A method for extracting and imaging target micro-motion features using non-coherent signal radar

Abstract

The invention discloses a method for extracting and imaging micro-motion features of non-coherent signal radar targets. The invention analyzes the influence of Doppler sensitivity of linear frequency modulation step signal on high-resolution imaging of targets, namely, the range movement and range broadening effects caused by target speed; utilizes the range movement and range broadening effects, and constructs a parameterized sparse representation dictionary containing unknown velocity vectors of multiple scattering points of the target and unknown phases of the signal under non-coherent signal conditions, and establishes a joint optimization model of target speed, signal phase and high-resolution one-dimensional range image of the target; solves the established optimization model to achieve accurate estimation of the moving speed of target scattering points; derives a mathematical expression of the scattering point speed according to the target movement mode, and fits the target scattering point speed to obtain the target micro-motion feature extraction result and high-resolution imaging result. The method can realize the extraction and imaging of micro-motion features of non-coherent signal radar targets and improve the overall working performance of the radar.

Description

Technical Field

[0001] The present invention relates to the technical field of signal and information processing, and in particular to a method for extracting and imaging micro-motion features of a non-coherent signal radar target. Background Art

[0002] Radar targets often have complex micro-motions while moving horizontally, such as rotation, precession, tumbling, and swinging. Target micro-motions will produce frequency modulation on the radar echo signal, thereby generating a physical phenomenon of Doppler spectrum sidebands about the moving target body, which is called the "micro-Doppler effect". On the one hand, the micro-Doppler effect will reduce the performance of the traditional range Doppler imaging algorithm, affect the target imaging quality, and thus reduce the target recognition performance based on the imaging results. On the other hand, the micro-Doppler effect reflects the instantaneous characteristics of the Doppler frequency shift, which contains the unique micro-motion feature information of the target. By analyzing the micro-Doppler effect and extracting the target micro-motion features, it can provide an important feature basis for target classification and recognition that is independent of prior information, highly reliable, and well separable. At present, the classification and recognition technology of moving targets based on micro-motion feature extraction has been widely studied in modern radar technology. In particular, the wideband radar can better distinguish the micro-Doppler modulation characteristics of each scattering point by utilizing the high-resolution capability of the wideband radar in the range direction, so that it can theoretically have a stronger ability to extract fine micro-motion features than narrowband radar. At the same time, it can also achieve high-resolution imaging of the target based on the micro-motion feature extraction results. In "Empirical mode decomposition of micro-Doppler signature" (IEEE International Radar Conference, Arlington, 2005, 895-899), the empirical mode decomposition method was used in micro-Doppler signal analysis and feature extraction, and the effective estimation of the vehicle surface vibration frequency was achieved. "Micro-Doppler feature extraction for wideband imaging radar based on complex image orthogonal matching pursuit decomposition" (IET Radar, Sonar and Navigation, 2013, 7 (8): 914-924) further studied the micro-Doppler effect of targets in wideband radar, and proposed a micro-motion feature extraction and imaging method based on complex image orthogonal matching pursuit decomposition by utilizing the sparse characteristics of radar echoes.In “Micro-motion feature extraction of spatial target based on track-before-detect” (Journal of Sensors, 2017, volume 2017, Article ID 8723042:1–14), a method for micro-motion feature extraction and imaging of spatial rotationally symmetric targets was proposed. The detection, tracking and real-time extraction of micro-motion features of the target were realized at the same time. On this basis, the target scattering distribution was reconstructed, and finally high-precision target micro-motion feature parameters and imaging results were obtained.

[0003] However, the existing research on the extraction and imaging of target micro-motion features of broadband radar is based on coherent signals. In practice, due to the high-speed motion of space targets, it is difficult for broadband radar to achieve long-term coherent accumulation, which brings great difficulties to the extraction and imaging of micro-motion features. Since the micro-Doppler effect is essentially caused by the echo phase change caused by the target micro-motion, an important prerequisite for analyzing the radar echo micro-Doppler effect and extracting the target micro-motion features is that the pulses emitted by the radar maintain phase coherence, that is, the radar is required to be a coherent system. In the case of non-coherence, many existing micro-Doppler effect analysis methods and micro-motion feature extraction technologies, such as time-frequency analysis, Chirplet decomposition, sine frequency modulation Fourier transform, etc., are difficult to apply directly. In broadband radar, even if the radar is a coherent radar, the coherence of the echo signal will be destroyed during the echo motion compensation process due to the limitation of compensation accuracy, thus transforming into an incoherent signal. For example, when pulse compression is performed, the coherence of the echo signal will be seriously damaged when the echo crosses multiple range gates due to the high-speed movement of the space target and the change of the reference distance is not accurately compensated. Therefore, if the effective extraction and high-resolution imaging of target micro-motion features in broadband radar are to be achieved, it is urgent to study the extraction and imaging methods of target micro-motion features in non-coherent signal radar. Summary of the invention

[0004] The purpose of the present invention is to overcome the above-mentioned shortcomings and provide a method for extracting and imaging micro-motion features of non-coherent signal radar targets, which can realize the extraction and imaging of micro-motion features of non-coherent signal radar targets and improve the overall working performance of the radar.

