A low signal-to-noise ratio inverse synthetic aperture radar imaging method

Abstract

The present invention discloses a low signal-to-noise ratio inverse synthetic aperture radar imaging method, which uses a sidelobe learning particle swarm optimization algorithm and a generalized Radon-Fourier transform to estimate the amplitude of a scattering point, thereby avoiding the image quality degradation caused by the traditional grid division. The traditional PSO algorithm is improved. After obtaining the maximum amplitude among multiple particles, a certain number of sidelobes of particles are extended to both sides of the particle to continue searching for the maximum amplitude as the amplitude of the scattering point, thereby improving the accuracy of the final maximum amplitude calculation. The signal component of the scattering point corresponding to the maximum amplitude is removed, and the amplitude of the remaining scattering points is repeatedly estimated to obtain the maximum amplitude and corresponding motion parameters of each scattering point. This avoids the problem of strong scattering points interfering with the detection of weak scattering points in the traditional constant false alarm rate detection process, and the problem of noise points appearing in the final ISAR image due to the presence of noise, thereby reducing the image quality.

Description

Technical Field

[0001] The present invention relates to the technical field of inverse synthetic aperture radar, and in particular to a low signal-to-noise ratio inverse synthetic aperture radar imaging method. Background Art

[0002] Inverse Synthetic Aperture Radar (ISAR) is a key branch of synthetic aperture radar (SAR). Unlike traditional radar, ISAR is a high-resolution imaging radar capable of acquiring detailed images of non-cooperative moving targets, such as aircraft, ships, and missiles, in all weather conditions and at all times, over long distances. This technology holds significant value for both military and civilian applications.

[0003] ISAR-based moving target imaging techniques compensate for the target's translational component. This involves removing the consistent motion between scattering elements (the target's translational motion). Then, using the non-uniform motion caused by target sway, they estimate the target's motion parameters through parameter estimation, ultimately achieving ISAR imaging. However, these methods assume that the target's translational component can be perfectly compensated, an assumption that places certain demands on the target echo's signal-to-noise ratio. When the target echo's signal-to-noise ratio is low, traditional translational compensation methods become ineffective, significantly reducing parameter estimation accuracy. To solve this problem, a low signal-to-noise ratio parameter estimation method based on generalized radon Fourier transform (GRFT) and constant false-alarm rate (CFAR) detection can be used to achieve ISAR imaging. However, this method still faces a series of problems: the first problem is that the accuracy of parameter estimation is related to the GRFT grid division accuracy. When the grid division accuracy is low, there will be errors in the estimation results of the target motion parameters, which will eventually lead to a decrease in the target image quality; the second problem is that during the CFAR detection process, strong scattering points will interfere with the detection of weak scattering points. A uniformly determined threshold is difficult to balance the retention of weak scattering points with the removal of strong scattering point interference; third, due to the presence of noise, the CFAR detection process will inevitably regard noise above the threshold as real signal, which will inevitably lead to the appearance of noise in the final ISAR image, reducing the image quality. Summary of the Invention

[0004] In view of this, the present invention provides a low signal-to-noise ratio inverse synthetic aperture radar imaging method, which can achieve instantaneous ISAR high-resolution imaging of ship targets based on obtaining the amplitude and motion parameters of all scattering points of the target.

[0005] The specific technical solutions adopted in the present invention are as follows:

[0006] A low signal-to-noise ratio inverse synthetic aperture radar imaging method, comprising:

[0007] Step 1: Perform range-direction pulse compression on the echo signal of the moving target on the sea surface to obtain a one-dimensional high-resolution range image of the echo signal;

[0008] Step 2: Using the motion parameter range of the scattering points in the one-dimensional high-resolution range image as the position parameter range of the particles, the generalized Radon-Fourier transform (GRFT) is used to obtain the amplitude of each particle. The sidelobe learning particle swarm optimization algorithm (SLL-PSO) is used to search for the particle with the largest amplitude, and the signal component of the scattering point corresponding to the particle with the largest amplitude is calculated.

