CN113435299A - Bistatic forward-looking SAR clutter suppression method based on space-time matching - Google Patents

Abstract

The invention discloses a dual-base forward-looking SAR clutter suppression method based on space-time matching, comprising the following steps: S1, establishing a BFSAR spatial geometric model, and initializing system parameters; S2, performing range pulse compression on echo signals , and perform preprocessing and migration correction on the echo signal after range pulse compression; S3, establish a space-time clutter model according to the BFSAR spatial geometric model, and obtain the space-time frequency information of the clutter of the unit to be detected; S4, treat The optimal matching space-time filter of the detection unit is designed to obtain the constrained optimization problem; S5, the particle swarm optimization algorithm is used to solve the constrained optimization problem, and the optimal solution is obtained; S6, the optimal matching space-time filter is reconstructed according to the optimal solution . The invention effectively avoids the clutter covariance matrix estimation in the traditional STAP algorithm, eliminates the influence of the non-stationarity of the bi-base forward looking SAR clutter, and can establish an optimally matched space-time filter in any configuration. Suppression of strong non-stationary clutter in bistatic forward looking SAR.

Description

A clutter suppression method for bistatic forward looking SAR based on space-time matching

technical field

The invention belongs to the technical field of radar, in particular to a dual-base forward-looking SAR clutter suppression method based on space-time matching.

Background technique

Bistatic forward-looking synthetic aperture radar (BFSAR) can break through the inherent limitations of traditional single-base SAR by placing the receiving and transmitting stations on different independent platforms, and obtain a two-dimensional view directly in front of the platform. High-resolution images. With the continuous improvement of the requirements for remote sensing systems in modern applications, the demand for BFSAR ground moving target detection technology in the military and civilian fields is increasingly urgent. However, when BFSAR detects the ground, the moving target echo will usually be submerged in the strong clutter background of stationary ground. Therefore, clutter suppression is one of the key steps in moving target detection.

In single-base front-looking SAR, the clutter echo is range-independent, that is, the clutter angle-Doppler trace has the same characteristics in different range units, but in the case of bi-base forward-looking SAR, the ground clutter The wave has strong non-stationarity, and the angle-Doppler trace of the clutter changes with the change of the distance unit, and there is a distance correlation, which brings great difficulty to the effective suppression of the clutter in the bistatic forward looking SAR.

At present, the research and literature of bistatic forward-looking SAR mainly focus on the imaging algorithm of stationary scene, see the literature "R. Wang, O. Loffeld, Y. Neo, et al. .IEEE Transactions on Geoscience and RemoteSensing, 2010, 48(1):452-465" and "H.Shin and J.Lim.Omega-k Algorithm for AirborneForward-Looking Bistatic Spotlight SAR Imaging[J].IEEE Geoscience and RemoteSensing Letter , 2009, 6(2):312-316”. For the imaging of moving targets with bistatic forward looking SAR, there have also been related research published in recent years. See document "Z.Li, J.Wu, Y.Huang, Z.Sun and J.Yang.Ground-Moving TargetImaging and Velocity Estimation Based on Mismatched Compression for BistaticForward-Looking SAR[J].IEEE Transactions on Geoscience and Remote Sensing , 2016, 54(6):3277-3291" and "Z.Li, J.Wu, Z.Liu, Y.Huang,H.Yang and J.Yang.An Optimal2-D Spectrum Matching Method for SAR Ground Moving Target Imaging. IEEE Transactions on Geoscience and Remote Sensing, 2018, 56(10):5961-5974”. These two methods can realize the refocusing and parameter estimation of moving targets, but they do not consider the influence of BFSAR ground clutter in the processing of moving target signals. When BFSAR clutter exists, the above methods will face severe performance loss. In order to effectively realize the BFSAR moving target indication, it is necessary to suppress the clutter in the echo first. The main clutter suppression methods include the phase center offset antenna (DPCA) method and the space-time adaptive processing (STAP) method. See references "D. Cerutti-Maori and I. Sikaneta. A Generalization of DPCA Processing for Multichannel SAR/GMTIRadars. IEEE Transactions on Geoscience and Remote Sensing, 2013, 51(1):560-572." and "Ender, J.H G. Space-time processing for multichannel synthetic apertureradar [J]. IEEE Electronics and Communication Engineering Journal, 2002, 11(1): 29-38”. The DPCA method achieves clutter suppression by canceling multi-channel echoes, but it requires the speed, channel spacing and pulse repetition frequency of the SAR system to meet strict conditions, so that the equivalent phase centers of different channels can overlap in different time domains ; However, in BFSAR, the separation of transceiver will make it difficult to meet the above conditions, and then affect the processing effect of BFSAR-DPCA. As an extension of DPCA, the STAP method extends the one-dimensional signal processing to the space-time two-dimensional domain for processing; however, due to the range migration, Doppler spectrum spread and non-stationary characteristics of the clutter echo of BFSAR, it will lead to this problem. The clutter suppression performance of the method is severely deteriorated.

