CN108196240A - A Trajectory Reconstruction Method for Ground Moving Targets Applicable to CSAR Imaging - Google Patents

Abstract

A ground moving target track reconstruction method suitable for CSAR imaging comprises the following steps: s1, distance pulse pressure is carried out on a received CSAR moving target echo signal, a full-aperture echo of a radar is uniformly divided into a plurality of sub-aperture echoes, the sub-aperture echoes are converted into a distance Doppler domain, moving target detection is realized through Doppler filtering and constant false alarm rate detection, and the position of a moving target in an RD domain is obtained; s2, realizing moving target association under different sub-apertures by using a multi-target tracking algorithm, and acquiring a motion track of the moving target in an RD (RD) domain; and S3, projecting the dynamic target RD domain track to a road grid, and realizing dynamic target track reconstruction by using a dynamic programming algorithm. The method adopts a CSAR mode, has the capability of observing the moving target for a long time, and can improve the reconstruction precision of the target track by increasing road prior information and solving an optimized equation by utilizing dynamic programming.

Description

A Trajectory Reconstruction Method for Ground Moving Targets Applicable to CSAR Imaging

technical field

The invention belongs to the field of synthetic aperture radar (Synthetic Aperture Radar, SAR) ground moving target indication (Ground Moving Target Indication, GMTI), and relates to a ground moving target track reconstruction ( Ground Moving Target Trajectory Reconstruction, GMTTR) method.

Background technique

Synthetic Aperture Radar (SAR) is a radar technology that can perform high-resolution microwave imaging of observation scenes. Initially, the main function of SAR was to obtain static radar images of observation scenes. In order to improve battlefield awareness, people hope that SAR can realize the reconnaissance and detection of ground moving targets while completing the imaging reconnaissance of stationary targets, that is, it has the function of GMTI. SAR-GMTI can complete the imaging reconnaissance of stationary/moving targets at the same time, which greatly expands the application range of SAR technology. Trajectory reconstruction of moving targets can provide accurate intelligence information on the moving state of moving targets, which is one of the important research contents of GMTI. However, the conventional SAR-GMTI system adopts the straight-line flight mode, which does not have the ability to observe the moving target for a long time, and it is difficult to realize the function of reconstructing the trajectory of the moving target.

CSAR refers to a radar system in which the radar platform (or radar station) moves in a large curve or wide-angle arc around the observation scene, and the beam is always pointed at the target scene for observation and imaging. CSAR has the ability to observe and reconnaissance areas for a long time. Correspondingly, CSAR-GMTI can realize long-term observation, tracking and trajectory reconstruction of moving targets in the scene.

While the special working mode of CSAR brings the advantage of all-round observation, it also increases the complexity of the system's detection geometry. How to realize the trajectory reconstruction of moving targets in the CSAR detection mode is a technical problem that needs to be solved urgently.

Contents of the invention

The purpose of the present invention is to provide a ground moving target trajectory reconstruction method suitable for CSAR imaging, so as to improve the performance and practical value of CSAR-GMTI.

In order to realize above-mentioned technical purpose, technical scheme of the present invention is:

A method for reconstructing the trajectory of a ground moving target suitable for CSAR imaging, comprising the following steps:

S1. Perform range pulse pressure on the received CSAR moving target echo signal, divide the full-aperture echo of the radar evenly into several sub-aperture echoes, and transform the sub-aperture echoes into the range Doppler domain, through Doppler Le filter and constant false alarm rate detection realize moving target detection and obtain the position of moving target in RD domain;

S2. Use the multi-target tracking algorithm to realize the association of moving targets under different sub-apertures, and obtain the trajectory of the moving target in the RD domain;

S3. Project the RD domain trajectory of the moving target to the road grid, and use a dynamic programming algorithm (ie Dynamic Programming, DP algorithm) to realize the reconstruction of the moving target trajectory.