[0009] The SLL-PSO method comprises the following steps: during an iteration, selecting a particle with the largest amplitude as the global optimal particle; then expanding the acceleration and jerk of the global optimal particle to the left and right to obtain an extended particle; wherein the expansion width is the total sidelobe width of a certain number of sidelobes; continuously searching for the particle with the largest amplitude between the global optimal particle and the expanded particles as a new global optimal particle; and using the position parameters of the new global optimal particle to represent the motion parameters of the scattering point, and using the amplitude of the new global optimal particle to represent the amplitude of the scattering point;

[0010] Step 3: Remove the signal component of the scattering point corresponding to the particle with the largest amplitude obtained in step 2 from the one-dimensional high-resolution range image;

[0011] Step 4: Repeat steps 2 and 3 for the one-dimensional high-resolution range image processed in step 3 until the residual energy of the one-dimensional high-resolution range image is less than a threshold, thereby obtaining the amplitude and motion parameters of all scattering points of the echo signal;

[0012] Step 5: Construct a radar image based on the motion parameters obtained in step 4.

[0013] Furthermore, in step 2, the signal component of the scattering point is calculated as follows:

[0014]

[0015] Where j is the sign of the imaginary part, t is the azimuth time, and λ is the wavelength; R 0i 、v Si 、a Si is the motion parameter of the scattering point, R 0i is the equivalent slant range constant of the radar to the i-th scattering point, v Si is the equivalent velocity of the i-th scattering point, a Si is the equivalent acceleration of the i-th scattering point; A rc is the signal gain brought by range pulse compression, σ i is the intensity of the i-th scattering point, tr Represents the distance to fast time, i∈M, M is the total number of scattering points on the moving target on the sea surface, and if M=1, the signal component of the scattering point can be obtained.

[0016] Furthermore, in step 2, the extended particle is represented as:

[0017]

[0018] Among them, R m , v m , a m , b m is the position parameter of the mth particle, R m represents the slant distance of the mth particle, v m represents the velocity of the mth particle, a m represents the acceleration of the mth particle, b m represents the acceleration of the mth particle, gbest represents the global optimal particle, W A Represents the sidelobe width of the acceleration dimension, W B Indicates the sidelobe width of the acceleration dimension, p is an integer, indicating the number of sidelobes in the acceleration dimension, q is an integer, indicating the number of sidelobes in the jerk dimension, pW A That is, the total sidelobe width of a certain number of sidelobes expanded by acceleration, qW B That is, the total sidelobe width of a certain number of sidelobes expanded by the jerk.

[0019] Furthermore, the number of acceleration side lobes p and the number of jerk side lobes q are determined as follows:

[0020]

[0021] Among them, a m0 with b m0 are the acceleration and jerk values of all particles at initialization, To round up.

[0022] Furthermore, the sidelobe width W of the acceleration dimension A and the sidelobe width W of the acceleration dimension B The method of determining is:

[0023]

[0024] Among them, A n With B n is the nth inflection point of acceleration dimension and jerk dimension, A n+1 With B n+1 It is the n+1th inflection point of the acceleration dimension and the jerk dimension, where n and n+1 are integers.

[0025] Furthermore, in step three, the signal component of the scattering point corresponding to the particle with the largest amplitude obtained in step two is removed from the one-dimensional high-resolution range image by subtracting the signal component of the scattering point corresponding to the particle with the largest amplitude from the one-dimensional high-resolution range image in step one.

[0026] Furthermore, in step 4, the threshold is set to 1% of the initial energy of the one-dimensional high-resolution range image.

[0027] Furthermore, in step 4, the motion parameters include the equivalent slant range constant R from the radar to the i-th scattering point 0i , the equivalent velocity v of the i-th scattering point Si and the equivalent acceleration a of the i-th scattering point Si .

[0028] Furthermore, in step 5, the radar image is constructed based on the motion parameters obtained in step 4 as follows: first, the equivalent velocity v of the i-th scattering point is calculated. Si and the equivalent acceleration a of the i-th scattering point Si The Doppler frequency history of the i-th scattering point is calculated, and then the equivalent slant range constant and Doppler frequency of all scattering points at any moment are used as the coordinate information of the corresponding scattering point to obtain the radar image.

[0029] Beneficial effects:

[0030] (1) A low signal-to-noise ratio inverse synthetic aperture radar imaging method uses the sidelobe learning particle swarm optimization algorithm SLL-PSO and the generalized Radon-Fourier transform GRFT to estimate the amplitude of a single scattering point, avoiding the problem of image quality degradation caused by the grid division accuracy in the traditional grid division method. At the same time, after obtaining the maximum amplitude among multiple particles, a certain number of sidelobe particles are extended to both sides of the particle, i.e., the global optimal particle, to continue to search for the maximum amplitude as the amplitude of the scattering point, avoiding the problem that the traditional PSO algorithm may find the local maximum as the global maximum, thereby improving the accuracy of the maximum amplitude calculation. By removing the signal component of the scattering point corresponding to the maximum amplitude and repeatedly estimating the amplitude of a single scattering point among the remaining scattering points, the maximum amplitude and corresponding motion parameters of each scattering point can be obtained, avoiding the problem that strong scattering points interfere with the detection of weak scattering points in the traditional constant false alarm rate CFAR detection process and the problem that noise appears in the final ISAR image due to the presence of noise, thereby reducing the image quality.