SUMMARY OF THE INVENTION

The purpose of the present invention is to overcome the deficiencies of the prior art, and to provide a processing strategy for directly designing and constructing a space-time filter matched to the clutter spectrum, avoiding the clutter covariance matrix estimation in the traditional STAP algorithm, and effectively The influence of clutter non-stationarity of bistatic forward-looking SAR is eliminated, and an optimally matched space-time filter can be established in any configuration to suppress non-stationary clutter in bistatic forward-looking SAR.

The object of the present invention is to be achieved through the following technical solutions: a dual-base forward-looking SAR clutter suppression method based on space-time matching, comprising the following steps:

S1. Establish a BFSAR spatial geometric model and initialize system parameters;

S2, perform range pulse compression on the echo signal, and perform preprocessing and migration correction on the echo signal after range pulse compression;

S3, establishing a space-time clutter model according to the BFSAR spatial geometric model, obtaining the space-time frequency information of the clutter of the unit to be detected, and obtaining the space-time distribution information of the clutter spectrum according to the coupling relationship of the space-time frequency of the clutter;

S4. According to the clutter space-time distribution information, the optimal matching space-time filter of the unit to be detected is designed to obtain a constrained optimization problem;

S5. Use the particle swarm optimization algorithm to solve the constrained optimization problem, and obtain the optimal solution of the constrained optimization problem;

S6. According to the optimal solution obtained by the particle swarm optimization algorithm, reconstruct the optimally matched space-time filter, and use the reconstructed optimally matched space-time filter to filter the unit to be detected to obtain non-stationary clutter suppression signal after.

Further, the specific implementation method of S2 is: for the BFSAR system, set the transmitted signal to be a linear frequency modulation signal, use a filtering method to preprocess the echo signal, and perform migration correction on the echo signal through trapezoidal distortion correction;

The filter H _pre (t,f _τ ) in preprocessing is:

Among them, f _ref is the Doppler centroid of the reference point, f _τ and f _c are the range frequency and the carrier frequency, respectively;

The keystone correction function is expressed as:

t=f _c t ₁ /(f _τ +f _c )

Among them, t ₁ is the new azimuth time after transformation;

After the above processing, the signal received by the nth channel is expressed as:

Among them, P represents the clutter scattering point or moving target in the observation scene, σ(P) is the backscattering coefficient of P, ω _a ( ) represents the azimuth envelope; parameters τ, t, B _τ , λ, c and t _P represent distance time, azimuth time, distance bandwidth, wavelength, speed of light and beam center time, respectively; R _s (0,n; P) is the dual base distance of point P relative to the nth channel when the azimuth time is 0; R _s (t ₁ ,n; P) is the bi-base distance of point P relative to the nth receiving channel;

Using time-sharing processing, the processing time is divided into several sub-time segments. In time-sharing processing, each sub-time segment satisfies the following relationship:

Among them, ΔT is the length of the time-sharing processing sub-segment, Ka is the Doppler modulation frequency, _{Δδ a is} the Doppler resolution, and T _syn is the synthetic aperture time;

After performing migration correction and time-division processing on the echo signal, column vectorization is performed on the data in each sub-period to obtain the space-time sample data in each sub-period, which is denoted as S _vec (t).

Further, the specific implementation method of S3 is: the space-time frequency information of the clutter of the unit to be detected is as follows:

分别是地面杂波散射点相对于接收机的方位角和俯仰角，θ_p表示接收机阵列与飞行方向的夹角，d和λ分别表示通道间距和信号波长；where f _d and f _s are the normalized Doppler frequency and normalized spatial frequency of the ground clutter scattering point, respectively, f _r is the pulse repetition frequency, and V _T and _VR are the flight of the transmitter and receiver, respectively Velocities, ψ _T and ψ _R are the spatial cone angles of the ground clutter scattering point relative to the transmitter and receiver, respectively, θ _R and

are the azimuth angle and pitch angle of the ground clutter scattering point relative to the receiver, θ _p represents the angle between the receiver array and the flight direction, and d and λ represent the channel spacing and signal wavelength, respectively;

The space-time distribution of the clutter spectrum in the unit to be detected is expressed as:

分别是杂波散射点相对于发射机的方位角和俯仰角，δ_T和δ_R分别表示发射机和接收机飞行方向相对于双基基线的夹角。θ _T and

are the azimuth and pitch angles of the clutter scattering point relative to the transmitter, respectively, and δT and _{δR represent} the included angles of the transmitter and receiver flight directions relative to the dual-base baseline, respectively.

