Space-based distributed video SAR moving target joint track-before-detect method

Abstract

The invention discloses a space-based distributed video SAR (Synthetic Aperture Radar) moving target joint track-before-detect method, which mainly solves the problem that the high-speed maneuvering target detection and tracking capability in a single-platform space-based video SAR system is limited. According to the implementation scheme, the method comprises the following steps: constructing a space-based distributed video SAR collaborative detection model; sequentially connecting short synthetic apertures formed by a plurality of space-based video SAR in the model in series to construct two equivalent long synthetic apertures, performing imaging focusing on the two equivalent long synthetic apertures to obtain an image, and performing cancellation processing to obtain a difference video SAR image sequence after static clutter suppression; performing pre-detection, clustering and matching on the difference image of the sequence to obtain a shadow pre-detection point set; performing speed matching on the pre-detection points to realize state initialization; shadow and energy joint value function accumulation is carried out based on the shadow initial state; and judging a target according to the accumulated value. According to the invention, the imaging time is greatly shortened, the detection capability on the moving target is improved, and the method can be used for detecting and tracking the ground moving target of the space-based distributed video synthetic aperture radar.

Description

Technical Field

[0001] The invention belongs to the field of radar technology, and in particular relates to a moving target pre-detection tracking method, which can be used for ground moving target shadow detection and tracking by space-based distributed video synthetic aperture radar. Background Art

[0002] Video synthetic aperture radar (SAR) is a typical high frame rate imaging system that can dynamically monitor hot spots and continuously track moving targets. Therefore, video SAR has great application potential in external environment situation awareness and ground moving target indication (GMTI). In recent years, space-based video SAR has attracted widespread attention due to its strong survivability, wide detection range and freedom from national restrictions.

[0003] However, space-based video SAR still faces some problems and challenges that have not been effectively solved. On the one hand, space-based video SAR faces the contradiction between high frame rate imaging and long-distance detection. Under the premise of ensuring the azimuth resolution of imaging, increasing the radar operating carrier frequency is an effective way to increase the imaging frame rate. However, due to the influence of atmospheric attenuation of electromagnetic waves and the current limitation of the power level of high-frequency devices, the detection range of high-frequency band SAR systems is severely limited. Conventional single-unit space-based video SAR cannot simultaneously meet the application requirements of high frame rate imaging and long-distance detection. On the other hand, space-based video SAR also faces the contradiction between high-resolution imaging and high-speed maneuvering target detection. An effective way for video SAR to achieve azimuth high resolution is to increase the length of the synthetic aperture to obtain a larger accumulation angle. However, a long synthetic aperture will lead to more serious range migration and imaging defocusing of high-speed maneuvering targets. In addition, in video SAR, a long synthetic aperture will also cause the shadow of a high-speed moving target to face problems such as blurred boundaries and reduced contrast with the background, and even unable to form a visible shadow. Therefore, the method of increasing the azimuth resolution by increasing the synthetic aperture is not conducive to the detection and tracking of high-speed maneuvering targets.

[0004] At present, there are few literature reports on moving target detection and tracking of space-based video SAR, and there are mainly two technologies:

[0005] First, Cai Xuelian proposed a moving target track generation method based on the high-low orbit collaborative video SAR imaging model in her paper "Research on high-low orbit collaborative dual-base Video-SAR imaging and target tracking methods". The solution is to determine the initial position and speed of the target based on the first frame of dual-channel detection results after clutter cancellation, and use a local weighted dynamic programming pre-detection tracking algorithm to generate the target track.

[0006] Second, Li Jinwei proposed a moving target detection process for a single-channel spaceborne video SAR image sequence in his paper "A Constant False Alarm Detection Method for Moving Targets in Spaceborne Video SAR". The main steps of this method include preprocessing, change detection factor calculation, single-frame detection threshold determination, single-frame detection fusion and motion trajectory detection. While ensuring a constant false alarm rate, the statistical characteristics of SAR images and the laws of target motion are reasonably used to achieve robust moving target detection.

[0007] However, both of the above methods study the problem of moving target detection in single-platform space-based video SAR, without considering the two major contradictions in actual space-based video SAR systems: high frame rate imaging and long-distance detection, high-resolution imaging and high-speed maneuvering target detection. This affects the detection and tracking performance of moving targets, especially the further improvement of the detection and tracking capabilities of high-speed maneuvering targets, thus limiting their application. Summary of the invention

[0008] The purpose of the present invention is to address the deficiencies of the above-mentioned prior art and propose a joint pre-detection tracking method for ground moving targets of space-based distributed video SAR to improve the detection and tracking performance of moving targets, especially to improve the detection and tracking capability of high-speed maneuvering targets.