[0031] (2) By using the one-dimensional high-resolution range image to subtract the signal component of the scattering point corresponding to the maximum amplitude, the influence of the signal component of the scattering point corresponding to the maximum amplitude on the estimation of the motion parameters and amplitude of the remaining scattering points can be avoided, thereby improving the efficiency and accuracy of the motion parameter estimation of each scattering point. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 Schematic diagram of the geometric model of inverse synthetic aperture radar ISAR.

[0033] Figure 2 Schematic diagram of GRFT results for a target point

[0034] Figure 3 This is a schematic diagram of a simplified ship model used in a specific embodiment of the present invention.

[0035] Figure 4 Schematic diagram of traditional ISAR imaging results.

[0036] Figure 5 Schematic diagram of ISAR imaging results of GRFT using constant false alarm rate (CFAR) detection.

[0037] Figure 6 The figure is a schematic diagram of ISAR imaging results using a low signal-to-noise ratio inverse synthetic aperture radar imaging method of the present invention.

[0038] Figure 7 This is a flow chart of a low signal-to-noise ratio inverse synthetic aperture radar imaging method of the present invention. DETAILED DESCRIPTION

[0039] The present invention provides a low signal-to-noise ratio inverse synthetic aperture radar imaging method. The method uses a sidelobe learning particle swarm optimization algorithm (SLL-PSO) to search in a one-dimensional high-resolution range image, uses a generalized Radon-Fourier transform (GRFT) to obtain the amplitude of each particle, and calculates the signal component of a scattering point corresponding to the maximum amplitude. After obtaining the amplitudes of multiple particles, the particle with the largest amplitude is selected as a global optimal particle. Then, a certain number of sidelobe particles are extended to both sides of the global optimal particle to continue searching for the particle with the largest amplitude as a new global optimal particle. The position parameters of the new global optimal particle are used to characterize the motion parameters of the scattering point, and the amplitude of the new global optimal particle is used to characterize the amplitude of the scattering point. Then, the signal component of the scattering point corresponding to the maximum amplitude obtained in step 2 is removed from the one-dimensional high-resolution range image. Steps 2 and 3 are repeated until the residual energy of the one-dimensional high-resolution range image is less than a threshold value, thereby obtaining the amplitudes and motion parameters of all scattering points of the echo signal and constructing a radar image. It avoids the problem of image quality degradation caused by grid division accuracy in the traditional grid division method, the problem of strong scattering points interfering with weak scattering point detection in the traditional constant false alarm rate CFAR detection process, and the problem of noise appearing in the final ISAR image due to the presence of noise, thereby reducing image quality.

[0040] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0041] A low signal-to-noise ratio inverse synthetic aperture radar imaging method, the process is as follows Figure 7 As shown, the following steps are included:

[0042] Step 1: Perform range-direction pulse compression on the echo signal of the moving target on the sea surface to obtain a one-dimensional high-resolution range image of the echo signal.

[0043] First, the geometric model of inverse synthetic aperture radar (ISAR) imaging of sea surface moving targets is introduced. Figure 1 The figure shows a schematic diagram of the ISAR geometric model, where the origin O of the OXYZ coordinate system is the center of mass of the moving target, the Y axis is the direction of motion of the ISAR platform, the Z axis is perpendicular to the sea level, and the X axis is determined by the right-hand rule. The origins of the OBWZ coordinate system and the ORHV coordinate system are also the center of mass of the moving target. The B axis and the W axis represent the length and width of the target, respectively, the R axis represents the radar line of sight (LOS), the H axis belongs to the XOY plane and is perpendicular to the R axis, and the V axis is determined by the right-hand rule. The angle ψ is the ground grazing angle, and the angle is the angle between the B axis and the X axis, the angle ζ is the angle between the projection of the R axis on the XOY plane and the X axis at the observation center. G is the intersection of the plumb line of the ISAR platform and the sea level, and H is the platform height. R 、ω Y 、ω PThey represent the angular velocities of roll, pitch, and yaw, respectively.