Further, the specific implementation method of S4 is: for a system with N number of channels and K number of pulses, let the designed filter weight coefficient be W _d , and W _d is a NK×1-dimensional complex vector, expressed as:

W _d =[W _d1 ,W _d2 ,...,W _dnk ,...,W _dNK ] ^T

Among them, W _dnk is the weight coefficient component of the nkth dimension;

The two-dimensional space-time frequency response of this filter is expressed as:

为空-时导向矢量；S_t∈K×1和S_s∈N×1分别为时间与空间导向矢量，分别表示为：in,

is the space-time steering vector; S _t ∈ K×1 and S _s ∈ N×1 are the time and space steering vectors, respectively, expressed as:

其中，V_t∈K×1和V_s∈N×1分别为运动目标的时间与空间导向矢量；The response of this filter to a moving target is expressed as

Among them, V _t ∈ K×1 and V _s ∈ N×1 are the temporal and spatial steering vectors of the moving target, respectively;

The receiving beam is evenly divided into Q sub-beams, and Q is determined by the synthetic aperture length L _syn and the azimuth resolution ρ _a , and is expressed as Q=L _syn /ρ _a ; after the beam division, it is considered that the clutter signal in the receiver is It is obtained by the superposition of the clutter echoes in the Q directions; therefore, in order to achieve effective suppression of the non-stationary clutter in the bistatic forward looking SAR, it is only necessary that the space-time frequency response of the filter in the direction of the Q sub-beams is zero; according to Based on the above ideas, the following constrained optimization problem is established:

Among them, ε is the error tolerance of observation noise; {f _u (W _d )∣u=1,2,3} is the objective function, which is expressed as:

Among them, f _di and f _si are the normalized Doppler frequency and normalized spatial frequency in the ith sub-beam direction; the objective functions f ₁ (W _d ) and f ₂ (W _d ) represent space-time filtering, respectively is the average and variance of the notches set by the filter in the Q sub-beam directions; the objective function f ₃ (W _d ) will determine the width of the space-time filter notch, and f ₃ (W _d ) consists of two parts: sumR _L ( W _d ) and sumR _R (W _d ), sumR _L (W _d ) and sumR _R (W _d ) are the sums of the two-dimensional frequency responses on the left and right sides of the filter notch, respectively, expressed as:

where Δf _d is a constant Doppler frequency.

Further, the specific implementation method of S5 is:

S51, initialize the population number X and the number of iterations G, and set the boundary of the decision variable in the particle swarm optimization algorithm;

S52. Split the filter weight coefficient W _di into a real part and an imaginary part:

W _di =x _i +jx _l , i=1,2,...,NK

Among them, the serial number index l=i+NK, the decision vector x=[x ₁ ,x ₂ ,x ₃ ,...,x _2NK ] ^T is a particle in the particle swarm algorithm, that is, an optimal solution, and the optimal solution adopts the particle swarm The algorithm is solved, and the optimal solution obtained is recorded as

重构最优匹配空-时滤波器权系数

如下，Further, the specific implementation method of S6 is: according to the optimal solution obtained by the particle swarm optimization algorithm

Reconstructing the optimal matching space-time filter weights

as follows,

For the unit to be detected, use the reconstructed optimal matching space-time filter to perform space-time filtering, and obtain the target signal after the non-stationary clutter suppression of the dual-base forward looking SAR.

The beneficial effects of the present invention are as follows: the present invention adopts the processing strategy of directly designing and constructing the space-time filter matched to the clutter spectrum, effectively solving the deterioration of the suppression performance caused by the non-stationary clutter of the bistatic forward looking SAR. question. The invention first obtains the space-time characteristics of clutter through BFSAR clutter modeling, and designs a clutter suppression filter according to the obtained clutter information. Then, the problem of solving the filter weight coefficients is transformed into a constrained optimization problem. Finally, using the particle swarm optimization algorithm, the optimal matching space-time filter of the unit to be detected is directly solved and constructed, and the non-stationary clutter suppression of the dual-base forward looking SAR is realized by the space-time filter. The innovation of the invention is that the estimation of the clutter covariance matrix in the traditional STAP algorithm is effectively avoided, the influence of the clutter non-stationarity of the bistatic forward looking SAR is effectively eliminated, and the optimal matching space-time can be established under any configuration. The filter realizes the suppression of strong non-stationary clutter in bistatic forward looking SAR.