[0009] To achieve the above object, the implementation steps of the technical solution of the present invention include the following:

[0010] (1) Constructing collaborative detection model:

[0011] Set up N video SARs, these space-based video SARs maintain a constant spacing and move in a straight line configuration, and use a spotlight mode to observe the same target area;

[0012] The difference frequency Δf is designed to transmit a series of instantaneous frequency-divided orthogonal linear frequency modulation continuous wave signals with time synchronization and very small difference frequency, and when receiving, each space-based video SAR only retains the echo signal corresponding to its own carrier frequency through orthogonal isolation;

[0013] (2) Assume that the space-based distributed video SAR system performs M observations in total. In the kth and k+1th observations, two equivalent long synthetic apertures α satisfying the DPCA condition are constructed by sequentially connecting multiple short synthetic apertures formed by multiple space-based video SARs. k and β k+1 ;

[0014] (3) Using the frequency domain imaging algorithm represented by polar coordinate formatting, the equivalent long synthetic aperture α k and β k+1 Focusing is performed to obtain the video SAR images corresponding to the two apertures. and Finally, two sets of video SAR image sequences are obtained and The difference video SAR image sequence Π={ΔI1,...,ΔI k ,...,ΔI M-1},in is the difference video SAR image obtained for the kth time;

[0015] (4) Pre-detection and clustering are performed on the difference video SAR images ΔI1 and ΔI2, and the shadow and energy matching of the same moving target is achieved based on the range coordinates, and the shadow pre-detection point sets Γ1 and Γ2 of the images ΔI1 and ΔI2 are obtained respectively:

[0016]

[0017] Where r = 1, ..., R, q = 1, ..., Q, R and Q are the total number of pre-detection points on ΔI1 and ΔI2 respectively, are the rth and qth shadow detection points on ΔI1 and ΔI2 respectively, and are the azimuth offsets of the shadow detection point and energy detection point of the same moving target on ΔI1 and ΔI2 respectively;

[0018] (5) Based on the shadow pre-detection point set Γ1 and Γ2, the state is initialized by velocity matching to obtain the shadow initial state set in is the initial state of the shadow of the mth candidate target in the first frame, M t is the total number of candidate targets to be tracked;

[0019] (6) Based on the initial state of the shadow of the mth candidate target in the first frame In M-1 frame video SAR image sequence The shadow and energy joint value function is accumulated on:

[0020] (6a) Let the current frame be the hth frame, where h∈[1,M-1);

[0021] (6b) For the shadow state in the hth frame Perform inter-frame state transfer to obtain the shadow state in the h+1th frame

[0022]

[0023] in I2 is the second-order identity matrix, represents the Kronecker product of the matrix, δ x and δ y are the acceleration process noise in the x-direction and y-direction respectively, and Δt is the time interval between two adjacent frames;

[0024] (6c) The shadow state of the mth candidate target in the h+1th frame Mapping to energy states

[0025] (6d) Joint Shadow State and energy status Perform value function accumulation to obtain the cumulative value

[0026]

[0027] in is the inverse shadow amplitude, is the energy amplitude, are all states that can be transferred to in the hth frame The state set of

[0028] (7) Repeat step (6) to set the value function Accumulate to the M-1th frame, and then compare it with the set detection threshold V α Compare and determine whether the target to be detected exists:

[0029] like Then the mth target to be detected exists, and its trajectory in the M-1 frame is traced back.

[0030] Otherwise, the target to be detected does not exist;

[0031] (8) Repeat step (7) to complete M t Joint pre-detection tracking of multiple targets to be detected.

[0032] Compared with the prior art, the present invention has the following advantages:

[0033] 1) The present invention greatly shortens the imaging time while ensuring the imaging resolution by sequentially connecting multiple short synthetic apertures formed by multiple space-based video SARs into an equivalent long synthetic aperture. This not only enables a higher imaging frame rate to be obtained in the low-frequency band, thus avoiding the contradiction between high-frame-rate imaging and long-distance detection of space-based video SAR, but also facilitates the detection of highly maneuverable targets, thus avoiding the contradiction between high-resolution imaging and high-speed maneuverable target detection of space-based video SAR.