In the present invention, the implementation method of S1 is as follows:

It is known that the center frequency of the CSAR transmit signal is f _c , and the bandwidth is B. Assuming that the origin of the Cartesian coordinate system is the center of the detection scene, the moving speed and position of the moving target are respectively and where t _a represents the slow time. radar platform at speed Curve movement around the detection scene, the position coordinates of the radar platform at the slow time t _a is

Assuming that the radar transmission signal is a linear frequency modulation (LFM) signal, the received CSAR moving target echo signal after quadrature demodulation and range pulse pressure is expressed as:

s(r,t _a )=sinc{πB[rR(t _a )]}exp[-j4πf _c R(t _a )/c] (1)

Among them, r is the slant distance, c is the speed of light, sinc(·) is the Singh function; R(t _a ) is the two-way distance slant distance from the radar platform to the moving target, that is:

where ||·|| ₂ represents the 2-norm.

In order to realize the trajectory reconstruction of the moving target, the radar full-aperture echo is evenly divided into K sub-aperture echoes, and the kth sub-aperture echo is

Among them, T _sub and t _a,k represent the duration and central moment of the kth sub-aperture respectively; rect(·) represents the rectangular window function.

In formula (3), the Taylor expansion expression of R(t _a ) at t _a =t _a,k is

Where r _k =R(t _ak ) is a constant term, representing the distance between the target and the radar at the center of the kth sub-aperture;

is the coefficient of the first-order term, and φ ₁ (t _a ,t _a,k ) represents the higher-order term, representing Residual Range Cell Migration (RRCM).

The first-order term in formula (4) can be corrected by keystone transformation; after correcting the first-order term by keystone transformation, the kth sub-aperture echo is transformed into the range Doppler domain to obtain

in represents the Fourier transform, λ _c =c/f _c , f _a represents the Doppler frequency, Represents a two-dimensional convolution. Indicates the instantaneous Doppler frequency of the moving target at t _a =t _a,k ; walking the RRCM in the residual range unit will cause the moving target to defocus in the RD domain, and the degree of defocus is represented by Ψ(r,f _a ).

After obtaining sS _k (r, f _a ), use the weighted average method to obtain the position measurement value of the moving target in the RD domain, and the measurement value is represented by (r _k , f _{a, k} ).

Generally speaking, the sub-aperture echo contains multiple moving targets, and the measurement result of the kth sub-aperture echo is expressed by the following formula

where N _k represents the number of moving targets in the kth sub-aperture echo, and are two sets whose elements and Represent the distance and Doppler measurement value of the mth moving target in the kth sub-aperture, respectively.

After obtaining the moving target detection results of multiple sub-apertures, it is necessary to correlate the same moving target among different sub-apertures. Since the association process is carried out in the RD domain, it is called the moving target trajectory tracking in the RD domain. The realization method of its S2 of the present invention is as follows:

Assuming that the kth sub-aperture echo measurement results need to be processed currently, the nth moving target trajectory obtained from the first k-1 sub-aperture echo measurement results can be expressed as

Among them, N represents the number of moving targets, i _n represents that the nth moving target is detected for the first time in the i _nth sub-aperture, and L _n represents the track length of the nth moving target. gather and Represents the trajectory of the moving target, and its elements and represent the distance and Doppler measurement value of the nth moving target in the kth sub-aperture, respectively.

Through the nearest neighbor search, the RD domain position of the nth moving target under the kth subaperture can be obtained, and the nearest neighbor search expression is

dis _min indicates the minimum distance, and m ₀ indicates the value of m when the minimum distance dis _min is reached. If dis _min is less than a given threshold T _dis , the and respectively added to the collection and , and update the corresponding trajectory length L _n =L _n +1, and update the set Otherwise, it means that the nth moving target does not appear in the kth sub-aperture, the reason may be that a false alarm occurs or the moving target leaves the observation scene, this phenomenon is called tracking loss; perform the above nearest neighbor search for each moving target , if the final collection and If there are still elements remaining in , the remaining elements will be used as the starting position of the new moving target trajectory, and the number of moving targets N=N+N _r will be updated at the same time, where N _r represents the number of remaining elements; if a moving target is in two consecutive sub-apertures If the track is lost, its trajectory will no longer be updated. If the moving target loses track in a certain sub-aperture, the position of the adjacent sub-aperture is estimated by interpolation.

In S3 of the present invention, the trajectory of the moving target in the RD domain is obtained via S2 After that, the RD domain trajectory needs to be mapped to the road grid with the help of road prior information; road prior information can be obtained from satellite maps or automatically extracted from CSAR imaging results.