Description of drawings

Fig. 1 is the flow chart of the bistatic forward looking SAR non-stationary clutter suppression method of the present invention;

Fig. 2 is the schematic diagram of the BFSAR space geometric model of the present embodiment;

FIG. 3 is an echo signal diagram of the present embodiment;

FIG. 4 is an echo domain signal after non-stationary clutter suppression in this embodiment;

Fig. 5 is the image domain signal after non-stationary clutter suppression in this embodiment;

Fig. 6 is the cross-sectional comparison along the X-axis before and after the treatment of the present embodiment;

FIG. 7 is a cross-sectional comparison along the Y-axis before and after treatment in this embodiment.

Detailed ways

The present invention mainly adopts the method of simulation experiment for verification, and the simulation verification platform is MATLAB2020a. The technical solutions of the present invention are further described below with reference to the accompanying drawings and specific embodiments.

As shown in FIG. 1 , a method for suppressing clutter based on space-time matching for dual-base forward looking SAR of the present invention includes the following steps:

S1, establish a BFSAR spatial geometric model, and initialize the system parameters; the BFSAR geometric configuration used in this example is shown in Figure 2, and the system parameters of BFSAR are shown in Table 1, wherein, the position coordinates of the transmitter at zero time are ( x _T , y _T , z _T ), the transmitter flies along the Y axis at the speed of V _T ; the position coordinate of the nth channel of the receiver at time zero is (x _R , y _R +(n-1)d, z _R ), the receiver flies along the Y axis at the speed of _VR ; the speed of light is c.

Table 1

S2, perform range pulse compression on the echo signal, and perform preprocessing and migration correction on the echo signal after range pulse compression;

The specific implementation method is as follows: for the BFSAR system, the transmitted signal is a linear frequency modulation signal, the echo signal is calculated and the echo signal is compressed by distance pulse; in order to eliminate the influence of the echo cross-range unit on the clutter suppression processing, the present invention adopts a filtering method The echo signal is preprocessed to eliminate Doppler centroid ambiguity; the echo signal will contain the coupling terms of range frequency and azimuth time after range-to-pulse compression, and the coupling is removed by keystone correction to realize echo range migration correction.

The filter H _pre (t,f _τ ) in preprocessing is:

Among them, f _ref is the Doppler centroid of the reference point, f _τ and f _c are the range frequency and the carrier frequency, respectively;

The keystone correction function is expressed as:

t=f _c t ₁ /(f _τ +f _c )

Among them, t ₁ is the new azimuth time after transformation;

After the above processing, the signal received by the nth channel is expressed as:

Among them, P represents the clutter scattering point or moving target in the observation scene, σ(P) is the backscattering coefficient of P, ω _a ( ) represents the azimuth envelope; parameters τ, t, B _τ , λ, c and t _P represent the distance time, azimuth time, distance bandwidth, wavelength, light speed and beam center time respectively; R _s (0,n; P) is the dual base distance of point P relative to the nth channel when the azimuth time is 0, That is, the sum of the distances from P to the transmitting station and the receiving station (channel n) at time 0; R _s (t ₁ ,n; P) is the dual-base distance of point P relative to the nth receiving channel;

Considering that the long observation time in BFSAR results in Doppler broadening, the traditional STAP method cannot be directly applied, the present invention adopts the means of time-division processing to divide the processing time into several sub-time segments, thereby eliminating the influence of Doppler broadening. In time-sharing processing, each sub-time segment satisfies the following relationship:

Among them, ΔT is the length of the time-sharing processing sub-segment, Ka is the Doppler modulation frequency, _{Δδ a is} the Doppler resolution, and T _syn is the synthetic aperture time;

After performing migration correction and time-sharing processing on the echo signal, column vectorization is performed on the data in each sub-period to obtain the space-time sample data in each sub-period, which is expressed as S _vec (t). In this embodiment, after S2 The resulting signal after processing is shown in Figure 3.

S3. In order to realize the design of the optimal matching space-time filter, the present invention first establishes a space-time clutter model according to the BFSAR space geometric model, obtains the space-time frequency information of the clutter of the unit to be detected, and obtains the space-time frequency information of the clutter of the unit to be detected. The coupling relationship of frequency, obtain the space-time distribution information of clutter spectrum;

The specific implementation method is as follows: the space-time frequency information of the clutter of the unit to be detected is as follows:

分别是地面杂波散射点相对于接收机的方位角和俯仰角，θ_p表示接收机阵列与飞行方向的夹角，d和λ分别表示通道间距和信号波长。where f _d and f _s are the normalized Doppler frequency and normalized spatial frequency of the ground clutter scattering point, respectively, f _r is the pulse repetition frequency, and V _T and _VR are the flight of the transmitter and receiver, respectively Velocities, ψ _T and ψ _R are the spatial cone angles of the ground clutter scattering point relative to the transmitter and receiver, respectively, θ _R and

are the azimuth and pitch angles of the ground clutter scattering point relative to the receiver, respectively, θ _p represents the included angle between the receiver array and the flight direction, and d and λ represent the channel spacing and signal wavelength, respectively.