[0034] 2) The present invention performs joint shadow and energy pre-detection tracking in ground moving target detection, and can use the multi-dimensional characteristics of moving targets to achieve robust detection, thereby improving the detection capability of space-based video SAR for ground moving targets. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flow chart of the implementation of the present invention;

[0036] Figure 2 This is the result of coarse imaging of moving target scene by frequency domain algorithm (PFA);

[0037] Figure 3 for Figure 2 A magnified image of the target area. DETAILED DESCRIPTION

[0038] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be described clearly and completely in conjunction with the drawings in the embodiment of the present invention. Obviously, the described embodiment is only a part of the embodiment of the present invention, not all of the embodiments. Based on the embodiments of the present invention, other embodiments obtained by ordinary technicians in this field without creative work should all fall within the scope of protection of the present invention.

[0039] It should be noted that the step numbers in the specification and claims of the present invention are only for a clear description of the implementation scheme of the present invention to facilitate understanding, and the order of other serial numbers is not limited.

[0040] Reference Figure 1 , the implementation steps of this example include the following:

[0041] Step 1: Build a space-based distributed video SAR collaborative detection model.

[0042] Set up N video SARs, these space-based video SARs maintain a constant spacing and move in a straight line configuration, and use a spotlight mode to observe the same target area;

[0043] The difference frequency Δf is designed to transmit a series of instantaneous frequency-divided orthogonal linear frequency modulation continuous wave signals that are time-synchronized and have a very small difference frequency, and when receiving, each space-based video SAR only retains the echo signal corresponding to its own carrier frequency through orthogonal isolation;

[0044] The difference frequency Δf is designed as follows:

[0045] To ensure that the echo signals of each space-based video SAR can be processed coherently, the difference frequency Δf should be much smaller than the transmission signal bandwidth B;

[0046] To ensure that the echo signals of different video SAR can be orthogonally isolated, the difference frequency Δf should be set larger than the detection scene bandwidth B. r =2w r γ / c, that is, B r <Δf<B, where w r is the imaging scene width, γ is the frequency modulation slope, and c is the speed of light.

[0047] Step 2: construct a long synthetic aperture that meets the DPCA conditions.

[0048] 2.1) Assume that the space-based distributed video SAR system conducts M observations in total;

[0049] 2.2) In the kth and k+1th observations, two equivalent long synthetic apertures α satisfying the DPCA condition are constructed by sequentially connecting multiple short synthetic apertures formed by multiple space-based video SARs. k and β k+1 :

[0050] 2.2.2) Within the kth observation time Δt, the short synthetic aperture L formed by the nth space-based video SAR is obtained nk , the equivalent long synthetic aperture α is obtained by sequentially connecting multiple short synthetic apertures of the second to the Nth space-based video SAR k :

[0051] α k =L 2k ||L 3k ||...||L nk ||...||L Nk

[0052] 2.2.3) After a time Δt, the k+1th observation is performed, and the short synthetic aperture formed by the nth space-based video SAR is obtained as L n(k+1) By sequentially connecting multiple short synthetic apertures of the first to the N-1th space-based video SAR, the equivalent long synthetic aperture β is obtained. k+1 :

[0053] β k+1 =L 1(k+1) ||L 2(k+1) ||...||L n(k+1) ||...||L (N-1)(k+1) .

[0054] Step three, perform frequency domain imaging focusing on the long synthetic aperture and cancel to obtain a difference SAR image.

[0055] Frequency domain imaging focusing can be achieved through a frequency domain imaging algorithm, that is, through a range Doppler algorithm, a polar coordinate formatting algorithm, etc. This example adopts but is not limited to a frequency domain imaging algorithm represented by polar coordinate formatting.

[0056] The specific implementation of this step is as follows:

[0057] 3.1) Using the frequency domain imaging algorithm represented by polar coordinate formatting, the equivalent long synthetic aperture α k and β k+1 Focusing is performed to obtain the video SAR images corresponding to the two apertures. and The equivalent length of the synthetic aperture αk and β k+1 Meet DPCA conditions;

[0058] 3.2) Video SAR images and Cancellation is performed, that is, according to the situation where the aperture spatial position is the same but there is a time difference of Δt, the corresponding video SAR image and The amplitude and phase of the stationary clutter are consistent, while the amplitude and phase of the moving target are different. The difference video SAR image after the stationary clutter is suppressed can be obtained by complex image subtraction.

