Scene DEM extraction method based on CSAR sub-aperture correlation

Abstract

The invention discloses a scene DEM extraction method based on CSAR sub-aperture correlation, and relates to the technical field of radar imaging, and the method comprises the steps: dividing sub-aperture echo data of different visual angles; performing imaging processing to obtain a complex numerical sub-aperture image; binarization processing is carried out to change the sub-aperture image into an image with only a target and a background, and the centroid position of a target area is extracted; registration of centroid points of the sub-aperture images of different viewing angles is carried out; calculating a target position offset and a target height according to the position offset geometrical relationship; dEM information of all the sub-aperture images is calculated based on the target height, the DEM information of the sub-aperture images at different visual angles is integrated, and omni-directional scene DEM information is obtained. According to the method, the offset of different sub-aperture imaging is calculated by using the center of mass of point target imaging, the calculation process is simple, the complexity is low, the calculation efficiency is improved, the target position offset is solved through the view angle difference, the omni-directional scene DEM information is calculated, and the method has higher precision and anti-noise capability.

Description

Technical Field

[0001] The present invention relates to the field of radar imaging technology, and more specifically, to a scene DEM extraction method based on CSAR sub-aperture correlation. Background Art

[0002] Synthetic Aperture Radar (SAR) obtains high resolution in the range direction by emitting a wide bandwidth signal. At the same time, the radar platform observes the scene target at a wide angle and can obtain high resolution in the azimuth direction. SAR imaging can obtain more electromagnetic scattering information of the observed target by reconstructing the target scattering function, which helps to analyze, classify and identify the target characteristics. SAR can provide all-weather and all-day reconnaissance capabilities, has significant advantages in remote sensing observation, and has a wide range of uses.

[0003] Circular SAR (CSAR) refers to a SAR platform that moves in a circular trajectory around the observation scene, with its antenna beam continuously pointing to the target observation scene. The all-round observation of the CSAR mode can obtain images with higher resolution, and can achieve multi-angle observation and imaging of the target, thereby obtaining more complete target information and improving image resolution.

[0004] In existing airborne CSAR imaging experiments, laser radar is usually used to obtain the digital elevation model (DEM) information of the reference scene as the reference data of the imaging plane. However, since CSAR needs to obtain DEM information over a large range, laser radar still faces challenges and limitations in obtaining DEM information. For example, the laser radar system usually requires expensive equipment and maintenance fees, the laser radar band is difficult to penetrate clouds and dense vegetation, and is easily affected by weather and flight detection altitude restrictions. It cannot effectively extract the DEM information of the observed scene, which affects the further application and development of CSAR technology. The method of extracting DEM information using interferometric SAR has a complex processing flow, high algorithm complexity, and high requirements for the accuracy of the registration and phase unwrapping algorithms. If the high resolution and omnidirectional observation capabilities of the CSAR system can be effectively utilized to directly extract the scene elevation information from the CSAR data itself, more accurate and detailed terrain height information can be provided. In addition, since the CSAR system can work under various weather conditions, it has obvious advantages over optical and infrared remote sensing technologies, and can realize integrated imaging processing, from data acquisition to image processing to result output, real-time imaging can be achieved in a relatively short time.

[0005] In order to solve the above problems, the present invention makes full use of the advantages of CSAR in extracting DEM information and proposes a scene DEM extraction method based on CSAR sub-aperture correlation. Summary of the invention

[0006] The purpose of the present invention is to overcome the problems of low precision and complex processing flow in extracting DEM information of target scene in the prior art, make full use of the multi-angle observation characteristics of CSAR, extract DEM information of target scene from sub-aperture imaging, and provide a scene DEM extraction method based on CSAR sub-aperture association.

[0007] To achieve the above object, the technical solution of the present invention is as follows:

[0008] A scene DEM extraction method based on CSAR sub-aperture correlation includes the following steps:

[0009] The full-aperture echo data acquired by CSAR is divided into multiple sub-aperture echo data of different viewing angles along the azimuth direction;

[0010] Using a CSAR imaging algorithm to perform imaging processing on each of the sub-aperture echo data to obtain a corresponding complex-valued sub-aperture image;

[0011] Binarizing each of the complex-valued sub-aperture images to convert them into images with only two values, target and background, and extracting the centroid position of the target area in each image;

[0012] According to the centroid position, performing centroid point registration between sub-aperture images of different viewing angles;

[0013] According to the focus position offset geometric relationship, the target position offset of all sub-aperture images at different viewing angles after the centroid point registration is calculated, and the target height is calculated based on the target position offset;

[0014] Based on the known information including the target height, the DEM information of each sub-aperture image is calculated, and the DEM information of sub-aperture images of different viewing angles is integrated to obtain the omnidirectional scene DEM information.