[0044] Based on the ISAR geometric model, after converting from the OBWZ coordinate system to the OXYZ coordinate system and then to the ORHV coordinate system, the slant range history expression between the i-th scattering point on the moving target and the radar can be obtained as follows:

[0045] R i (t)≈R0+R T (t)+r i (t) (1)

[0046] Where R0 is the distance from the center of mass of the moving target to the radar, R T (t) is the slant range introduced by the moving target and the ISAR platform's own motion, and it is the same for all scattering points on the moving target, r i (t) is the slant range introduced by the time-varying three-dimensional rotation of the moving target driven by the waves, and t represents the azimuth time.

[0047] On this basis, using Taylor series expansion, a second-order slant range model that is more suitable for actual imaging processing is obtained:

[0048] R i (t)≈R 0i +v Ri t+a Ri t 2 +v T t+a Ri t 2 =R 0i +v Si t+a Si t 2 (2)

[0049] And the corresponding echo signal model after range pulse compression, that is, the one-dimensional high-resolution range image of the echo signal:

[0050]

[0051] Where, j is the sign of the imaginary part, t is the azimuth time, λ is the wavelength, R 0i is the equivalent slant range constant of the radar to the i-th scattering point, v Ri is the velocity component of the i-th scattering point caused by the target rotation, a Ri is the acceleration component of the i-th scattering point caused by the target rotation, b Ri is the acceleration component of the i-th scattering point caused by the target rotation, v T is the velocity component of the i-th scattering point caused by the target and the platform's own motion, a Tis the acceleration component of the i-th scattering point caused by the target and the platform's own motion, b T is the acceleration component of the i-th scattering point caused by the target and the platform's own motion. In terms of quantity, v Si =v Ri +v T , a Si =a Ri +a T , b Si =b Ri +b T . i∈M, M is the number of scattering points on the target, A rc =T p B is the signal gain brought by range pulse compression, where T p is the pulse width, B is the signal bandwidth, σ i is the intensity of the i-th scattering point, t r Indicates distance to fast time.

[0052] Step 2: Using the motion parameter range of the scattering points in the one-dimensional high-resolution range image as the position parameter range of the particles, the generalized Radon-Fourier transform (GRFT) is used to obtain the amplitude of each particle. The sidelobe learning particle swarm optimization algorithm (SLL-PSO) is used to search for the particle with the largest amplitude, and the signal component of the scattering point corresponding to the particle with the largest amplitude is calculated.

[0053] The SLL-PSO algorithm is as follows: during the iteration process, the particle with the largest amplitude is selected as the global optimal particle, and then the acceleration and jerk of the global optimal particle are expanded to the left and right to obtain an extended particle; the expansion width is the total sidelobe width of a certain number of sidelobes; the particle with the largest amplitude is continued to be searched between the global optimal particle and the expanded particles as the new global optimal particle, and the position parameters of the new global optimal particle are used to represent the motion parameters of the scattering point, and the amplitude of the new global optimal particle is used to represent the amplitude of the scattering point.

[0054] Among them, the extended particle is expressed as:

[0055]

[0056] Among them, R m , v m , a m , b m is the position parameter of the mth particle, R m represents the slant distance of the mth particle, v m represents the velocity of the mth particle, a m represents the acceleration of the mth particle, b m represents the acceleration of the mth particle, gbest represents the global optimal particle, W Ais the sidelobe width of the acceleration dimension of the target GRFT result, W B is the sidelobe width of the target GRFT result acceleration dimension, p is the number of sidelobes of the target GRFT result acceleration dimension, q is the number of sidelobes of the target GRFT result acceleration dimension, pW A That is, the total sidelobe width of a certain number of sidelobes expanded by acceleration, qW B That is, the total sidelobe width of a certain number of sidelobes expanded by the jerk.

[0057] In addition, due to limited space, some of the expanded content of the expanded particles is replaced by ellipsis. For the first example, from the middle particle gbest (R m ,v m ,a m -pW A ,b m ) Expand a W to the left B , until q W are expanded B Get gbest(R m ,v m ,a m -pW A ,b m -qW B ), then extend a W to the right B Start, until q W are expanded B Get gbest(R m ,v m ,a m -pW A ,b m +qW B ). And so on, for each row. And so on, for each column, it is to expand a W A , until p W are expanded A So far, no further description is given here.