[0059] 3.3) Repeat step 3.1) for a total of M-1 times to obtain two sets of video SAR image sequences and

[0060] 3.4) Repeat step 3.2) for a total of M-1 times, and perform image cancellation to obtain the difference video SAR image sequence Π={ΔI1,...,ΔI k ,...,ΔI M-1}.

[0061] Step 4: perform pre-detection and clustering on the difference SAR image.

[0062] Pre-detection and clustering are performed on the difference video SAR images ΔI1 and ΔI2, and the shadow and energy of the same moving target are matched based on the range coordinates, and the shadow pre-detection point sets Γ1 and Γ2 of the two difference images ΔI1 and ΔI2 are obtained respectively:

[0063] 4.1) Set the detection threshold V T , compare it with the difference video SAR image ΔI1 to determine whether shadow or energy exists:

[0064] If the SAR image value ΔI1(x,y) at the position ΔI1 with the azimuth x and the range y is greater than the set detection threshold, then shadow or energy exists.

[0065] Otherwise, the shadow or energy does not exist;

[0066] The mathematical expression of this judgment is as follows:

[0067]

[0068] where x=1,...,N x ,y=1,...,N y , N x and N yThey represent the number of resolution units in the azimuth and range directions of the SAR image, respectively. H1 indicates the presence of shadow or energy, and H0 indicates the absence of shadow or energy.

[0069] 4.2) Clustering the moving target shadow and energy detection results obtained in step 4.1) using a density clustering algorithm, and then averaging and fusing multiple points in the same class to obtain an accurate shadow or energy detection point;

[0070] 4.3) Matching the shadow and energy of the same moving target based on the range coordinates, respectively obtaining the shadow pre-detection point set Γ1 of the difference image ΔI1 and the shadow pre-detection point set Γ2 of the difference image ΔI2:

[0071] 4.3.1) Based on SAR images The neighborhood pixel mean value of the upper detection point is used to distinguish the results, and all detection points are divided into two categories: shadow detection points and energy detection points;

[0072] 4.3.2) Let the rth shadow detection point be The jth energy detection point is Let Δr be the distance threshold and compare it with the detection point:

[0073] like Then the rth shadow detection point and the jth energy detection point belong to the same moving target, achieving the matching of the shadow and energy of the same moving target, and the azimuth offset is recorded as

[0074] 4.3.3) Repeat steps 4.3.1) to 4.3.2) to traverse all shadow detection points and finally obtain the shadow pre-detection point set

[0075] 4.3.4) Repeat steps 4.3.1) to 4.3.3) on the difference image ΔI2 to obtain a shadow pre-detection point set

[0076] Step 5: Initialize the state of the candidate target through velocity matching based on the shadow pre-detection point sets Γ1 and Γ2 to obtain the shadow initial state set C.

[0077] 5.1) Assume and are the detection points in the shadow pre-detection point sets Γ1 and Γ2 respectively, and the inter-frame coordinate difference Δx in the azimuth direction and the inter-frame coordinate difference Δy in the distance direction are calculated:

[0078]

[0079] 5.2) Assume that the inter-frame velocity range in the azimuth direction is The speed range in the distance direction is Determine the coordinate difference Δx between the azimuth frames and the coordinate difference Δy between the distance frames:

[0080] If the coordinate difference between frames satisfies and Then the detection point is considered as a candidate target, and the initial state of the shadow of the mth candidate target in the first frame is Expressed as:

[0081]

[0082] Otherwise, no action is taken;

[0083] 5.3) Repeat steps 5.1) to 5.2), traverse all detection points in Γ1 and match them with all detection points in Γ2 to obtain the shadow initial state set Among them, M t is the total number of candidate targets to be tracked.

[0084] Step 6: Initial state of the shadow based on the mth candidate target In M-1 frame video SAR image sequence The shadow and energy joint value function is accumulated.