[0015] As a preferred solution, the sub-aperture echo data of the multiple different viewing angles have the same size, and each sub-aperture echo data contains a phase history for subsequent sub-aperture imaging. When dividing, the sub-aperture length and overlap amount need to be determined, and the resolution needs to be weighed.

[0016] As a preferred solution, the CSAR imaging algorithm includes a back projection (BP) algorithm, and GPU acceleration is used for parallel processing during the imaging process.

[0017] As a preferred solution, an adaptive binarization algorithm based on Otsu's method is used to perform the binarization process.

[0018] As a preferred solution, the step of extracting the centroid position includes: calculating the connected area, setting an area threshold of the connected area, excluding isolated noise points whose area is smaller than the connected area threshold, and extracting the centroid position of the connected area.

[0019] As a preferred solution, the centroid point registration includes: according to the centroid position extracted from each sub-aperture image, finding the corresponding centroid point for registration through the nearest neighbor matching metric, and using chain association calculation to traverse all sub-aperture images to complete the centroid point registration of all sub-aperture images.

[0020] As a preferred solution, the chain association calculation is specifically as follows: a certain sub-aperture image is selected as the first sub-aperture image and the centroid point of the first sub-aperture image is used as a reference. The first sub-aperture image has centroid points with a sequence of 1, 2, 3, ..., n. The centroid points with a sequence of 1, 2, 3, ..., n in the first sub-aperture image are traversed. The centroid point closest to the centroid point with a sequence of 1 in the first sub-aperture image is found in the second sub-aperture image. The sequence number of this centroid point in the second sub-aperture image is also marked as 1. The other centroid points are similarly performed. All centroid points of all sub-aperture images are traversed to complete the one-to-one correspondence of centroid points.

[0021] As a preferred solution, the target position offset of all sub-aperture images at different viewing angles after the centroid point registration is calculated according to the focus position offset geometric relationship, and the target height is calculated based on the target position offset, specifically:

[0022] Ignore the change of CSAR beam pitch angle θ in the target scene area and set the imaging reference plane to H ref =z p -Δh, the imaging projection positions of the point target P in the two sub-aperture images are P A and P B , according to the geometric relationship, their coordinates are respectively expressed as:

[0023]

[0024] From the above formula, it can be seen that when there is an imaging position offset between the two sub-aperture images of the point target P in CSAR, it can be obtained that the height deviation Δh and the imaging position offset Δr = |P A -P B The relationship between | is:

[0025]

[0026] in, is the azimuth angle interval of the sub-aperture centers.

[0027] As a preferred solution, the height deviation |Δh| should satisfy:

[0028]

[0029] Among them, k max The maximum wave number of the signal transmitted by the CSAR system.

[0030] As a preferred solution, after integrating the DEM information of the sub-aperture images of different viewing angles, the average value of the integrated DEM information is calculated, and the average value is used as the omnidirectional scene DEM information.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] The present invention calculates the offset of different sub-aperture imaging by using the centroid of point target imaging, without the need for point target position coordinates as prior information. The calculation process is simple, the complexity is low, and the calculation efficiency is improved. The multi-perspective information of sub-aperture imaging is used to obtain the target position offset through the perspective difference, and then the omnidirectional scene DEM information is calculated. Compared with single-perspective imaging, it has higher accuracy and noise resistance. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 A flowchart of the steps of a scene DEM extraction method based on CSAR sub-aperture association in Example 1 of the present application;

[0034] Figure 2 This is a schematic diagram of CSAR imaging when there is a height deviation of the target in Example 2 of the present application;

[0035] Figure 3 A schematic diagram of the three-dimensional geometry of the CSAR mode of Example 2 of the present application;

[0036] Figure 4 The CSAR sub-aperture imaging result diagrams of the target in Example 2 of the present application at 1° to 60°, 61° to 120°, 121° to 180°, 181° to 240°, 241° to 300°, and 301° to 360°;