The spatial cone angles _ψT and _ψR are obtained by,

分别是杂波散射点相对于发射机的方位角和俯仰角，δ_T和δ_R分别表示发射机和接收机飞行方向相对于双基基线的夹角。因此，待检测单元内杂波谱的空-时分布表示为：where _θT and

are the azimuth and pitch angles of the clutter scattering point relative to the transmitter, respectively, and δT and _{δR represent} the included angles of the transmitter and receiver flight directions relative to the dual-base baseline, respectively. Therefore, the space-time distribution of the clutter spectrum in the unit to be detected is expressed as:

相关，将随距离环的变化而变化，双基前视SAR的杂波将具有非平稳特性。此外，杂波的空-时分布信息将由散射点相对于双平台的空间位置决定，以下将对角度θ_R、

θ_T和

进行求解。It can be seen that the space-time distribution and angle of clutter

The correlation will vary with the change of the range loop, and the clutter of the bistatic forward looking SAR will have non-stationary characteristics. In addition, the space-time distribution information of the clutter will be determined by the spatial position of the scattering point relative to the double platform, and the following will discuss the angle θ _R ,

θ _T and

to solve.

For the unit to be detected, the instantaneous bistatic distance history is expressed as:

Expand the above formula, and rewrite the double base distance sum as:

in,

The general ellipse expression is:

ax ² +bxy+cy ² +dx+ey+1=0

Comparing the general expression of the ellipse, the coefficients of the non-standard ellipse corresponding to the equidistant ring are obtained as:

Further, the inclination angle θ of the major axis, the geometric center (x _c , y _c ), the major semi-axis L _ma and the minor semi-axis L _mi of the non-standard ellipse can be obtained as:

According to the geometric parameters obtained above, according to the relationship between the non-standard ellipse and the standard ellipse (rotation and translation), the coordinate information of each point on the non-standard ellipse can be obtained as:

为非标准椭圆对应的标准椭圆上的点坐标，可根据标准椭圆的参数方程获取。in,

is the point coordinates on the standard ellipse corresponding to the non-standard ellipse, which can be obtained according to the parametric equation of the standard ellipse.

为

和

的统一表示)Therefore, according to the obtained point coordinates on the equidistant ring, the spatial relationship between each point and the carrier platform can be obtained as (θ _{R, T} is the unified representation of θ _R and θ _T ,

for

and

unified representation)

L_RP,TP和

表示如下(L_RP,TP为L_RP和L_TP的统一表示，

为

和

的统一表示)where ||·|| represents the 2-norm, and R ₀ and T ₀ represent the projections of the receiver and transmitter on the ground, respectively. The coordinates x _{R, T} are the unified representation of x _R and x _T , y _{R, T} is the unified representation of y _R and y _T , and z _{R, T} is the unified representation of z _R and z _T. The vector in the above formula

L _RP , TP and

Represented as follows (L _{RP, TP} is the unified representation of L _RP and L _TP ,

for

and

unified representation)

S4. According to the clutter space-time distribution information, the optimal matching space-time filter of the unit to be detected is designed to obtain a constrained optimization problem;

The specific implementation method is: for a system with N channels and K pulses, let the designed filter weight coefficient be W _d , and W _d is a complex vector of NK×1 dimension, which is expressed as:

W _d =[W _d1 ,W _d2 ,...,W _dnk ,...,W _dNK ] ^T

Among them, W _dnk is the weight coefficient component of the nkth dimension;

The two-dimensional space-time frequency response of this filter is expressed as:

为空-时导向矢量；S_t∈K×1和S_s∈N×1分别为时间与空间导向矢量，分别表示为：in,

is the space-time steering vector; S _t ∈ K×1 and S _s ∈ N×1 are the time and space steering vectors, respectively, expressed as:

其中，V_t∈K×1和V_s∈N×1分别为运动目标的时间与空间导向矢量；The response of this filter to a moving target is expressed as

Among them, V _t ∈ K×1 and V _s ∈ N×1 are the temporal and spatial steering vectors of the moving target, respectively;