Taking the nth moving target as an example, the trajectory of the RD domain After mapping to the road grid, L _n sets can be obtained, and the mth set can be expressed as

in Indicates the coordinates of the center of the road grid, Indicates the radar position.

The number of elements in the set S _m is inversely proportional to the size of the road grid, and the elements in the set S _m represent all possible positions of the moving target at the slow time t _{a, k} . In order to select the optimal position from them, consider the following optimization problem

in

Indicates the speed of the aircraft at slow time t _a,k , and T indicates the position collection of

Obviously, the optimal solution That is, the trajectory reconstruction result of the nth moving target; the optimization problem can be solved by using the DP algorithm.

Perform the above operations on all RD domain trajectories of moving objects, and then the reconstruction results of all moving object trajectories can be obtained.

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

1) The present invention adopts the CSAR mode, which has the ability to observe moving targets for a long time;

2) By adding road prior information and using dynamic programming to solve the optimization equation, the present invention can improve the reconstruction accuracy of the target trajectory.

Description of drawings

FIG. 1 is a schematic diagram of an application scenario of the present invention;

Fig. 2 is a flow chart of the present invention;

Figure 3 is a schematic diagram of the simulated scene optical image and the CSAR imaging results of the measured data;

Fig. 4 is a schematic diagram of detection results of a sub-aperture moving target.

Fig. 5 is a schematic diagram of the track tracking result of a single moving target in the RD domain;

Fig. 6 is a schematic diagram of road grid extraction results;

Fig. 7 is a schematic diagram of the DP algorithm.

Fig. 8 is a schematic diagram of the reconstruction result of the moving target trajectory.

Detailed ways

In order to make the object, technical solution and advantages of the present invention clearer, the implementation manner of the present invention will be further described in detail below in conjunction with the accompanying drawings.

Circular motion is a special form of curved motion, which has the largest observation angle and can observe the scene for a long time. The technical solution of the present invention will be described below by taking circular motion as an example.

Fig. 1 is a schematic diagram of the application scene of the present invention. In the figure, the radar platform flies along a circle, and the beam always points to the detection area to observe the moving target in the detection area for a long time.

Fig. 2 is a flowchart of the present invention. A method for reconstructing the trajectory of a ground moving target suitable for CSAR imaging, comprising the following steps:

S1. Perform range pulse pressure on the received CSAR moving target echo signal, evenly divide the full aperture echo of the radar into several sub-aperture echoes, and transform the sub-aperture echoes into the range Doppler domain, through Doppler Le filter and constant false alarm rate detection are used to detect the moving target and obtain the position of the moving target in the RD domain.

It is known that the center frequency of the CSAR transmit signal is f _c , and the bandwidth is B. Assuming that the origin of the Cartesian coordinate system is the center of the detection scene, the moving speed and position of the moving target are respectively and where t _a represents the slow time. radar platform at speed Curve movement around the detection scene, the position coordinates of the radar platform at the slow time t _a is

Assuming that the radar transmission signal is a linear frequency modulation (LFM) signal, the received CSAR moving target echo signal after quadrature demodulation and range pulse pressure is expressed as:

s(r,t _a )=sinc{πB[rR(t _a )]}exp[-j4πf _c R(t _a )/c] (1)

Among them, r is the slant distance, c is the speed of light, sinc(·) is the Singh function; R(t _a ) is the two-way distance slant distance from the radar platform to the moving target, that is:

where ||·|| ₂ represents the 2-norm.

In order to realize the trajectory reconstruction of the moving target, the radar full-aperture echo is evenly divided into K sub-aperture echoes, and the kth sub-aperture echo is

Among them, T _sub and t _a,k represent the duration and central moment of the kth sub-aperture respectively; rect(·) represents the rectangular window function.

In formula (3), the Taylor expansion expression of R(t _a ) at t _a =t _a,k is

Where r _k =R(t _a,k ) is a constant term, representing the distance between the target and the radar at the kth sub-aperture center moment;

is the coefficient of the first-order term, and φ ₁ (t _a ,t _a,k ) represents the higher-order term, representing Residual Range Cell Migration (RRCM).