[0037] Figure 5 This is a diagram illustrating the processing of a scene DEM extraction method based on CSAR sub-aperture association in Example 2 of the present application;

[0038] Figure 6 This is a diagram of the simulation imaging results of 9 point targets in Example 3 of the present application;

[0039] Figure 7 This is a CSAR sub-aperture imaging result diagram of the target at 1° to 60° in Example 3 of the present application;

[0040] Figure 8 This is a CSAR sub-aperture imaging result diagram of the target at 61° to 120° in Example 3 of the present application. DETAILED DESCRIPTION

[0041] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0042] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0043] Example 1

[0044] See also Figure 1 Embodiment 1 of the present application provides a scene DEM extraction method based on CSAR sub-aperture association, comprising:

[0045] S1: Divide the full-aperture echo data acquired by CSAR into multiple sub-aperture echo data with different viewing angles along the azimuth direction.

[0046] The full-aperture echo data is divided into multiple sub-aperture echo data of the same size along the azimuth direction. Each sub-aperture echo data contains a phase history for subsequent sub-aperture imaging. When dividing, the sub-aperture length and overlap amount must be determined, and factors such as resolution must be weighed.

[0047] S2: Use the CSAR imaging algorithm to perform imaging processing on each sub-aperture echo data to obtain the corresponding complex-valued sub-aperture image.

[0048] The CSAR imaging algorithm is used to perform imaging processing on each sub-aperture echo data to obtain the corresponding complex-valued sub-aperture image, and then multiple continuous sub-aperture image sequences are obtained. In view of the good imaging performance of the BP algorithm, this embodiment 1 uses the BP algorithm for sub-aperture imaging processing, and in order to improve the imaging processing efficiency, the GPU-accelerated BP algorithm is used for parallel processing during the processing.

[0049] like Figure 4 As shown in FIG. 1 , six sub-aperture images are obtained in Example 1, which are the sub-aperture imaging results corresponding to 1° to 60°, 61° to 120°, 121° to 180°, 181° to 240°, 241° to 300°, and 301° to 360°, respectively, where the color bar unit is dB. Figure 4It can be found that due to the height deviation between the target and the imaging plane, the focus position of the target in different sub-aperture images will be different, and the sidelobe directions of the sub-aperture imaging will also be different.

[0050] S3: Binarize each complex-valued sub-aperture image into an image with only two values, target and background, and extract the centroid position of the target area in each image.

[0051] Since the ultimate goal is to extract the contour of the target imaging, and the binary image has only two values ​​of 0 and 1, the amount of calculation is small, so each complex value sub-aperture image needs to be binarized and converted into only two values ​​of target and background. In view of the fact that the binarization algorithm based on the Otsu method can adaptively count the pixel distribution value of each image, and select the binarization threshold value that makes the variance between the foreground and background image classes the largest, so that the effect of image binarization is good. Therefore, the present embodiment 1 adopts the adaptive binarization algorithm based on the Otsu method to binarize the sub-aperture image.

[0052] Since the sub-aperture imaging result has a large area value of target imaging, if the offset of the point target under different sub-aperture imaging is calculated in the form of image window sliding, the amount of calculation is too large. In order to reduce the calculation cost of the sub-aperture imaging offset, the present invention calculates the centroid position of the point target area in the sub-aperture image as a reference point for subsequent sub-aperture image registration. Therefore, this step first calculates the connected area, sets the area threshold of the connected area, then excludes isolated noise points whose area is less than the threshold of the connected area, and finally extracts the centroid of the connected area. Subsequent calculations only use the centroid of the point target imaging to calculate the offset of different sub-aperture imaging, which can improve the calculation efficiency.

[0053] S4: Perform centroid point registration between sub-aperture images of different viewing angles according to the centroid position;

[0054] In order to establish the correspondence between target scattering points at different viewing angles, it is necessary to perform centroid registration between different sub-aperture images. In this embodiment 1, according to the centroid position extracted from each sub-aperture image, the corresponding feature point is found for registration through the nearest neighbor matching metric.