[0058] The definition of GRFT is as follows:

[0059]

[0060] Where G(r, v, a, b) is the result of GRFT. In the specific embodiment of the present invention, the result of GRFT is the estimation result of the motion parameters and the signal component with the largest amplitude. ob is the observation time, r, v, a and b are the process parameters of GRFT. During the search process of PSO, the process parameters match the motion parameters of the scattering points, that is, they are equal; [γ min γ max ]γ=r,v,a,b is the motion parameter search range, γ min with γ maxThe lower and upper bounds of the target motion parameter search range are as follows. When the process parameters r, v, a, b are equal to the motion parameters R of the scattering point, 0i ,v Si ,a Si ,b Si When , the GRFT result, that is, the amplitude at the process parameter, will reach the maximum value, that is, a related focusing peak will appear, so that the amplitude and motion parameters of each scattering point can be estimated.

[0061] R 0i ,v Si ,a Si ,b Si They represent the equivalent slant range constant from the radar to the i-th scattering point, the equivalent velocity, equivalent acceleration and equivalent jerk of the i-th scattering point respectively.

[0062] In order to avoid the problem of insufficient parameter estimation accuracy caused by the grid division step in the GRFT process, the present invention uses the PSO algorithm to find the global maximum of GRFT. However, since the target GRFT result has the following problems: Figure 2 The local maxima corresponding to the many side lobes shown. Figure 2 (a) is the projection on the range-velocity plane, Figure 2 (b) in the figure is the projection on the distance-acceleration plane. Figure 2 (c) in the figure is the projection on the velocity-acceleration plane. Figure 2 (d) in the equation is the distance plane projection. Figure 2 (e) in the figure is the velocity plane projection. Figure 2 (f) in the equation is the acceleration plane projection. PSO may also reach local maxima due to factors such as premature particle maturation. To address this issue, the present invention improves the PSO algorithm and proposes a sidelobe learning particle swarm optimization algorithm (SLL-PSO). To estimate the parameters of each scattering point, the global maximum value among multiple particles is searched as the signal component of the single scattering point corresponding to the maximum amplitude. After obtaining the maximum amplitude of the single scattering point, i.e., the global maximum value among multiple particles, i.e., the amplitude of the globally optimal particle, the search for the global maximum is continued by expanding a certain number of sidelobe particles to both sides of the globally optimal particle. This avoids the possibility of PSO reaching local maxima. The expanded global maximum value is then used as the output to obtain the motion parameters of the single scattering point and the signal component with the maximum amplitude.

[0063] Estimating the motion parameters and maximum amplitude of each scattering point includes the following steps:

[0064] 1. Initialize particles:

[0065] x m =(R m,v m ,a m ,b m ),m=1,2,…,NP

[0066] x m , R m , v m , a m , b m are the position parameters of the m-th particle, respectively, representing the position, equivalent slant distance, equivalent velocity, equivalent acceleration, and equivalent jerk of the m-th particle, and NP is the total number of particles.

[0067] Initialization is to use the range of motion parameters of the scattering points of the one-dimensional high-resolution range image to initialize the range of the particle's position parameters and the particle's own speed range, and then randomly select a particle position within the range of the particle's position parameters as the initialization position, that is, set the current position of each particle to its initial optimal position, and the position parameter of the current position particle is the initialized position parameter; randomly select a particle speed within the range of the particle's speed parameters as the initialization speed.

[0068] Substitute the above parameters into formula (4) to calculate the GRFT value G(x m ).

[0069] 2. Set the current position of each particle to its initial best position pbest m , take the best position pbest of all particles in the current scattering point m The maximum value is taken as the global optimal particle, that is, the global maximum value gbest, which is expressed as follows:

[0070]

[0071] Among them, argmax means to find the function G(pbest m ) reaches the maximum value, the corresponding parameter is pbest m .

[0072] 4. Because each particle has its own moving direction and speed, the particle's own speed needs to be updated:

[0073] υ m,k+1 =wυ m,k +c p r1(x m,k -pbest m )+c g r2(x m,k -gbest)

[0074] Among them, m,k+1 and υm,k are the velocities of the mth particle at the k+1th and kth iterations, c r and c g is the learning factor, generally set to 2, r1 and r2 are random numbers between (0, 1).

[0075] 5. Use the updated velocity v m,k+1 , calculate the new position of the current particle and update pbest m and gbest.