[0085] (6.1) Let the current frame be the hth frame, where h∈[1,M-1);

[0086] (6.2) For the shadow state in the hth frame Perform inter-frame state transfer to obtain the shadow state in the h+1th frame

[0087]

[0088] in represents a four-dimensional matrix, the symbol T represents transpose, I2 is the second-order unit matrix, represents the Kronecker product of the matrix, δ x and δ y are the acceleration process noise in the x-direction and y-direction respectively;

[0089] (6.3) The shadow state of the mth candidate target in the h+1th frame Mapping to energy states

[0090] 6.3.1) According to the azimuth offset of the mth candidate target The shadow state after transferring the state between frames Mapping to initial energy state

[0091]

[0092] in, is the azimuth coordinate of the mth candidate target in the h+1th frame, is the distance coordinate of the mth candidate target in the h+1th frame;

[0093] 6.3.2) Compensate for the deviation between the initial estimated energy state and the actual state caused by the change in the speed of the moving target, that is, The azimuth coordinates are randomly expanded to obtain the expanded pixel coordinate set:

[0094]

[0095] in is the azimuthal coordinate after expansion, which revolves around the original position By taking a uniform distribution U D (x a ,x b ) is sampled D times; is the distance coordinate after expansion;

[0096] 6.3.3) Expand the coordinates and The pixel values ​​of all corresponding states are arranged in descending order, and the state with the highest pixel value is taken. As shadow state The energy state after mapping;

[0097] (6.4) Joint Shadow State and energy status Perform value function accumulation to obtain the cumulative value

[0098]

[0099] in is the inverse shadow amplitude, is the energy amplitude, are all states that can be transferred to in the hth frame The state collection.

[0100] Step seven: Make target decisions based on the joint accumulation value.

[0101] When the target does not exist, the detection threshold V is set by Monte Carlo simulation according to the signal-to-noise ratio. α ,

[0102] Repeat step 6 and accumulate the value function to the M-1th frame to obtain Then compare it with the set detection threshold V αCompare and determine whether the target to be detected exists:

[0103] like Then it is determined that the mth target to be detected exists, and its trajectory in the M-1 frame is traced back.

[0104] Otherwise, it is determined that the target to be detected does not exist.

[0105] Step 8, repeat step 7 to complete M t Joint pre-detection tracking of multiple targets to be detected.

[0106] The effect of the present invention can be further illustrated by the following simulation results:

[0107] 1. Simulation experiment parameters, as shown in Table 1

[0108] Table 1 Space-based cluster SAR simulation parameters

[0109] parameter value Carrier frequency 94GHz Track height 550Km Platform speed 7579m / s Range 800Km Transmit power 500W Transmitting and receiving antenna gain 50dBi Pulse repetition frequency 9KHz Number of sub-platforms 10

[0110] 2. Simulation experiment content

[0111] Under the above simulation parameters, the frequency domain algorithm PFA is used to perform coarse imaging of the moving target scene, that is, to simulate the long aperture imaging of a single platform and cluster SAR. The results are as follows: Figure 2 ,in:

[0112] Figure 2 (a) is the result of single-platform long-aperture imaging.

[0113] Figure 2 (b) is the result of cluster SAR long aperture imaging;

[0114] from Figure 2 (a) and Figure 2 (b) By comparison, we can see that:

[0115] Figure 2 In the single-platform long-aperture imaging process shown in (a), the synthetic aperture length required to achieve a resolution of 1 m is 1212 m, and the accumulation time is 0.1667 s; Figure 2 The cluster SAR multi-platform long-aperture imaging shown in (b) has the same resolution as single-platform long-aperture imaging, but its synthetic aperture time is only one-tenth of that of single-platform long-aperture imaging.

[0116] The comparison results show that the present invention can greatly shorten the imaging time while ensuring the imaging resolution by sequentially connecting multiple short synthetic apertures formed by multiple space-based video SARs into an equivalent long synthetic aperture. It can not only obtain a higher imaging frame rate in the low-frequency band, avoiding the contradiction between high-frame rate imaging and long-distance detection of space-based video SAR, but also is more conducive to the detection of highly maneuverable targets, avoiding the contradiction between high-resolution imaging of space-based video SAR and detection of high-speed maneuverable targets.

[0117] The target area of ​​the above imaging results is enlarged, such as Figure 3 As shown, where:

[0118] Figure 3 (a) is a magnified image of the target area as a result of single-platform long-aperture imaging.

[0119] Figure 3 (b) is a magnified image of the target area as a result of cluster SAR long-aperture imaging;

[0120] from Figure 3 (a) and Figure 3 (b) By comparison, we can see that:

[0121] Figure 3 (a) The single-platform long-aperture imaging leads to more serious range migration and imaging defocus of high-speed maneuvering targets; Figure 3 In (b), by sequentially connecting multiple short synthetic apertures formed by multiple space-based video SARs into an equivalent long synthetic aperture, the target's movement distance can be reduced, that is, the movement of high-speed targets along the distance can be reduced to one-tenth of the original. This shows that by sequentially connecting multiple short synthetic apertures formed by multiple space-based video SARs into an equivalent long synthetic aperture, the detection capability of space-based video SAR for ground moving targets can be improved.