[0055] This embodiment 1 adopts chain correlation calculation to traverse all sub-aperture images. Figure 5Step 4 is a schematic diagram of chain-association calculation in the present invention. Specifically, the centroid point of the first sub-aperture image is used as a reference. The first sub-aperture image has centroid points of a sequence of 1, 2, 3, ..., n. Traverse the centroid points of a sequence of 1, 2, 3, ..., n in the first sub-aperture image, and find the centroid point closest to the centroid sequence 1 in the first sub-aperture image in the second sub-aperture image. In the second sub-aperture image, the sequence number of this centroid point is also marked as 1. The other centroid points are similar, and all sub-aperture images are traversed to complete the one-to-one correspondence of the centroid points.

[0056] S5: calculating the target position offset of all sub-aperture images at different viewing angles after the centroid point registration according to the focus position offset geometric relationship, and calculating the target height based on the target position offset;

[0057] By using the geometric relationship of the focus position offset of the target in different sub-aperture imaging, the viewing angle difference between the sub-aperture images and the known viewing angle parameters are solved to calculate the sight direction offset (target position offset) of the corresponding scattering point at different viewing angles, and then calculate the target height. Figure 5 As shown in step 5, the deviation between sub-aperture image 2 and sub-aperture image 1 is calculated, the deviation between sub-aperture image 3 and sub-aperture image 2 is calculated, and so on, traversing all sub-aperture images to obtain the focus position offset of the target in different sub-aperture images (target position offset).

[0058] S6: Based on the known information including the target height, calculate the DEM information of each sub-aperture image, integrate the DEM information of sub-aperture images of different viewing angles, and obtain the omnidirectional scene DEM information.

[0059] In order to reduce the influence of noise, the DEM information obtained from the associated sub-aperture images is integrated and the average value is calculated to obtain a more accurate and comprehensive scene DEM extraction result. The core of the present invention is to use the multi-view information of sub-aperture imaging, obtain the target position offset through the perspective difference, and then calculate the omnidirectional scene DEM information. Compared with single-view imaging, the present invention has higher accuracy and noise resistance.

[0060] Example 2

[0061] This embodiment 2 further explains (1) the impact of target height deviation on CSAR full-aperture imaging quality and (2) the principle of target scene DEM information extraction based on CSAR sub-aperture imaging in a scene DEM extraction method based on CSAR sub-aperture association in embodiment 1, including:

[0062] (1) Impact of target height deviation on CSAR full-aperture imaging quality:

[0063] The focusing effect of CSAR imaging is easily affected by the height of the target. Ideally, when the target is at the preset imaging plane height, the phase change of the echo signal can be fully compensated, thereby obtaining a high-quality CSAR image that can truly reflect the geometric characteristics of the target. On the contrary, if there is a deviation in the imaging plane height, that is, the actual height of the target does not match the imaging plane height, the resulting phase error will cause the target imaging result to undergo geometric deformation, causing the shape, size and position of the target imaging result to change, thereby reducing the imaging quality and the accuracy of target recognition.

[0064] CSAR determines the distance between the target and CSAR by measuring the arrival time of the echo signal. The measured distance is the slant distance of the target along the radar line of sight, not the actual ground projection distance of the target. Therefore, during the imaging process, the target is projected onto an inclined plane with an angle to the ground, which is the slant range projection mode. For targets with a certain height, the slant range projection will cause the target to be projected onto the imaging plane along the line of sight, resulting in an offset between its position in the CSAR image and the actual ground projection position. The offset increases with the increase of the target height and the line of sight angle.

[0065] Figure 2 Schematic diagram of CSAR imaging when there is a deviation in target height. Assuming that the imaging plane height is 0m, the airborne CSAR performs circular motion observation with a radius of R on a plane with a height of H from the imaging plane. The center of the beam always points to the target observation scene, and its circular observation flight trajectory is shown by the blue curve. A is the CSAR in azimuth The position at time, the coordinates are A' is the projection of the CSAR position A on the imaging plane, and its coordinates are I is the target at a height Δh from the imaging plane. Due to the slant range projection mechanism of CSAR, the target I will be projected onto I of the imaging plane during imaging. A Place.