The receiving beam is evenly divided into Q sub-beams, and Q is determined by the synthetic aperture length L _syn and the azimuth resolution ρ _a , and is expressed as Q=L _syn /ρ _a ; after the beam division, it is considered that the clutter signal in the receiver is It is obtained by the superposition of the clutter echoes in the Q directions; therefore, in order to achieve effective suppression of the non-stationary clutter in the bistatic forward looking SAR, it is only necessary that the space-time frequency response of the filter in the direction of the Q sub-beams is zero; according to Based on the above ideas, the following constrained optimization problem is established:

Among them, ε is the error tolerance of observation noise; {f _u (W _d )∣u=1,2,3} is the objective function, which is expressed as:

Among them, f _di and f _si are the normalized Doppler frequency and normalized spatial frequency in the ith sub-beam direction; the objective functions f ₁ (W _d ) and f ₂ (W _d ) represent space-time filtering, respectively The average value and variance of the notch set by the filter in the Q sub-beam directions; when the function values of f ₁ (W _d ) and f ₂ (W _d ) are smaller, the filter will generate deep and deep corresponding positions in the space-time domain. Smooth notches, thus ensuring that the filter's two-dimensional frequency response matches the clutter spectral distribution and that clutter is adequately suppressed after space-time filtering. The objective function f ₃ (W _d ) will determine the width of the space-time filter notch, f ₃ (W _d ) consists of two parts: sumR _L (W _d ) and sumR _R (W _d ), sumR _L (W _d ) ) and sumR _R (W _d ) are the sum of the two-dimensional frequency responses on the left and right sides of the filter notch, respectively, and are expressed as:

where Δf _d is a constant Doppler frequency. Minimizing the objective function f ₃ (W _d ) can make the frequency responses on the left and right sides of the filter notch have as large a gain as possible, and ensure a certain notch width, so as to avoid the result of notch widening during the solution process, Therefore, the suppression performance of non-stationary clutter in bistatic forward looking SAR is improved.

So far, the problem of solving the space-time filter has been transformed into a mathematical optimization problem with constraints. When the minimum value of the objective function F(W _d ) is obtained, the space-time filter for the optimal matching of the bistatic forward looking SAR can be obtained. device.

S5. Use the particle swarm optimization algorithm to solve the constrained optimization problem, and obtain the optimal solution of the constrained optimization problem;

The specific implementation method is:

S51, initialize the population number X and the number of iterations G, and set the boundary of the decision variable in the particle swarm optimization algorithm;

S52. Split the filter weight coefficient W _di into a real part and an imaginary part:

W _di =x _i +jx _l , i=1,2,...,NK

Among them, the serial number index l=i+NK, the decision vector x=[x ₁ ,x ₂ ,x ₃ ,...,x _2NK ] ^T is a particle in the particle swarm algorithm, that is, an optimal solution, and the optimal solution adopts the particle swarm The algorithm is solved, and the optimal solution obtained is recorded as

The specific method of particle swarm algorithm is:

S521. Initialize the relevant parameters of the particle swarm algorithm, including the decision space V _D , the particle dimension D, the particle number Ω, and the maximum number of iterations G;

表示第g代粒子群体中的第i个粒子，共包含D个自变量；

为该粒子在决策空间中的速度。在初始化过程中，第i个粒子的位置将在其自变量最大值和最小值之间按照均匀分布随机生成，所有粒子的初始速度为零。S522. Initialize the particle group Γ ₁ . Let the number of iterations g=1, generate Ω particles in the decision space V _D to form a particle group Γ ₁ . make

represents the i-th particle in the g-th generation particle population, including D independent variables in total;

is the velocity of the particle in the decision space. During initialization, the position of the ith particle will be randomly generated according to a uniform distribution between the maximum and minimum values of its independent variables, and the initial velocity of all particles is zero.

After the particle population Γ ₁ is initialized, the fitness value F(x _i (g)) of each particle in the population is calculated.

S523. When the number of iterations satisfies g∈[1, G], proceed to step S524; otherwise, end the iteration and proceed to step S526.

S524. According to the calculated fitness value, record the extreme value of each particle as the individual optimal solution pBest of the particle. Through the information sharing between particles, the individual optimal solutions are compared to find the group optimal solution gBest of the current population.

S525 , update the position and velocity of the particle according to the individual optimal solution pBest and the group optimal solution gBest. The particle velocity is updated as follows

v _i (g+1)=κv _i (g)+Δv _i (g)

Among them, κ is a non-negative inertia factor, Δv _i (g) is the velocity update variable of the i-th particle in the g-th generation particle population, expressed as Δv _i (g)=C ₁ ×rand[0 _, 1]× (pBest _{i -xi} (g))+C ₂ ×rand[0 _, 1]×(gBest(g) _-xi (g)), C ₁ and C ₂ are the individual learning factor and social learning factor of the particle, respectively . rand[0,1] is a random number uniformly distributed between 0 and 1.