The first-order term in formula (4) can be corrected by keystone transformation; after correcting the first-order term by keystone transformation, the kth sub-aperture echo is transformed into the range Doppler domain to obtain

in represents the Fourier transform, λ _c =c/f _c , f _a represents the Doppler frequency, Represents a two-dimensional convolution. Indicates the instantaneous Doppler frequency of the moving target at t _a =t _a,k ; walking the RRCM in the residual range unit will cause the moving target to defocus in the RD domain, and the degree of defocus is represented by Ψ(r,f _a ).

After obtaining sS _k (r, f _a ), use the weighted average method to obtain the position measurement value of the moving target in the RD domain, and the measurement value is represented by (r _k , f _{a, k} ).

Generally speaking, the sub-aperture echo contains multiple moving targets, and the measurement result of the kth sub-aperture echo is expressed by the following formula

where N _k represents the number of moving targets in the kth sub-aperture echo, and are two sets whose elements and Represent the distance and Doppler measurement value of the mth moving target in the kth sub-aperture, respectively.

S2. Use the multi-target tracking algorithm to realize the association of moving targets under different sub-apertures, and obtain the trajectory of the moving target in the RD domain.

After obtaining the moving target detection results of multiple sub-apertures, it is necessary to correlate the same moving target among different sub-apertures. Since the association process is carried out in the RD domain, it is called the moving target trajectory tracking in the RD domain.

Assuming that the kth sub-aperture echo measurement results need to be processed currently, the nth moving target trajectory obtained from the first k-1 sub-aperture echo measurement results can be expressed as

Among them, N represents the number of moving targets, i _n represents that the nth moving target is detected for the first time in the i _nth sub-aperture, and L _n represents the track length of the nth moving target. gather and Represents the trajectory of the moving target, and its elements and represent the distance and Doppler measurement value of the nth moving target in the kth sub-aperture, respectively.

Through the nearest neighbor search, the RD domain position of the nth moving target under the kth subaperture can be obtained, and the nearest neighbor search expression is

dis _min indicates the minimum distance, and m ₀ indicates the value of m when the minimum distance dis _min is reached. If dis _min is less than a given threshold T _dis , the and respectively added to the collection and , and update the corresponding trajectory length L _n =L _n +1, and update the set Otherwise, it means that the nth moving target does not appear in the kth sub-aperture, the reason may be that a false alarm occurs or the moving target leaves the observation scene, this phenomenon is called tracking loss; perform the above nearest neighbor search for each moving target , if the final collection and If there are still elements remaining in , the remaining elements will be used as the starting position of the new moving target trajectory, and the number of moving targets N=N+N _r will be updated at the same time, where N _r represents the number of remaining elements; if a moving target is in two consecutive sub-apertures If the track is lost, its trajectory will no longer be updated. If the moving target loses track in a certain sub-aperture, the position of the adjacent sub-aperture is estimated by interpolation.

S3. Project the RD domain trajectory of the moving target to the road grid, and use a dynamic programming algorithm (ie Dynamic Programming, DP algorithm) to realize the reconstruction of the moving target trajectory.

Obtain the trajectory of the moving target in the RD domain via S2 After that, the RD domain trajectory needs to be mapped to the road grid with the help of road prior information; road prior information can be obtained from satellite maps or automatically extracted from CSAR imaging results.

Taking the nth moving target as an example, the trajectory of the RD domain After mapping to the road grid, L _n sets can be obtained, and the mth set can be expressed as

in Indicates the coordinates of the center of the road grid, Indicates the radar position.

The number of elements in the set S _m is inversely proportional to the size of the road grid, and the elements in the set S _m represent all possible positions of the moving target at the slow time t _{a, k} . In order to select the optimal position from them, consider the following optimization problem

in

Indicates the speed of the aircraft at slow time t _a,k , and T indicates the position collection of

Obviously, the optimal solution That is, the trajectory reconstruction result of the nth moving target; the optimization problem can be solved by using the DP algorithm. Perform the above operations on all RD domain trajectories of moving objects, and then the reconstruction results of all moving object trajectories can be obtained.

The method of the invention is verified through half-physical simulation, that is, the static target echo is obtained from the radar image simulation, and the preset moving target echo is added at the same time. The moving target is moving at a speed of 5m/s, traveling from southwest to northeast along the road. Experimental results prove the effectiveness of the present invention.

In the experiment, the system parameters in the present invention are shown in Table 1 below.