Claims

1. A space-based distributed video SAR moving target joint detection and tracking method, characterized in that: These include: (1) Constructing collaborative detection model: Set up N video SARs, these space-based video SARs maintain a constant spacing and move in a straight line configuration, and use a spotlight mode to observe the same target area; The difference frequency Δf is designed to transmit a series of instantaneous frequency-divided orthogonal linear frequency modulation continuous wave signals with time synchronization and very small difference frequency, and when receiving, each space-based video SAR only retains the echo signal corresponding to its own carrier frequency through orthogonal isolation; (2) Assume that the space-based distributed video SAR system performs M observations in total. In the kth and k+1th observations, two equivalent long synthetic apertures α satisfying the DPCA condition are constructed by sequentially connecting multiple short synthetic apertures formed by multiple space-based video SARs. k and β k+1 ; (3) Using the frequency domain imaging algorithm represented by polar coordinate formatting, the equivalent long synthetic aperture α k and β k+1 Focusing is performed to obtain the video SAR images corresponding to the two apertures. and Finally, two sets of video SAR image sequences are obtained and The difference video SAR image sequence Π={ΔI1,...,ΔI k ,...,ΔI M-1 },in is the difference video SAR image obtained for the kth time; (4) Pre-detection and clustering are performed on the difference video SAR images ΔI1 and ΔI2, and the shadow and energy matching of the same moving target is achieved based on the range coordinates, and the shadow pre-detection point sets Γ1 and Γ2 of the images ΔI1 and ΔI2 are obtained respectively: Where r = 1, ..., R, q = 1, ..., Q, R and Q are the total number of pre-detection points on ΔI1 and ΔI2 respectively, are the rth and qth shadow detection points on ΔI1 and ΔI2 respectively, and are the azimuth offsets of the shadow detection point and energy detection point of the same moving target on ΔI1 and ΔI2 respectively; (5) Based on the shadow pre-detection point set Γ1 and Γ2, the state is initialized by velocity matching to obtain the shadow initial state set in is the initial state of the shadow of the mth candidate target in the first frame, M t is the total number of candidate targets to be tracked; (6) Based on the initial state of the shadow of the mth candidate target in the first frame In M-1 frame video SAR image sequence The shadow and energy joint value function is accumulated on: (6a) Let the current frame be the hth frame, where h∈[1,M-1); (6b) For the shadow state in the hth frame Perform inter-frame state transfer to obtain the shadow state in the h+1th frame in I2 is the second-order identity matrix, represents the Kronecker product of the matrix, δ x and δ y are the acceleration process noise in the x-direction and y-direction respectively, and Δt is the time interval between two adjacent frames; (6c) The shadow state of the mth candidate target in the h+1th frame Mapping to energy states (6d joint shadow state and energy status Perform value function accumulation to obtain the cumulative value in is the inverse shadow amplitude, is the energy amplitude, are all states that can be transferred to in the hth frame The state set of (7) Repeat step (6) to set the value function Accumulate to the M-1th frame, and then compare it with the set detection threshold V α Compare and determine whether the target to be detected exists: like Then the mth target to be detected exists, and its trajectory in the M-1 frame is traced back. Otherwise, the target to be detected does not exist; (8) Repeat step (7) to complete M t Joint pre-detection tracking of multiple targets to be detected.

2. The method according to claim 1, wherein the design principle of the difference frequency Δf in step (1) is as follows: To ensure that the echo signals of each space-based video SAR can be processed coherently, the difference frequency Δf should be much smaller than the transmission signal bandwidth B; To ensure that the echo signals of different video SAR can be orthogonally isolated, the difference frequency Δf should be set larger than the detection scene bandwidth B. r =2w r γ / c, that is, B r <Δf<B, where w r is the imaging scene width, γ is the frequency modulation slope, and c is the speed of light.

