2, 5 and 8 are all 0, and the heights of target points 3, 6 and 9 are 50. Figure 6 The target point coordinates on different baselines are Figure 7 The gradient value of the offset of the corresponding target point. From Figure 7 It can be seen that the gradient value of the offset of target points 1, 4, 7, 2, 5 and 8 is 0, and the gradient value of the offset of target points 3, 6 and 9 changes with the baseline, which presents a first-order linear curve, that is, because the height of the target point is not 0, the offset presents an approximate linear change with the change of the baseline. Therefore, according to this principle, the flat area and the slope area can be distinguished with the aid of SAR images under different baselines.

[0093] Figure 8 The nine target points are shown in the figure, wherein the heights of target points 1, 4 and 7 are 50, the heights of target points 2, 5 and 8 are 100, and the heights of target points 3, 6 and 9 are 150. After BP imaging processing, the target point coordinate values and offset gradient values under different baselines are shown in Figure 9 and Figure 10 From Figure 9 and Figure 10 It can be seen that with the change of the baseline, the coordinate value decreases, and the gradient of the coordinate value is not equal to zero. That is, the offset of the coordinates is not consistent at different heights.

[0094] Therefore, based on the above analysis, by estimating the offset of the same scattering point under different baselines, the ground area can be divided into flat areas and slope areas, as well as the slope surface direction of the slope area. The flat areas and slope areas are accurately imaged separately, and the boundaries of the flat areas and slope areas are accurately fitted using the least squares method, thus completing the precise imaging of complex site areas. Figure 11 This is the implementation flow chart of the baseline-assisted scene partitioning SAR imaging method proposed in this paper. The specific implementation steps include:

[0095] Coarse imaging: Receive echo signals and obtain more than two SAR images of the monitoring area under more than one baseline;

[0096] Strong scattering point screening: Determine the strong scattering points of each SAR image based on the pixel amplitude of each SAR image;

[0097] Strong scattering point offset estimation: Select one SAR image as the main image and the remaining images as auxiliary images, and estimate the offset of the strong scattering points in the main image in the auxiliary images under each baseline;

[0098] Terrain zoning: Strong scattering points whose offsets under each baseline are greater than the preset offset are defined as being located in the flat area, and other strong scattering points are defined as being located in the slope area;

[0099] Partitioned precision imaging: Images of the flat area and slope area are taken separately to obtain images of the flat area and slope area;

[0100] Image stitching: stitch the flat area image and the slope area image to obtain the SAR imaging of the entire monitoring area.

[0101] (1) Coarse imaging

[0102] In this embodiment, a multi-baseline-assisted approach is used to divide the monitoring area into a flat area and a slope area. The first requirement is to image the entire area. Since the initial elevation of the entire monitoring scene is not determined, the traditional "flat ground assumption" imaging mode is used for imaging processing, and the algorithm used is the back projection algorithm. In this embodiment, two moving tracks form a baseline, and two moving tracks have two SAR images. Assuming that N B / 2 baselines to get N B SAR images.

[0103] The length of the baseline plays an important role in determining the target points on the slope within the monitoring area. Figure 4 As shown, for a height h p The projection offset Δr of the two SAR images when the baseline length is B can be expressed as:

[0104]

[0105] The baseline length can be represented as:

[0106]

[0107] The value of baseline B<0 is removed.

[0108] In this paper, the flat area and slope area are determined by using multi-baseline assistance, and the direction of the slope surface is determined, wherein the number of baselines determines the complexity of the proposed algorithm. According to the above analysis, based on one baseline and enough high SNR scattering points, the flat area and slope area can be determined.

[0109] (2) Strong scattering point screening

[0110] In order to distinguish the flat area and slope area in the detection area, in this embodiment, the strong scattering points are screened, and the offset of the strong scattering points projected under different baseline conditions is used to judge whether the strong scattering points belong to the flat area or the slope area, so as to distinguish the flat area and the slope area. In this embodiment, the pixel point with an amplitude greater than the average amplitude is taken as a strong scattering point I, which can be represented as:

[0111]

[0112] Wherein, mean(·) represents the average value, represents the amplitude of the pixel point converted into dB form.