[0005] In order to achieve the above object, the present invention comprises the following steps:

[0006] S1, analyze the impact of the linear frequency modulation step signal SFCS Doppler sensitivity on the target high-resolution imaging, and determine the range movement and range broadening effects caused by the target speed;

[0007] S2, using the range walk and range broadening effects, under the condition of signal incoherence, a parameterized sparse representation dictionary containing the unknown velocity vectors of multiple scattering points of the target and the unknown phase of the signal is constructed, and a joint optimization model of the target velocity, signal phase and target high-resolution one-dimensional range profile HRRP is established;

[0008] S3, solving the joint optimization model of target speed, signal phase and target HRRP to estimate the moving speed of the target scattering point;

[0009] S4, derive the mathematical expression of the scattering point velocity according to the target movement mode, fit the target scattering point velocity, and obtain the target micro-motion feature extraction result and high-resolution imaging result.

[0010] In S1, in the inverse synthetic aperture radar ISAR imaging, the radar transmission signal adopts a linear frequency modulation step signal composed of several linear frequency modulation sub-pulses;

[0011] Perform dechirp processing on the target echo signal;

[0012] After dechirp processing of the target echo signal, analysis of the three phase terms yields:

[0013] The first term is the Doppler term; the second term is the residual video phase (RVP) term unique to dechirp processing; and the third term is the echo envelope tilt term.

[0014] Effectively compensate the phases of the residual video phase RVP item and the echo envelope tilt item, remove the residual video phase RVP item and the echo envelope tilt item, and obtain the target coarse resolution one-dimensional range image CRRP;

[0015] After performing secondary sampling and Fourier transformation on the target CRRP and determining the peak position of the target HRRP, the three mathematical expressions that affect the HRRP peak are analyzed and obtained:

[0016] The first item is used to reflect the real position information of the target scattering point; the second item is the HRRP range movement item caused by the radial motion of the target; and the third item is the range broadening item caused by the radial motion of the target.

[0017] The specific methods for analyzing the distance walking effect and the distance broadening effect are as follows:

[0018] Ignoring the range broadening effect, the peak position of the target scattering point in HRRP is obtained;

[0019] Determine the distance coordinates and distance travel of the target scattering point;

[0020] Determine the conditions under which no winding occurs at the scattering point;

[0021] Determine the range of target motion speed when the target scattering point does not wrap around;

[0022] When the target moving speed exceeds the value range, the position of the target scattering point in HRRP is determined;

[0023] Qualitatively analyze the range-broadening effect and determine the signal parameters that affect the degree of range-broadening.

[0024] The specific method of S2 is as follows:

[0025] The radar observation area is discretized into a number of distance units along the distance direction with the preset distance resolution as the unit increment;

[0026] Combined with the representation of the target echo observation vector, the unknown velocity vector and the unknown phase of the signal at each scattering point are taken as the parameters to be optimized, and the mathematical expression of the parameterized sparse representation dictionary is derived;

[0027] Under the theoretical framework of compressed sensing, a joint optimization model of target speed, signal phase and target HRRP is established.

[0028] The specific methods of S3 are as follows:

[0029] S31, given the initial target speed estimate and phase estimate, initialize the number of iterations, and set the iteration convergence threshold;

[0030] S32, constructing a parameterized sparse representation dictionary based on the current target speed estimate and phase estimate;

[0031] S33, using the orthogonal matching pursuit (OMP) algorithm to obtain the reconstruction result of the target HRRP;

[0032] S34, using the least square method to update the target motion speed and phase estimation value;

[0033] S35, calculating the estimated parameter increment.

[0034] If the iteration termination condition |Δp is met k |<η, Δpk is the estimation parameter increment, η is the iterative convergence threshold, the iteration stops, and the final estimated value of the target motion speed, the final estimated value of the target phase and the final reconstruction result of the target HRRP are obtained; otherwise, go to S32 to continue iteratively solving the motion speed estimation value, the phase estimation value and the target HRRP reconstruction result.