The particle positions are updated as follows:

x _i (g+1)=x _i (g)+v _i (g+1)

After updating the particle attributes, the next generation particle population Γ _g+1 is obtained. Recalculate the fitness value of each particle, update g=g+1, and then return to step S523.

S526. After the iteration, obtain the particle group Γ _G of the last generation, each particle in the group will gather at the position of the global optimal solution, and the global optimal solution is the solution of the constrained optimization problem

S6. According to the optimal solution obtained by the particle swarm optimization algorithm, reconstruct the optimally matched space-time filter, and use the reconstructed optimally matched space-time filter to filter the unit to be detected to obtain non-stationary clutter suppression signal after.

重构最优匹配空-时滤波器权系数

如下，The specific implementation method is: according to the optimal solution obtained by the particle swarm optimization algorithm

Reconstructing the optimal matching space-time filter weights

as follows,

最后，可以对目标信号S_MSTF(t)进行后续的参数估计与归位聚焦等处理。For the unit to be detected, use the reconstructed optimal matching space-time filter to perform space-time filtering, and obtain the target signal after the non-stationary clutter suppression of the dual-base forward looking SAR.

Finally, subsequent processing such as parameter estimation and homing focusing can be performed on the target signal S _MSTF (t).

Figure 4 shows the echo domain signal after non-stationary clutter suppression in this embodiment, and Figure 5 shows the image domain signal. Figure 6 and Figure 7 show the comparison results before and after clutter suppression. The base forward-looking SAR clutter has been fully suppressed, and only the moving target signal is retained in the echo domain and image domain, which can achieve highly reliable moving target detection.

Those of ordinary skill in the art will appreciate that the embodiments described herein are intended to assist readers in understanding the principles of the present invention, and it should be understood that the scope of protection of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations without departing from the essence of the present invention according to the technical teaching disclosed in the present invention, and these modifications and combinations still fall within the protection scope of the present invention.

Claims (6)

1. The space-time matching-based bistatic forward-looking SAR clutter suppression method is characterized by comprising the following steps:

s1, establishing a BFSAR space geometric model and initializing system parameters;

s2, performing range pulse compression on the echo signal, and performing preprocessing and migration correction on the echo signal after the range pulse compression;

s3, establishing a space-time clutter model according to the BFSAR space geometric model, acquiring space-time frequency information of the clutter of the unit to be detected, and acquiring space-time distribution information of the clutter spectrum according to the coupling relation of the space-time frequency of the clutter;

s4, designing an optimal matching space-time filter of the unit to be detected according to clutter space-time distribution information to obtain a constraint optimization problem;

s5, solving the constrained optimization problem by utilizing a particle swarm optimization algorithm to obtain an optimal solution of the constrained optimization problem;

s6, solving the obtained optimal solution according to the particle swarm optimization algorithm, reconstructing the optimal matching space-time filter, and filtering the unit to be detected by using the reconstructed optimal matching space-time filter to obtain a signal after non-stationary clutter suppression.

2. The space-time matching-based bistatic forward-looking SAR clutter suppression method according to claim 1, wherein the S2 is specifically realized by the following steps: preprocessing the echo signal by adopting a filtering mode, and performing migration correction on the echo signal through trapezoidal distortion correction;

filter H in preprocessing_pre(t,f_τ) Comprises the following steps:

wherein f is_refIs the Doppler centroid of the reference point, f_τAnd f_cRespectively distance frequency and carrier frequency;

the keystone correction function is expressed as:

t＝f_ct₁/(f_τ+f_c)

wherein, t₁Is the new azimuth time after the transformation;

after the above processing, the signal received by the nth channel is represented as:

wherein, P represents clutter scattering points or moving targets in an observation scene, and σ (P) is a backscattering coefficient of P, ω_a() represents an azimuthal envelope; parameter τ, t, B_τλ, c and t_PRespectively representing distance time, azimuth time, distance bandwidth, wavelength, light speed and beam center time; r_s(0, n; P) is the bistatic distance of the point P relative to the nth channel at the azimuth moment of 0; r_s(t₁N; p) is the double base distance of point P relative to the nth receive channel;

the processing time is divided into a plurality of sub-time periods by adopting a time-sharing processing means, and in the time-sharing processing, each sub-time period satisfies the following relation:

wherein, Delta T is the length of the time-sharing processing sub-time interval, K_aFor adjusting the frequency, Δ δ, by Doppler_aFor Doppler resolution, T_synIs the synthetic aperture time;

after the migration correction and the time-sharing processing are carried out on the echo signal, the data in each sub-period is subjected to column vectorization to obtain space-time sample data in each sub-period, which is expressed as S_vec(t)。