[0113] (3) Strong scattering point offset estimation

[0114] The first image N1 is taken as the main image, and the remaining (N B -1) amplitude images are taken as auxiliary images. The correlation coefficient is used to estimate the offset of the strong scattering point in the main image in other baseline images. A matching window I m with a side length of N is constructed with the coordinate of a strong scattering point in the main image as the center. s As shown in the offset estimation schematic diagram of Figure 12 , the correlation coefficient τ between the matching window and the block with the size of the matching window in the search window is calculated:

[0115]

[0116] Wherein, E(X) represents the average value of X, cov(I m ,I s ) represents the covariance of I m and I scovariance between the two windows, var() denotes variance, I m (i,j) denotes the pixel in the i-th row and j-th column of the matching window, I s (i,j) denotes the pixel in the i-th row and j-th column of the search window.

[0117] According to equation (13), the correlation coefficient of each pixel in the search window I s with the matching window I m is calculated, and the coordinate value of the strongest scattering point with the largest correlation coefficient is determined:

[0118] (m max ,n max ) = max{τ (M-N+1,M-N+1)} (14)

[0119] τ (M-N+1,M-N+1) represents the (M-N+1) x (M-N+1) correlation coefficient array between the matching window and the search window, and max{} represents the maximum value.

[0120] The offset (r max ,r max ) of the matching window I m of the primary image and the search window I s of the secondary image can be calculated by subtracting the coordinate of the strongest scattering point in the matching window from (m m ,n n ), thereby obtaining the offset of each strongest scattering point in the primary image on different secondary images.

[0121] (4) Terrain partitioning

[0122] According to the correlation coefficient method, the offset of the strongest scattering point under different baselines is obtained wherein the subscripts m and n represent the coordinates of the strongest scattering point, the superscript represents the label of the baseline, and represent the offset of the strongest scattering point with coordinates (m, n) in the N s th baseline.

[0123] When , the strongest scattering point corresponding to the coordinates (m, n) can be defined as being located in a flat area, wherein ε is a preset offset. When , the strongest scattering point corresponding to the coordinates (m, n) can be defined as being located in a slope area.

[0124] When partitioning the flat area and the slope area in the SAR image, how to determine the threshold value of partitioning, i.e., the preset offset, is a problem worth considering. In this paper, the preset offset is set to 1 / 4 of the resolution unit.

[0125] (5) Partitioning precise imaging

[0126] 1) Slope area imaging

[0127] Due to the influence of the height, the scattering points in the slope area have a serious defocus phenomenon under the influence of the coupling phase error of the height and the motion error. Therefore, the accurate height information of the target point is needed in the imaging. By establishing the model of the height of the target point on the slope surface and the ground distance, the image entropy or contrast is used to complete the fitting of the optimal imaging surface. However, the premise of the method is to know the direction of the slope surface. However, in practice, the direction of the slope surface is unknown, so the model of the height of the target point on the slope surface and the ground distance cannot be established.

[0128] For the target point with a height of z p , the coordinate X c of the target point on the imaging projection surface can be expressed as:

[0129]

[0130] In the case of the baseline B, assuming that the baseline angle is β, for the target point with a height of z p , the coordinate of the target point on the imaging projection surface can be expressed as:

[0131]

[0132] Then, for the target point with a height of h p , the distance difference Δd of the projection coordinate on the imaging surface under the condition of the baseline B can be expressed as:

[0133]

[0134] The derivative of Δd with respect to z p can be expressed as:

[0135]

[0136] In order to further analyze the sign of , H = 200, β = 30°, B = 5, x p ∈(200, 1500), z p ∈(1, 100), The values of Figure 13 are shown in the table below. It can be seen that in the entire range, that is, under the same baseline, for the target point on the slope, as z p increases, Δd decreases. Therefore, based on this principle, the direction of the slope surface can be determined.