[0066] When CSAR is performing observation imaging, its position coordinate vector is assumed to be Assume the position coordinates of the point target P in the observation area are r p =(x p ,y p ,z p ), then the instantaneous distance from CSAR to point target P is:

[0067]

[0068] When the imaging plane height z0 = 0m, the BP algorithm is used to image the point target P echo signal, then the point spread function of the point target in the imaging plane can be expressed as:

[0069]

[0070] Where k is the wave number of the signal transmitted by the CSAR system, R xy is the distance from CSAR to each grid point (x, y, 0) of the imaging plane grid, and its expression is:

[0071]

[0072] Performing a two-dimensional Fourier transform on equation (2) yields:

[0073]

[0074] Substituting the point spread function g(x,y) into equation (4), we can obtain:

[0075]

[0076] Let Φ = -2kR lp +2kR xy -k x xk y y, then formula (5) can be transformed into:

[0077]

[0078] Calculate the first-order partial derivative of the phase Φ in equation (6) with respect to the variables x and y respectively, and the instantaneous frequency is:

[0079]

[0080] Let the first-order partial derivative be zero, then k x ,k y The relationship between x and y is:

[0081]

[0082] Therefore, formula (6) can be expressed as:

[0083] G(k x ,k y )=exp{jΦ(x(k x ),y(k y ))} (9)

[0084] Using k in formula (8) x ,k y The relationship between x and y is:

[0085]

[0086] Formula (10) is the relationship between the wave number and distance of any grid point (x, y, 0) on the CSAR imaging plane. For a point target P, we have:

[0087]

[0088] Substituting equation (10) and equation (11) into equation (9), we can obtain:

[0089]

[0090] in, Substituting into formula (12), we have:

[0091]

[0092] Performing a two-dimensional inverse Fourier transform on equation (13), we have:

[0093]

[0094] make:

[0095]

[0096] For equation (15), find Ψ(k x ,k y )About (k x ,k y ), then we have:

[0097]

[0098] Let the first-order partial derivative be zero, then:

[0099] (xx p ) 2 +(yy p ) 2 =(z p tanθ) 2 (17)

[0100] By analyzing equation (17), we can know that if there is a height difference between the target position and the set imaging plane, the CSAR imaging quality will be reduced, and a specific geometric deformation will be generated on the imaging plane, which is a (x p ,y p ) is the center of the circle, |z p ·tanθ| is the radius of the circular blur area, which results in the inability to effectively gather energy and low imaging quality.

[0101] In order to ensure that the point target P can be focused with high precision during CSAR full-aperture imaging, it is necessary to ensure that the height difference between the point target and the imaging plane is Δh = |z p-z0| should satisfy: Δh≤Δr·cotθ / 2, where Δr is the resolution of the imaging plane grid. At this time, the imaging result can accurately reflect the actual position of the point target.

[0102] Impact of target height deviation on CSAR sub-aperture imaging quality:

[0103] According to formula (15), the phase error is Ψ e , then:

[0104]

[0105] Will e (k x ,k y ) is converted to polar coordinate format Ψ e (ρ,φ), then:

[0106]

[0107] Where ρ = 2ksinθ, then equation (19) can be rewritten as:

[0108]

[0109] After transformation, the phase error Ψ e It can be expressed as:

[0110] Ψ e =ρtanθ·Δh (21)

[0111] Phase error Ψ e It is a key factor affecting the imaging quality, which can be decomposed into a first-order phase error component Ψ e1 and quadratic and higher-order error components Ψ e2 .

[0112] Ψ e =Ψ e1 +Ψ e2 (twenty two)

[0113] Due to the position error of the CSAR system, the motion of the target, or the instability of the CSAR platform motion, a first-order phase error component Ψ is generated. e1 , directly affects the position offset of the target imaging result on the imaging plane and is the main factor determining the imaging accuracy. Error component Ψ e1 The expression is:

[0114] Ψ e1 =k x Δx+k y ·Δy (23)

[0115] The circular full aperture is divided into multiple sub-apertures, and the center of each sub-aperture is The azimuth is Then we have:

[0116]

[0117] Therefore, formula (23) can be rewritten as:

[0118]

[0119] in, The value range is make Then formula (25) can be rewritten as:

[0120]

[0121] The quadratic and higher-order error components Ψ e2 It usually originates from the nonlinear effect in the signal propagation process, the inhomogeneity of the propagation medium, or the non-ideal characteristics of the imaging system itself, which in turn affects the imaging quality. Combined with equation (22), the quadratic and higher-order error components Ψ e2 for:

[0122]

[0123] Phase error Ψ e2 Should meet |Ψ e2 |<π / 4, so that the imaging result can have a better focusing effect. You can get |Ψ e2 | max for:

[0124]

[0125] At this time, the height deviation |Δh| should satisfy:

[0126]

[0127] According to formula (29), we can know that the sub-aperture azimuth accumulation angle is The size of directly affects the sensitivity of the imaging result to the target height deviation. ), the amount of data and calculation is large, and the imaging result is less sensitive to height deviation, which means that the imaging quality is relatively good and the target can be better focused. ), the amount of data and calculation is small, and the imaging result is more sensitive to the height deviation, which means that the imaging quality is poor and the target imaging may appear blurred or defocused. Therefore, choose a suitable sub-aperture azimuth accumulation angle It is critical to achieve high-quality imaging and to select the appropriate value to balance imaging quality and height measurement accuracy.

[0128] The division of sub-apertures determines the grouping method of CSAR data, which in turn affects the resolution and accuracy of extracting elevation information from these data. Reasonable sub-aperture division can maximize the correlation between CSAR data while reducing the impact of low CSAR imaging quality caused by altitude deviation.

[0129] From the above analysis, it can be seen that when there is an error between the actual height of a point target and the imaging plane, a circular blurred area will appear when the point target is imaged in the CSAR full aperture, affecting the imaging quality. The size of the circular blurred area is closely related to the imaging height deviation, reflecting the importance of measuring the target elevation information.

[0130] (2) Principle of target scene DEM information extraction based on CSAR sub-aperture imaging:

[0131] From the above analysis, it can be seen that the focusing quality of target imaging is closely related to the imaging height deviation. When there is a deviation between the imaging reference plane and the actual height of the scene target, the CSAR slant range projection cannot be projected to the exact image point, and a position offset will occur, thereby affecting the position accuracy of the target imaging and the overall practicality of the imaging results. At the same time, the offset may be different in each sub-aperture imaging. By comparing the offset differences of these images, the height of the target can be estimated. The following analyzes the method of extracting the elevation information of the observation area using sub-aperture imaging in CSAR mode. Figure 3 The three-dimensional geometric diagram of the CSAR mode is shown in Figure 1, and a spatial rectangular coordinate system OXYZ is established. The CSAR platform observes the target by making a circular motion around the OZ axis on a plane at a height of H, and its speed is V.

[0132] Figure 3 (a) is a stereoscopic view of the CSAR mode. Figure 3 (b) is a top view of the CSAR mode, where the blue curve is the circular flight trajectory curve of the CSAR platform, and A is the azimuth of the CSAR. The position of CSAR at P is the point target with a height difference of Δh from the imaging plane, and its coordinates are P(x p ,y p ,z p ). Due to the slant range projection observation mechanism of CSAR, CSAR has When observing at P, P will be projected onto the imaging plane P A At; CSAR in azimuth When observing at P, P will be projected onto P on the imaging plane. B At all points, they deviate from the projection position of P on the imaging plane, causing geometric deformation of the CSAR image and resulting in poor imaging quality. Under CSAR full-aperture imaging, the point target P will be projected as a circular area on the reference imaging plane due to the height difference, such as Figure 3 The green area in (a) and (b) has a radius of:

[0133]

[0134] Where θ is the beam pitch angle of CSAR. From equation (30), it can be seen that the geometric deformation of the subaperture image increases with the increase of the height difference. When the target is exactly on the imaging plane, that is, when the height deviation is zero, the radius of the defocus ring is zero, which means that the point target is well focused in this case and the imaging quality is the highest.

[0135] In order to further illustrate the method of extracting target height, the top view of the geometric model of the target with height error in CSAR two-dimensional imaging is analyzed, as shown in Figure 3 Assume that the beam pitch angle θ changes negligibly in the target scene area and set the imaging reference plane to H ref =z p -Δh, the imaging projection positions of the point target P in the two sub-aperture images are P A and P B According to the geometric relationship, their coordinates are expressed as:

[0136]

[0137] From formula (31), we can see that when the imaging position of the point target P between the two CSAR sub-aperture images is offset, we can get the height difference Δh and the imaging position offset Δr = |P A -P B |, that is:

[0138]

[0139] in, is the azimuth angle interval of the sub-aperture centers.