3. According to the method of claim 1, in step (2), two equivalent long synthetic apertures α satisfying the DPCA condition are constructed. k and β k+1 , which is expressed as follows: 2a) Within the kth observation time Δt, the short synthetic aperture L formed by the nth space-based video SAR is obtained nk , the equivalent long synthetic aperture α is obtained by sequentially connecting multiple short synthetic apertures of the second to the Nth space-based video SAR k : alpha k =L 2k ||L 3k ||...||L nk ||...||L Nk 2b) After a time Δt, the k+1th observation is performed to obtain the short synthetic aperture L formed by the nth space-based video SAR n(k+1) , by sequentially connecting multiple short synthetic apertures of the first to the N-1th space-based video SAR, the equivalent long synthetic aperture β is obtained. k+1 : β k+1 =L 1(k+1) ||L 2(k+1) ||...||L n(k+1) ||...||L (N-1)(k+1) .

4. According to the method of claim 1, in step (4), pre-detection and clustering are performed on the difference video SAR images ΔI1 and ΔI2 respectively, and matching of shadow and energy of the same moving target is achieved based on the range coordinates, and the implementation steps include the following: 4a) Set the detection threshold V T , compare it with the difference video SAR image ΔI1 to determine whether there is a shadow or energy: if the SAR image value ΔI1(x,y) at the position x in the azimuth direction and y in the range direction on ΔI1 is greater than the set detection threshold, then the shadow or energy exists, otherwise, the shadow or energy does not exist. Its mathematical expression is as follows: where x=1,...,N x ,y=1,...,N y , N x and N y They represent the number of resolution units in the azimuth and range directions of the SAR image, respectively. H1 indicates the presence of shadow or energy, and H0 indicates the absence of shadow or energy. 4b) clustering the moving target shadow and energy detection results obtained in step 4a) using a density clustering algorithm, and then averaging and fusing multiple points in the same class to obtain an accurate shadow or energy detection point; 4c) Matching the shadow and energy of the same moving target based on the range coordinates, and obtaining the shadow pre-detection point set Γ1 of the difference image ΔI1 and the shadow pre-detection point set Γ2 of the difference image ΔI2 respectively: 4c1) Based on SAR images The neighborhood pixel mean value of the upper detection point is used to distinguish the results, and all detection points are divided into two categories: shadow detection points and energy detection points; 4c2) Let the rth shadow detection point be The jth energy detection point is Let Δr be the distance threshold and compare it with the detection point: like Then the rth shadow detection point and the jth energy detection point belong to the same moving target, achieving the matching of the shadow and energy of the same moving target, and the azimuth offset is recorded as 4c3) Repeat steps 4c1) to 4c2) to traverse all shadow detection points and finally obtain the shadow pre-detection point set 4c4) Repeat steps 4c1) to 4c3) on the difference image ΔI2 to obtain a shadow pre-detection point set 5. According to the method of claim 1, step (5) realizes state initialization through velocity matching based on the shadow pre-detection point sets Γ1 and Γ2, and the realization steps include the following: 5a) Set and They are the detection points in the shadow pre-detection point sets Γ1 and Γ2 respectively. If the coordinate difference between frames satisfies and Then the detection point is considered as a candidate target, and its initial state is expressed as: in, is the initial state of the shadow of the mth candidate target in the first frame, That is, the azimuth frame coordinate difference, That is, the distance coordinate difference between frames, and They are the inter-frame speed ranges set in azimuth and distance directions respectively; 5b) Repeat step 5a), traverse all detection points in Γ1 and match them with all detection points in Γ2 to obtain the shadow initial state set Among them, M t is the total number of candidate targets to be tracked.

6. The method according to claim 1, wherein in step (6c), the shadow state of the mth candidate target in the h+1th frame is Mapping to energy states The implementation steps include the following: 6c1) According to the azimuth offset of the mth candidate target The shadow state after transferring the state between frames Mapping to initial energy state in, is the azimuth coordinate of the mth candidate target in the h+1th frame, is the distance coordinate of the mth candidate target in the h+1th frame; 6c2) Compensate for the deviation between the initial energy state and the actual state caused by the change in the speed of the moving target, that is, The azimuth coordinates are randomly expanded to obtain the expanded pixel coordinate set: in is the azimuthal coordinate after expansion, which revolves around the original position By taking a uniform distribution U D (x a ,x b ) is sampled D times; is the distance coordinate after expansion 6c3) Expand the coordinates and The pixel values ​​of all corresponding states are arranged in descending order, and the state with the highest pixel value is taken. As shadow state The energy state after mapping.

Images