[0076] x m =x m +v m,k+1

[0077] 6. Since PSO may also obtain local maxima, after obtaining the global maximum value among multiple particles at a single scattering point, we expand the number of sidelobe particles on both sides of the global optimal particle to continue searching for the global maximum value: the acceleration and jerk of the global optimal particle are expanded by the sidelobe width of a certain number of particles on both sides, and the particle with the largest amplitude is found in the expanded search range as the new global optimal particle, that is, gbest is expanded and updated to gbest(p,q):

[0078]

[0079] Among them, W A is the sidelobe width in the acceleration dimension, W B is the sidelobe width of the acceleration dimension, p is an integer, indicating the number of sidelobes in the acceleration dimension, and q is an integer, indicating the number of sidelobes in the jerk dimension.

[0080] The acceleration and the number of jerk sidelobes can be determined by the following formula:

[0081]

[0082] In the above formula, a m0 with b m0 are the acceleration and jerk values of NP particles at initialization, For W A With W B The value of can be obtained by the following formula:

[0083]

[0084] Among them, A n With B n It is the nth inflection point of a and b in the acceleration dimension and jerk dimension in the process parameters, A n+1 With B n+1It is the n+1th inflection point of process parameters a and b in the acceleration dimension and jerk dimension.

[0085] Since the motion parameter R of the scattering point 0i ,v Si ,a Si ,b Si The width of the side lobe of the amplitude of the scattering point will not be affected, so A n With A n+1 , B n With B n+1 It will only appear in different positions of the acceleration dimension and the jerk dimension as the motion parameters of the scattering point change, and A n With A n+1 , B n With B n+1 Therefore, let R in formula (4) 0i =v Si =a Si =b Si =0 has:

[0086]

[0087] Then let the partial derivatives of equation (5) with respect to a and b be 0 to obtain:

[0088]

[0089] Then solve equation (6) to obtain A n With B n .

[0090] 7. Find the values of p and q that maximize equation (4) in the obtained gbest(p,q) and update gbest:

[0091]

[0092] 8. The difference between the current and the next two G(gbest) is less than 10 -3 , that is, when convergence occurs, end the while loop.

[0093] 9. Output the global optimal particle, i.e., the global maximum value gbest, which is used to characterize the motion parameters of a single scattering point, including slant range, velocity, and acceleration. The GRFT value corresponding to the global maximum value is the maximum amplitude G(gbest) of a single scattering point.

[0094] Step 3: Remove the signal component of the scattering point corresponding to the particle with the largest amplitude obtained in step 2 from the one-dimensional high-resolution range image.

[0095] The process of removing the estimated signal component is completed by subtracting the signal component of the scattering point corresponding to the particle with the largest amplitude from the one-dimensional high-resolution range image in step 1.

[0096] The signal component of a single scattering point is calculated by substituting the motion parameters corresponding to the scattering point into formula (3). When M = 1, the signal component of the scattering point corresponding to the particle with the largest amplitude is obtained. The process of removing the signal component of a single scattering point also completes the process of removing the scattering point.

[0097] Step 4: Using the CLEAN technique, repeat steps 2 and 3 for the one-dimensional high-resolution range image processed in step 3 until the residual energy of the one-dimensional high-resolution range image is less than the threshold, thereby obtaining the amplitude and motion parameters of all scattering points of the echo signal.

[0098] The motion parameters include the equivalent slant range constant R from the radar to the i-th scattering point 0i , the equivalent velocity v of the i-th scattering point Si and equivalent acceleration a Si .

[0099] The threshold is set to 1% of the initial energy of the one-dimensional high-resolution range image. The residual energy of the one-dimensional high-resolution range image is calculated by averaging the square values of the echo signals of the remaining scattering points in a traditional way.

[0100] Step 5: Construct a radar image based on the motion parameters obtained in step 4.

[0101] According to the equivalent velocity v of the i-th scattering point caused by the target rotation obtained in step 4 Si and equivalent acceleration a Si , the Doppler frequency history of the i-th scattering point is calculated as:

[0102]

[0103] Among them, t a Indicates azimuth slow time.

[0104] According to the equivalent slant range constant R of the i-th scattering point at any time 0i and the Doppler frequency f of the i-th scattering point diThe radar image at that moment is obtained. The radar image is then obtained by using the equivalent slant ranges and Doppler frequencies of all remaining scattering points as the coordinate information for the corresponding scattering points. By using the equivalent slant ranges and Doppler frequencies as the coordinate information for the corresponding scattering points, translation compensation and migration through resolution cell (MTRC) correction are performed on the target's one-dimensional range image, resulting in more accurate imaging results.