[0140] It can be seen from formula (32) that the height of the target can be calculated by analyzing the imaging position offset between the two sub-aperture images, and the DEM information of the entire scene area can be obtained by traversing all sub-aperture observation angles.

[0141] Example 3

[0142] In order to prove the effectiveness of the scene DEM extraction method based on CSAR sub-aperture association proposed in Example 1, it is verified through point target simulation experiments of the observation scene. This Example 3 is based on Example 1 and Example 2.

[0143] The parameters of the point target experiment simulation system are shown in Table 1. It is easy to calculate that the CSAR beam look-down observation angle is 45°.

[0144] Table 1 Point target simulation system parameters

[0145]

[0146]

[0147] There are 9 point targets with the same scattering coefficient and different heights in the observation scene, and the imaging plane height is 0m. Their two-dimensional positions and heights are shown in Table 2.

[0148] Table 2 Point target coordinate position

[0149]

[0150] The imaging results of 9 point targets in the observation scene are as follows: Figure 6 As shown. It can be found that when the imaging plane height is 0m, the imaging quality of the point target with a height of 2m is better than that of the point target with a height greater than 2m. As the height of the point target increases, the energy distribution becomes more dispersed, the resolution becomes lower, and the radius of the defocus ring becomes larger, indicating that the intensity of the reflected signal decreases with the increase of the height deviation, which is consistent with the theoretical analysis. At the same time, the defocus ring shows a consistent change trend, indicating that CSAR imaging has good consistency and predictability in the entire observation range, which is very important for calibrating and optimizing the imaging process. Among them, the actual height of the point target C is 18m. Substituting the system parameters set by the simulation into formula (30), the theoretical defocus ring radius is 18m, and it is found from the imaging results that the defocus ring radius is also 18m, which is consistent with the theoretical analysis.

[0151] In order to verify the effectiveness of the method of extracting scene DEM using sub-aperture imaging, the sub-aperture imaging results of 1° to 60° and 61° to 120° are given, as shown in Figure 7 , Figure 8 As shown. Figure 7 , Figure 8 It can be found that due to the height difference between the target and the imaging plane, the same target has an obvious position offset in different sub-aperture imaging, and the sidelobe direction of the target imaging is also different. The sidelobe direction is consistent with the sub-aperture center angle, which is consistent with the theoretical analysis.

[0152] It can be found that the greater the difference between the actual height of the target and the height of the imaging plane, the greater the position offset. The height difference between target C and the imaging plane is 18m, and the imaging position offset of target C under the sub-apertures of 1°~60° and 61°~120° is 32.79m; the height difference between target H and the imaging plane is 4m, and the imaging position offset of target H under the sub-apertures of 1°~60° and 61°~120° is 7.31m, which is smaller than the offset of target C. Therefore, according to the relationship between the offset and the sub-aperture angle of the target under different sub-aperture imaging, the true height of the target can be extracted, and then the DEM information of the scene can be extracted.

[0153] The DEM information of 9 point targets in the scene is extracted by using the scene DEM extraction method based on CSAR sub-aperture association of the present invention. The actual height of the point target and the height extraction results are shown in Table 3. It can be found that for multiple point targets with height deviations from the imaging plane, this method can extract more accurate actual heights of the point targets, with an error within 0.1m, and for some point targets, the height extraction error is within 0.05m, which reflects the effectiveness of the method proposed by the present invention.

[0154] Table 3. True height of point targets and extraction results

[0155]

[0156] The other technical details and steps of this embodiment 3 are the same as those of embodiments 1 and 2 and will not be repeated here.

[0157] It can be seen from the above embodiments that the present invention proposes a scene DEM extraction method based on CSAR sub-aperture association, and the idea of ​​proposing the method is described in detail from the impact of target height deviation on CSAR full-aperture imaging and sub-aperture imaging, and the principle of DEM extraction. The DEM information extracted by this method is more accurate and has better adaptability. In addition, this method does not require point target position coordinates as prior information, the calculation process is simple, the complexity is low, and the multi-view information of sub-aperture imaging is used to obtain the target position offset through the view difference, and then calculate the omnidirectional scene DEM information. Compared with single-view imaging, it has higher accuracy and noise resistance. The target scene DEM information extracted by the present invention can provide important source data for terrain mapping and terrain undulation error compensation.