3. The space-time matching-based bistatic forward-looking SAR clutter suppression method according to claim 1, wherein the S3 is specifically realized by the following steps: the space-time frequency information of the clutter of the unit to be detected is as follows:

wherein f is_dAnd f_sNormalized Doppler frequency and normalized spatial frequency, f, of ground clutter scattering points, respectively_rIs the pulse repetition frequency, V_TAnd V_RThe flight speeds of the transmitter and receiver, respectively,. psi_TAnd psi_RThe spatial cone angle, theta, of the ground clutter scattering point with respect to the transmitter and receiver, respectively_RAnd

azimuth and elevation angles, theta, respectively, of ground clutter scattering points relative to the receiver_pThe included angle between the receiver array and the flight direction is shown, and d and lambda respectively represent the channel spacing and the signal wavelength;

the space-time distribution of the clutter spectrum in the cells to be examined is expressed as:

θ_Tand

azimuth and elevation angles, delta, respectively, of clutter scattering sites relative to the transmitter_TAnd delta_RRespectively, the angles of the transmitter and receiver flight directions relative to the bistatic baseline.

4. The space-time matching-based bistatic forward-looking SAR clutter suppression method according to claim 3, wherein the S4 is specifically realized by the following steps: for the system with N channels and K pulses, the filter is designedThe filter weight coefficient is W_d，W_dIs an NK × 1 dimensional complex vector, expressed as:

W_d＝[W_d1,W_d2,...,W_dnk,...,W_dNK]^T

wherein, W_dnkIs the weight coefficient component of the nk dimension;

the two-dimensional space-time frequency response of this filter is expressed as:

wherein,

is a space-time steering vector; s_tE.g. Kx 1 and S_sE N × 1 are respectively time and space steering vectors, which are respectively expressed as:

the response of the filter to a moving object is represented as

Wherein, V_tE.g. K x 1 and V_sThe epsilon is N multiplied by 1 and is respectively the time and space guide vector of the moving target;

uniformly dividing a receive beam into Q sub-beams, Q being the synthetic aperture length L_synAnd azimuth resolution ρ_aDetermining, expressed as Q ═ L_syn/ρ_a(ii) a After beam division, the clutter signals in the receiver are obtained by superposing clutter echoes in the Q directions; therefore, only filtering is needed to realize effective suppression of bistatic forward-looking SAR non-stationary clutterThe space-time frequency response of the wave filter in the Q sub-wave beam directions is zero; according to the above thought, the following constraint optimization problem is established:

wherein epsilon is the error tolerance of the observation noise; { f_u(W_d) | u ═ 1,2,3} is an objective function expressed as:

wherein f is_diAnd f_siNormalized Doppler frequency and normalized space frequency in the ith sub-beam direction; objective function f₁(W_d) And f₂(W_d) Respectively representing the notch mean and variance of the space-time filter arranged in the direction of the Q sub-beams; objective function f₃(W_d) Will determine the width of the space-time filter notch, f₃(W_d) The device consists of two parts: sumR_L(W_d) And sumR_R(W_d)，sumR_L(W_d) And sumR_R(W_d) The sum of the two-dimensional frequency responses, left and right of the filter notch, respectively, is expressed as:

wherein, Δ f_dIs a constant doppler frequency.

5. The space-time matching-based bistatic forward-looking SAR clutter suppression method according to claim 4, wherein the S5 is specifically realized by the following steps:

s51, initializing the population quantity X and the iteration times G, and setting the boundary of a decision variable in a particle swarm optimization algorithm;

s52, weighting the filter weight coefficient W_diSplitting into a real part and an imaginary part:

W_di＝x_i+jx_l,i＝1,2,…,NK

where, the index of sequence number l ═ i + NK, and the decision vector x ═ x₁,x₂,x₃,…,x_2NK]^TIs a particle in the particle swarm optimization, namely an optimized solution, the optimized solution is solved by adopting the particle swarm optimization, and the obtained optimal solution is recorded as

6. The space-time matching-based bistatic forward-looking SAR clutter suppression method according to claim 5, wherein the S6 is specifically realized by the following steps: solving the optimal solution obtained according to the particle swarm optimization algorithm

Reconstructing optimal matching space-time filter weight coefficients

As follows below, the following description will be given,

for the unit to be detected, performing space-time filtering by using the reconstructed optimal matching space-time filter to obtain a target signal after the bistatic forward-looking SAR non-stationary clutter suppression

Images