Table 1

The optical image and CSAR imaging results of the detection scene used in the simulation are shown in Figure 3. There is a roundabout intersection in the scene, which is located near the entrance of the expressway, and the scene contains many moving targets mainly vehicles. The size of the observation scene is 300m×300m (distance×azimuth).

Fig. 4 is a schematic diagram of detection results of a sub-aperture moving target. The horizontal direction is the range sampling point, and the vertical direction is the Doppler sampling point. From the left figure of Figure 4, it can be seen that after Doppler filtering, the stationary target clutter is suppressed, and then through CFAR detection, the corresponding binary image is obtained, as shown in the right figure of Figure 4. Through weighted average processing, the position of the moving target in the RD domain under the current sub-aperture can be obtained, that is, the set and

Figure 5 is a schematic diagram of the tracking results of a single moving target trajectory in the RD domain; it can be seen that the obtained moving target trajectory is relatively smooth, which is consistent with the characteristics of the moving target trajectory in the RD domain. The side proves the correctness of the RD domain target tracking algorithm and the correctness of the moving target detection results.

Figure 6 is a schematic diagram of the road grid extraction results; the road grid is obtained from prior information, the denser the road grid division, the higher the reconstruction accuracy of the subsequent moving target trajectory, and the greater the calculation amount of the corresponding DP algorithm. In actual operation, a compromise can be made according to application requirements.

Fig. 7 is a schematic diagram of the DP algorithm. The DP algorithm contains L _n decision-making stages, and the state of each stage is determined by the road grid position after projection. path weight The corresponding minimum path corresponds to the reconstruction result of the moving target trajectory.

Fig. 8 is a schematic diagram of the reconstruction result of the moving target trajectory. The white cross symbol represents the real trajectory of the target (simulation preset trajectory), and the white solid box represents the estimated trajectory, that is, the trajectory reconstruction result. It can be seen that the reconstruction result of the trajectory of the moving target has high precision and a high degree of coincidence with the simulation preset trajectory.

The above descriptions are only preferred implementations of the present invention, and the scope of protection of the present invention is not limited to the above examples, and all technical solutions that fall under the idea of the present invention belong to the scope of protection of the present invention. It should be pointed out that for those skilled in the art, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims (4)

1. A kind of 1. ground moving target trajectory reconstruction method suitable for CSAR imagings, which is characterized in that include the following steps：

   S1. to the CSAR transient echos signal that receives into row distance pulse pressure, by the full aperture echo of radar be evenly dividing for Several sub-aperture echoes, and sub-aperture echo is transformed into range-Dopler domain, it is examined by doppler filtering and constant false alarm rate It surveys and realizes that moving-target detects and obtains moving-target in the position in RD domains；

   S2. it realizes that moving-target is associated under different sub-apertures using multiple target tracking algorithm, obtains movement rail of the moving-target in RD domains Mark；

   S3. moving-target RD domains track is projected into road grid, and realize moving-target trajectory reconstruction using dynamic programming algorithm.
2. 2. the ground moving target trajectory reconstruction method according to claim 1 suitable for CSAR imagings, which is characterized in that S1 Implementation method it is as follows：

   Known CSAR transmittings signal center frequency is f_c, bandwidth B；Assuming that cartesian coordinate system origin is detection scene center, move Target speed and position are respectivelyWithWherein t_aRepresent the slow time；Radar Platform is with speedCurvilinear motion, slow time t are around detection scene_aThe position coordinates of the radar platform at moment areIf radar emission signal is linear FM signal, then the CSAR transient echos signal received passes through Quadrature demodulation and after pulse pressure, is expressed as：

   s(r,t_a)=sinc { π B [r-R (t_a)]}exp[-j4πf_cR(t_a)/c] (1)

   Wherein, r represents oblique distance, and c is the light velocity, and sinc () represents sinc function；R(t_a) it is round trip of the radar platform to moving-target Apart from oblique distance, i.e.,：

   Wherein | | | |₂Represent 2 norms；

   In order to realize moving-target trajectory reconstruction, radar full aperture echo is divided evenly as K sub-aperture echo, k-th of sub-aperture Diameter echo is

   Wherein T_subAnd t_a,kDuration and the central instant of k-th sub-aperture are represented respectively；Rect () represents rectangular window letter Number；