[0158] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the embodiments here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A scene DEM extraction method based on CSAR sub-aperture correlation, characterized in that: The steps include: The full-aperture echo data acquired by CSAR is divided into multiple sub-aperture echo data of different viewing angles along the azimuth direction; Using a CSAR imaging algorithm to perform imaging processing on each of the sub-aperture echo data to obtain a corresponding complex-valued sub-aperture image; Binarizing each of the complex-valued sub-aperture images to convert them into images with only two values, target and background, and extracting the centroid position of the target area in each image; According to the centroid position, performing centroid point registration between sub-aperture images of different viewing angles; According to the focus position offset geometric relationship, the target position offset of all sub-aperture images at different viewing angles after the centroid point registration is calculated, and the target height is calculated based on the target position offset; Based on the known information including the target height, the DEM information of each sub-aperture image is calculated, and the DEM information of sub-aperture images of different viewing angles is integrated to obtain the omnidirectional scene DEM information.

2. The scene DEM extraction method based on CSAR sub-aperture correlation according to claim 1, characterized in that: The sub-aperture echo data of the multiple different viewing angles have the same size, and each sub-aperture echo data contains a phase history for subsequent sub-aperture imaging. When dividing, the sub-aperture length and overlap amount need to be determined, and the resolution needs to be weighed.

3. The scene DEM extraction method based on CSAR sub-aperture correlation according to claim 1 is characterized in that: The CSAR imaging algorithm includes a back-projection BP algorithm, and GPU acceleration is used for parallel processing during the imaging process.

4. The scene DEM extraction method based on CSAR sub-aperture correlation according to claim 1 is characterized in that: The binarization process is performed using an adaptive binarization algorithm based on the Otsu method.

5. The scene DEM extraction method based on CSAR sub-aperture correlation according to claim 1 is characterized in that: The step of extracting the centroid position includes: calculating the connected area, setting a connected area area threshold, excluding isolated noise points whose area is smaller than the connected area threshold, and extracting the centroid position of the connected area.

6. The scene DEM extraction method based on CSAR sub-aperture correlation according to claim 5, characterized in that: The centroid point registration includes: according to the centroid position extracted from each sub-aperture image, finding the corresponding centroid point for registration through nearest neighbor matching measurement, and using chain association calculation to traverse all sub-aperture images to complete the centroid point registration of all sub-aperture images.

7. The scene DEM extraction method based on CSAR sub-aperture correlation according to claim 6 is characterized in that: The chain association calculation is specifically as follows: a certain sub-aperture image is selected as the first sub-aperture image and the centroid point of the first sub-aperture image is used as a reference, the first sub-aperture image has centroid points with a sequence of 1, 2, 3, ..., n, the centroid points with a sequence of 1, 2, 3, ..., n in the first sub-aperture image are traversed, the centroid point closest to the centroid point with a sequence of 1 in the first sub-aperture image is found in the second sub-aperture image, the sequence number of this centroid point is also marked as 1 in the second sub-aperture image, and the other centroid points are similarly performed, all centroid points of all sub-aperture images are traversed, and a one-to-one correspondence of centroid points is completed.

8. The scene DEM extraction method based on CSAR sub-aperture correlation according to claim 1, characterized in that: The target position offset of all sub-aperture images at different viewing angles after the centroid point registration is calculated according to the focus position offset geometric relationship, and the target height is calculated based on the target position offset, specifically: Ignore the change of CSAR beam pitch angle θ in the target scene area and set the imaging reference plane to H ref =z p -Δh, the imaging projection positions of the point target P in the two sub-aperture images are P A and P B , according to the geometric relationship, their coordinates are respectively expressed as: From the above formula, it can be seen that when there is an imaging position offset between the two sub-aperture images of the point target P in CSAR, it can be obtained that the height deviation Δh and the imaging position offset Δr = |P A -P B The relationship between | is: in, is the azimuth angle interval of the sub-aperture centers.

9. The scene DEM extraction method based on CSAR sub-aperture correlation according to claim 8, characterized in that: The height deviation |Δh| should satisfy: Among them, k max The maximum wave number of the signal transmitted by the CSAR system.

10. The scene DEM extraction method based on CSAR sub-aperture correlation according to claim 1, characterized in that: After integrating the DEM information of the sub-aperture images of different viewing angles, the average value of the integrated DEM information is calculated, and the average value is used as the omnidirectional scene DEM information.

Images