[0105] By selecting any imaging time t0, we can get the ISAR sea surface ship range-instantaneous Doppler radar image at that time, which is expressed as:

[0106]

[0107] Among them, f a represents the azimuth frequency domain, The number of coherent peaks that meet the signal-to-noise ratio requirements obtained by the above parameter estimation is the number of remaining scattering points, Represents the intensity of the estimated i-th scattering point among the remaining scattering points, represents the equivalent slant range constant estimated by the radar to the i-th scattering point among the remaining scattering points, and δ represents the impulse function.

[0108] To demonstrate the effectiveness of the proposed method, ISAR imaging was implemented using computer simulation. The simulation parameters are shown in Table 1, and the motion parameters are shown in Table 2.

[0109] Table 1 Computer simulation parameters

[0110] Parameter / Unit Numerical Parameter / Unit Numerical Band Ku Sampling frequency / MHz 640 Platform speed / (m·s) 18.94 Wipe the corner / deg 20 Height / m 400 Pulse duration / μs 6 Signal bandwidth / MHz 480 Observation time / s 1.8 PRF / Hz 350 Echo signal noise ratio / dB -35

[0111] Table 2 Computer simulation motion parameters

[0112] Parameter / Unit Numerical Parameter / Unit Numerical ISAR platform speed / (m·s) 150 <![CDATA[Target speed / m·s -1 > 5 Rolling motion range / deg 1.0 Pitch motion range / deg 0.31 Yaw motion amplitude / deg 0.35 Roll motion period / s 8.6 Pitch motion period / s 4.7 Yaw motion cycle / s 10.0

[0113] The target model used in computer simulation is as follows: Figure 3 As shown, Figure 3 (a) in the figure shows the three-dimensional structure diagram of the target model. Figure 3 (b) in the figure shows the top view of the target model. Figure 3 (c) in the figure shows the side view of the target model. The conventional ISAR imaging results are shown in Figure 2. Figure 4 The ISAR imaging results of GRFT detected by CFAR are shown in Figure 5 As shown, the ISAR imaging results of the method of the present invention are as follows Figure 6 By comparison, it can be seen Figure 5 and Figure 6 The imaging results were better than Figure 4 Clear, and Figure 5 and Figure 6 Compared with the imaging results, it is obvious that Figure 5 Noise points exist in the displayed imaging result, and the image quality is even lower, which proves the effectiveness of the low signal-to-noise ratio inverse synthetic aperture radar imaging method of the present invention.

[0114] The present invention provides a low-SNR inverse synthetic aperture radar imaging method, which can realize translation compensation under low SNR and has better SNR performance than traditional ISAR imaging. The present invention can solve the problem of imaging error caused by GRFT grid division in the existing low-SNR parameter estimation technology based on GRFT. The present invention can solve the problem of side lobes and unnecessary main lobe values introduced by CFAR, and realize ISAR imaging with strong robustness under low SNR. The present invention can solve the problem of image noise introduced by CFAR detection, and improve the quality of ISAR imaging under low SNR.

[0115] The above specific embodiments merely illustrate the design principles of the present invention. The shapes and names of the components described herein may vary and are not limiting. Therefore, those skilled in the art may modify or substitute equivalents for the technical solutions described in the above embodiments. Such modifications and substitutions, without departing from the inventive spirit and technical solutions of the present invention, shall fall within the scope of protection of the present invention.

Claims

1. A low signal-to-noise ratio inverse synthetic aperture radar imaging method, characterized in that: include: Step 1: Perform range-direction pulse compression on the echo signal of the moving target on the sea surface to obtain a one-dimensional high-resolution range image of the echo signal; Step 2: Using the motion parameter range of the scattering points in the one-dimensional high-resolution range image as the position parameter range of the particles, the generalized Radon-Fourier transform (GRFT) is used to obtain the amplitude of each particle. The sidelobe learning particle swarm optimization algorithm (SLL-PSO) is used to search for the particle with the largest amplitude, and the signal component of the scattering point corresponding to the particle with the largest amplitude is calculated. The SLL-PSO method comprises the following steps: during an iteration, selecting a particle with the largest amplitude as the global optimal particle; then expanding the acceleration and jerk of the global optimal particle to the left and right to obtain an extended particle; wherein the expansion width is the total sidelobe width of a certain number of sidelobes; continuously searching for the particle with the largest amplitude between the global optimal particle and the expanded particles as a new global optimal particle; and using the position parameters of the new global optimal particle to represent the motion parameters of the scattering point, and using the amplitude of the new global optimal particle to represent the amplitude of the scattering point; Step 3: Remove the signal component of the scattering point corresponding to the particle with the largest amplitude obtained in step 2 from the one-dimensional high-resolution range image; Step 4: Repeat steps 2 and 3 for the one-dimensional high-resolution range image processed in step 3 until the residual energy of the one-dimensional high-resolution range image is less than a threshold, thereby obtaining the amplitude and motion parameters of all scattering points of the echo signal; Step 5: Construct a radar image based on the motion parameters obtained in step 4.