   R (t in formula (3)_a) in t_a=t_a,kLocating Taylor expansion expression formula is

   Wherein r_k=R (t_a,k) it is constant term, represent the distance of target and radar in k-th of sub-aperture central instant；

   For single order term coefficient, φ₁(t_a,t_a,k) represent higher order term, represent remaining Range cell migration RRCM；

   Single order item in formula (4) can be converted by keystone to be corrected；It, will after converting correction single order item by keystone K-th of sub-aperture echo transforms to range-Dopler domain and obtains

   WhereinRepresent Fourier transformation, λ_c=c/f_c, f_aRepresent Doppler frequency,Represent two-dimensional convolution,Represent moving-target in t_a=t_a,kInstantaneous Doppler frequency；Remaining Range cell migration RRCM can lead to dynamic mesh It is marked on RD domains to defocus, defocusing degree Ψ (r, f_a) represent；

   Obtain sS_k(r,f_a) after, using average weighted method, moving-target is obtained in the position measuring value in RD domains, the measuring value With (r_k,f_a,k) represent；

   Comprising multiple moving-targets in sub-aperture echo, k-th of sub-aperture echo measurement is represented with following formula

   Wherein N_kRepresent moving-target number in k-th of sub-aperture echo,WithFor two set, elementWithPoint The distance and Doppler measurement of m-th of moving-target in k-th of sub-aperture are not represented.
3. 3. the ground moving target trajectory reconstruction method according to claim 1 suitable for CSAR imagings, which is characterized in that S2 Implementation method it is as follows：

   It is measured assuming that need to currently handle k-th sub-aperture echo as a result, surveying the of result acquisition by preceding k-1 sub-aperture echo volume N moving-target track is represented by

   Wherein N represents moving-target number, i_nRepresent n-th of moving-target i-th_nA sub-aperture is detected for the first time, L_nRepresent n-th A moving-target path length；SetWithRepresent moving-target track, elementWithRepresent that n-th of moving-target exists respectively Distance and Doppler measurement in k-th of sub-aperture；

   By nearest neighbor search, RD domain position of n-th of moving-target under k-th of sub-aperture, nearest neighbor search table can be obtained It is up to formula

   dis_minRepresent minimum range, m₀Expression reaches minimum range dis_minWhen m value；If dis_minLess than one given threshold Value T_dis, then willWithIt is respectively added to gatherWithIn, and update corresponding path length L_n=L_n+ 1 and Update setOtherwise mean that n-th of moving-target does not appear in k-th of sub-aperture In, reason may be that false dismissal either moving-target occurs to be driven out to observation scene, this phenomenon is known as with losing；To each moving-target Above-mentioned nearest neighbor search is performed, if last setWithIn still have element remaining, then using surplus element as new moving-target Track initial position, while update moving-target number N=N+N_r, N_rRepresent surplus element number；If a moving-target is continuous two Occur in a sub-aperture with losing phenomenon, then no longer update its track.If moving-target occurs in a certain sub-aperture with losing phenomenon, profit Pass through its position of Interpolate estimation with adjacent sub-aperture position.
4. 4. the ground moving target trajectory reconstruction method according to claim 3 suitable for CSAR imagings, which is characterized in that S3 In, moving-target is obtained in the track in RD domains through S2Later, it needs by road prior information, by RD domains trajectory map To road grid；Road prior information can be obtained by satellite map or be automatically extracted by CSAR imaging results；

   By taking n-th of moving-target as an example, by the track in RD domainsIt is mapped to after road grid, L can be obtained_nA set, Wherein m-th set can be expressed as

   WhereinRepresent road grid centre coordinate, Represent radar site；

   Set S_mMiddle element number is inversely proportional with road grid size, set S_mIn element representation moving-target in slow time t_a,kWhen All possible positions carved；In order to therefrom select optimal location, following optimization problem is considered

   Wherein

   Represent aircraft in slow time t_a,kSpeed, T represent positionSet

   Obviously, optimal solutionThe trajectory reconstruction result of i.e. n-th moving-target；Optimization problem can utilize DP algorithm to solve；

   Aforesaid operations are carried out to all moving-target RD domains tracks, you can obtain all moving-target trajectory reconstruction results.