[0137] According to the relationship between the projection point offset of the scattering point under different baselines, when the baseline increases, the imaging projection offset of the scattering point with height will decrease when the baseline lengthens. Therefore, in the embodiment, the direction of the slope is determined according to the change of the imaging projection offset value of the target point with height under different baselines under the assistance of multiple baselines.

[0138] As shown in Figure 14 , it is assumed that the distance of the starting point of the slope region is x s , where the blue curve represents the real ground of the real monitoring region, the black straight line represents the constructed slope surface, and the included angle of the slope is α. The relationship between the height h p of the slope surface and the distance X from the ground can be expressed as:

[0139] h p = (x p -x s ) · tan (α) (19)

[0140] where x s = N · c / 2f s , c represents the speed of light, f s represents the sampling rate, and N represents the starting coordinate point of the slope region.

[0141] In order to accurately fit the trend of the slope surface, as shown in the green straight line in Figure 14 , the included angle is α', and the search of α' can be realized by changing the size of α, which can be expressed as:

[0142] h' p = (x' p -x s ) · tan (α') (20)

[0143] According to the above analysis, when the direction of the slope surface is determined, the high-quality radar image imaging of the slope region can be completed.

[0144] 2) Imaging of flat region

[0145] For the flat region, based on the principle of "flat ground assumption", the back projection imaging algorithm (BP) can be used to obtain high-quality SAR images.

[0146] (6) Image stitching

[0147] When the SAR images of each sub-region of the detection region are determined, the SAR imaging of the entire monitoring region can be completed by stitching. In order to ensure the stitching effect of the sub-SAR images, the imaging region has a certain overlapping region when imaging the sub-region. The SAR images of the overlapping region are weighted and averaged, and the phase information between the sub-regions is combined, so as to maintain the phase continuity at the boundary.

[0148] It should be noted that the various forms of flow shown above can be used to reorder, add or delete steps. For example, each step described in the present application can be performed in parallel, sequentially or in a different order, as long as the desired results of the technical solutions of the present application can be achieved, and the present embodiment does not limit here.

[0149] Based on the baseline-assisted scene partition SAR imaging method described above, the embodiment of the present application further provides a baseline-assisted scene partition SAR imaging system, which comprises a coarse imaging module, a strong scattering point screening module, a strong scattering point offset estimation module, a terrain partition module, a partitioned fine imaging module and an image splicing module.

[0150] The coarse imaging module is used to receive echo signals and obtain two or more SAR images of a monitoring area under one or more baselines. The strong scattering point screening module is used to determine strong scattering points of each SAR image based on the amplitudes of the pixel points of each SAR image. The strong scattering point offset estimation module is used to select one SAR image as a main image and the remaining SAR images as auxiliary images, and estimate the offsets of the strong scattering points of the main image in the auxiliary images under each baseline. The terrain partition module is used to define the strong scattering points whose offsets in the auxiliary images of the main image under each baseline are all greater than a preset offset as being located in a flat area, and define other strong scattering points as being located in a slope area. The partitioned fine imaging module is used to image the flat area and the slope area respectively to obtain a flat area image and a slope area image. The image splicing module is used to splice the flat area image and the slope area image to obtain SAR imaging of the entire monitoring area.

[0151] The embodiments described herein can be implemented in a computing system that includes a back-end component, e.g., as a data server, or that includes a middleware component, e.g., an application server, or that includes a front-end component, e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here, or any combination of such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication, e.g., a communication network. Examples of communication networks include a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0152] Computer programs implementing the methods and systems of the present application can be written in any combination of one or more programming languages, all of which are set out in computer readable program code. These computer programs can be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the computer program running on the processor implements the functions / operations specified in the flow diagrams and / or block diagrams. The computer program can execute entirely on a machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.

[0153] The computer readable storage medium can be a tangible medium that can include or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer readable storage medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, the computer readable storage medium can be a machine readable signal medium. More specific examples of the computer readable storage medium will include one or more lines of a program of instructions in a transitory signal form, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0154] The effectiveness of the proposed method and system is verified by taking 8 isolated points as imaging objects, Figure 15 for the motion trajectory of the UAV, Figure 16 for the target points in the whole scene, 4 of which are located in the flat area, i.e. the elevation is 0, and the other 4 are located in the slope area, i.e. the elevation is not 0.