2. The low signal-to-noise ratio inverse synthetic aperture radar imaging method according to claim 1, wherein: In step 2, the signal component of the scattering point is calculated as follows: Where j is the sign of the imaginary part, t is the azimuth time, and λ is the wavelength; R 0i 、v Si 、a Si is the motion parameter of the scattering point, R 0i is the equivalent slant range constant of the radar to the i-th scattering point, v Si is the equivalent velocity of the i-th scattering point, a Si is the equivalent acceleration of the i-th scattering point; A rc is the signal gain brought by range pulse compression, σ i is the intensity of the i-th scattering point, t r Represents the distance to fast time, i∈M, M is the total number of scattering points on the moving target on the sea surface, and if M=1, the signal component of the scattering point can be obtained.

3. The low signal-to-noise ratio inverse synthetic aperture radar imaging method according to claim 2, wherein: In step 2, the extended particle is represented as: Among them, R m , v m , a m , b m is the position parameter of the mth particle, R m represents the slant distance of the mth particle, v m represents the velocity of the mth particle, a m represents the acceleration of the mth particle, b m represents the acceleration of the mth particle, gbest represents the global optimal particle, W A Represents the sidelobe width of the acceleration dimension, W B Indicates the sidelobe width of the acceleration dimension, p is an integer, indicating the number of sidelobes in the acceleration dimension, q is an integer, indicating the number of sidelobes in the jerk dimension, pW A That is, the total sidelobe width of a certain number of sidelobes expanded by acceleration, qW B That is, the total sidelobe width of a certain number of sidelobes expanded by the jerk.

4. The low signal-to-noise ratio inverse synthetic aperture radar imaging method according to claim 3, wherein: The number of acceleration side lobes p and the number of jerk side lobes q are determined as follows: Among them, a m0 with b m0 are the acceleration and jerk values of all particles at initialization, To round up.

5. The low signal-to-noise ratio inverse synthetic aperture radar imaging method according to claim 4, wherein: The sidelobe width W of the acceleration dimension A and the sidelobe width W of the acceleration dimension B The method of determining is: Among them, A n With B n is the nth inflection point of acceleration dimension and jerk dimension, A n+1 With B n+1 It is the n+1th inflection point of the acceleration dimension and the jerk dimension, where n and n+1 are integers.

6. The low signal-to-noise ratio inverse synthetic aperture radar imaging method according to claim 1, wherein: In step three, the signal component of the scattering point corresponding to the particle with the largest amplitude obtained in step two is removed from the one-dimensional high-resolution range image by subtracting the signal component of the scattering point corresponding to the particle with the largest amplitude from the one-dimensional high-resolution range image obtained in step one.

7. The low signal-to-noise ratio inverse synthetic aperture radar imaging method according to claim 1, wherein: In step 4, the threshold is set to 1% of the initial energy of the one-dimensional high-resolution range image.

8. The low signal-to-noise ratio inverse synthetic aperture radar imaging method according to claim 1, wherein: In step 4, the motion parameters include the equivalent slant range constant R from the radar to the i-th scattering point 0i , the equivalent velocity v of the i-th scattering point Si and the equivalent acceleration a of the i-th scattering point Si .

9. The low signal-to-noise ratio inverse synthetic aperture radar imaging method according to claim 8, wherein: In step 5, the radar image is constructed based on the motion parameters obtained in step 4 as follows: first, the equivalent velocity v of the i-th scattering point is calculated. Si and the equivalent acceleration a of the i-th scattering point Si The Doppler frequency history of the i-th scattering point is calculated, and then the equivalent slant range constant and Doppler frequency of all scattering points at any moment are used as the coordinate information of the corresponding scattering point to obtain the radar image.

Images