[0155] Figure 17 The imaging results of different methods are shown in FIGS. 8-11, wherein Figure 17 (a) of FIG. 8 is the imaging result of the BP algorithm under the "flat assumption" condition, Figure 17 (b) of FIG. 8 is the imaging result of the proposed method and system. From Figure 17 it can be seen that in the areas with a distance of 400 meters and 600 meters, the azimuth can well focus the results because the target points are located in the flat area, i.e. the elevation is 0; in the areas with a distance of 800 meters and 1000 meters, the elevation of the target points is not 0, as shown in (a) of FIG. 8, and the four points can see obvious defocusing phenomenon in the azimuth, and the higher the elevation of the target points, the more obvious the defocusing phenomenon; in order to solve the azimuth defocusing phenomenon caused by different elevations, the partition imaging method is proposed in this paper, and the optimal elevation of the slope surface is fitted, as shown in (b) of FIG. 8, and the imaging result of the proposed method and system is shown in (c) of FIG. 8. Figure 17 Figure 17 ​As shown in (b) of FIG. 6, four scattering points have no defocusing phenomenon in the azimuth direction.

[0156] In order to further verify the effectiveness of the proposed method and system, the imaging results are sliced, and the results are as shown in FIG. 7. Figure 18 As shown in FIG. 7, (a) is the azimuth direction slice of the imaging result under the condition of the “flat ground assumption”, (b) is the azimuth direction slice of the imaging result of the proposed method and system. Figure 18 As shown in (a) of FIG. 7, when the elevation of the target point is 0, the focusing effect is good, as shown by points A and B; when the elevation of the target point is not 0, the azimuth direction defocusing is serious, as shown by points C and D, which indicates that under the condition of the flat ground assumption, when the target has an elevation, the imaging result will have a defocusing phenomenon, which is consistent with the theoretical derivation. Figure 18 Figure 18 As shown in FIG. 8, (a) is a three-dimensional display diagram of the surface domain structure, and (b) is a two-dimensional display diagram of the surface domain isolated point structure.

[0157] As shown in FIG. 8, (a) is a three-dimensional display diagram of the surface domain structure, and (b) is a two-dimensional display diagram of the surface domain isolated point structure. Figure 19 In the range of 0 to 120 meters in the distance direction, the elevation coordinate of the isolated target point is 0; in the range of 120 meters to 220 meters in the distance direction, it is a slope area, and the elevation of the isolated point increases linearly with the change of the distance direction.

[0158] Figure 20 As shown in FIG. 9, (a) is the imaging result of the “flat ground assumption”, and (b) is the imaging result of the proposed method. Figure 2 As shown in (a) of FIG. 9, because the imaging is performed under the flat ground assumption, the scattering points have a deviation in the distance direction; as shown in (b) of FIG. 9, because the imaging is performed on the real terrain plane, the deviation caused by the elevation error of the target point is compensated, and the scattering points are projected onto the real ground. Figure 20

[0159] As shown in FIG. 9, (a) is the imaging result of the “flat ground assumption”, and (b) is the imaging result of the proposed method.

[0160] ​​The above embodiments are the preferred embodiments of the present application, but the embodiments of the present application are not limited to the above embodiments, and any changes, modifications, substitutions, combinations, simplifications, etc. made without departing from the spirit and principles of the present application should be equivalent replacement manners and should be included in the protection scope of the present application.

Claims

1. A baseline-assisted scene partitioning SAR imaging method, characterized in that: Including steps: Receive echo signals and obtain more than two SAR images of the monitoring area under more than one baseline; Determine the strong scattering points of each SAR image based on the pixel amplitude of each SAR image; Select one SAR image as the main image and the rest of the SAR images as auxiliary images, and estimate the offset of the strong scattering points of the main image in the auxiliary images under each baseline; Strong scattering points of the main image whose offsets of the auxiliary image under each baseline are greater than the preset offset are defined as being located in the plane area, and other strong scattering points are defined as being located in the slope area; Imaging the flat area and the slope area respectively to obtain a flat area image and a slope area image; The flat area images and slope area images are spliced ​​together to obtain the SAR imaging of the entire monitoring area.

2. The baseline-assisted scene partitioning SAR imaging method according to claim 1, characterized in that: Estimate the offset of the auxiliary image under each baseline of the strong scattering point of the main image, specifically: A matching window with a side length of N is constructed with the coordinates of a strong scattering point in the main image as the center, and a search window with a side length of M is constructed with the coordinates of the strong scattering point in the auxiliary image as the center; Calculate the correlation coefficient between the matching window and the block of the matching window size in the search window pixel by pixel, and find the coordinates of the strong scattering point with the largest correlation coefficient; The offset between the matching window and the search window is calculated based on the coordinates of the strong scattering point with the largest correlation coefficient; Following the same steps as above, the offset of each strong scattering point in the main image on different auxiliary images is obtained.

3. The baseline-assisted scene partitioning SAR imaging method according to claim 2, characterized in that: The offset between the matching window and the search window is calculated according to the coordinates of the strong scattering point with the largest correlation coefficient. Specifically, the offset between the matching window and the search window is obtained by subtracting the coordinates of the strong scattering point in the matching window from the coordinates of the strong scattering point with the largest correlation coefficient.

4. The baseline-assisted scene partitioning SAR imaging method according to claim 3, characterized in that: The correlation coefficient τ between the matching window and the block of the matching window size in the search window is defined as Among them, I m represents the matching window, I s Represents the search window, cov(I m ,I s ) indicates I m and I s The covariance between them, var() represents the variance.

5. The baseline-assisted scene partitioning SAR imaging method according to claim 4, characterized in that: The strong scattering points of each SAR image are determined based on the pixel amplitude of each SAR image, specifically: The pixels in the SAR image with amplitudes greater than the average amplitude are regarded as strong scattering points.

6. The baseline-assisted scene partitioning SAR imaging method according to claim 1, characterized in that: The preset offset is defined as 1 / 4 pixel unit.

7. The baseline-assisted scene partitioning SAR imaging method according to claim 6, characterized in that: Imaging the slope area includes the following steps: Determine the direction of the slope surface; A model of the elevation and ground distance of a target point on the slope surface is established based on the determined direction of the slope surface; Image entropy or contrast is used to fit the model of the elevation and ground distance of the target point on the slope surface to obtain the optimal imaging surface.

8. The baseline-assisted scene partitioning SAR imaging method according to claim 7, characterized in that: Determining the direction of the slope surface specifically includes the following steps: Calculate the distance difference Δd of the projection coordinates on the imaging surface under baseline conditions; Find the value of Δd with respect to the height z p The derivative of The direction of the slope surface is determined based on the value of the derivative.

9. The baseline-assisted scene partitioning SAR imaging method according to claim 8, characterized in that: When stitching the flat area image and the slope area image, weighted averaging is performed on the overlapping areas.

10. A baseline-assisted scene partitioning SAR imaging system, characterized in that: It includes a coarse imaging module, a strong scattering point screening module, a strong scattering point offset estimation module, a terrain partitioning module, a partition fine imaging module and an image stitching module; The coarse imaging module is used to receive the echo signal and obtain two or more SAR images of the monitoring area under one or more baselines; The strong scattering point screening module is used to determine the strong scattering points of each SAR image based on the pixel amplitude of each SAR image; The strong scattering point offset estimation module is used to select one SAR image as the main image and the remaining SAR images as auxiliary images, and estimate the offset of the strong scattering points of the main image in the auxiliary images under each baseline; The terrain partitioning module is used to define strong scattering points of the main image whose offsets of the auxiliary images under each baseline are greater than a preset offset as being located in a plane area, and other strong scattering points as being located in a slope area; The partitioned precise imaging module is used to image the flat area and the slope area respectively to obtain a flat area image and a slope area image; The image stitching module is used to stitch the flat area image and the slope area image to obtain the SAR imaging of the entire monitoring area.

Images