[0035] The specific method of S4 is as follows:

[0036] Assume that the radar is located at the origin O of the coordinate system OXYZ;

[0037] According to the target motion mode, determine the relationship between the velocity of each scattering point and the slow time;

[0038] The least squares method is used to fit the velocity information of the scattering points to obtain the target micro-motion feature extraction results and high-resolution imaging results.

[0039] Compared with the prior art, the present invention analyzes the influence of the Doppler sensitivity of the linear frequency modulation step signal on the high-resolution imaging of the target, that is, the range movement and range broadening effect caused by the target speed; utilizes the range movement and range broadening effect, under the condition of signal incoherence, constructs a parameterized sparse representation dictionary containing the unknown velocity vectors of multiple scattering points of the target and the unknown phase of the signal, and establishes a joint optimization model of the target speed, signal phase and target HRRP; solves the established optimization model to achieve accurate estimation of the target scattering point movement speed; derives the mathematical expression of the scattering point speed according to the target movement mode, fits the target scattering point speed, and obtains the target micro-motion feature extraction result and high-resolution imaging result. This method can realize the micro-motion feature extraction and imaging of non-coherent signal radar targets and improve the overall working performance of the radar. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 is a flow chart of the present invention;

[0041] Figure 2 is the velocity estimation result of the target scattering point in the present invention; wherein (a) is scattering point 1, (b) is scattering point 2, and (c) is scattering point 3;

[0042] Figure 3 It is the target micro-motion feature extraction result in the present invention;

[0043] Figure 4 This is the target imaging result in the present invention. DETAILED DESCRIPTION

[0044] The present invention will be further described below in conjunction with the accompanying drawings.

[0045] The present invention comprises the following steps:

[0046] Step 1: Analyze the impact of the Doppler sensitivity of the linear frequency modulation stepped signal (SFCS) on the high-resolution imaging of the target, that is, the range movement and range broadening effects caused by the target speed;

[0047] Step 2: Using the range walk and range broadening effects, under the condition of signal incoherence, a parameterized sparse representation dictionary containing the unknown velocity vectors of multiple scattering points of the target and the unknown phase of the signal is constructed, and a joint optimization model of the target velocity, signal phase and target high-resolution one-dimensional range profile (HRRP) is established;

[0048] Step 3: Solve the optimization model established in the second step to achieve accurate estimation of the moving speed of the target scattering point;

[0049] Step 4: Derive the mathematical expression of the scattering point velocity according to the target movement mode, fit the target scattering point velocity, and obtain the target micro-motion feature extraction result and high-resolution imaging result.

[0050] The specific steps to implement the above method are as follows:

[0051] Step 1: Assume that the SFCS signal consists of N linear frequency modulated sub-pulses. The expression of the i-th sub-pulse signal in each pulse train is:

[0052]

[0053] Where T1 is the sub-pulse width, μ=B1 / T1 is the sub-pulse modulation frequency, B1 is the sub-pulse bandwidth, T r is the sub-pulse repetition period, f c is the starting carrier frequency of the pulse train, Δf is the carrier frequency step value, f c +iΔf is the carrier frequency of the ith sub-pulse, and t is the fast time, i.e., the time within the pulse train.

[0054] In inverse synthetic aperture radar (ISAR) imaging, high-resolution imaging of the target range can be achieved by performing range-direction pulse compression processing on the target echo signal. Assuming that the target consists of P scattering points, based on the "stop-go" model, the target echo of the i-th sub-pulse in the SFCS signal can be expressed as

[0055]

[0056] Where c is the speed of light, σ p is the scattering coefficient of the pth scattering point, R p is the distance from the pth scattering point to the radar, i = 0, 1, ..., N-1. Suppose the reference signal is

[0057]

[0058] Among them, R ref is the distance from the reference point to the radar, i=0,1,…,N-1, T ref is the pulse width of the reference signal, which is slightly larger than T1. The carrier frequency of the reference signal is the same as that of the transmitted signal.

[0059] Performing “dechirp” processing on the target echo signal, we can get

[0060]

[0061] Where i = 0, 1, ..., N-1, R Δp =R p -R refrepresents the distance from the pth scattering point to the reference point. Let t′=t-iT r -2R ref / c, and Fourier transform of equation (4) with respect to t′, we can get

[0062]

[0063] Where i = 0, 1, ..., N-1, f is the frequency domain representation corresponding to t′.

[0064] Analysis of the three phase terms in equation (5) shows that the first term is the normal Doppler term; the second term is the residual video phase (RVP) term unique to dechirp processing, which will cause changes in the Doppler value; the third term is the echo envelope tilt term, which will cause inconvenience to subsequent imaging processing. Therefore, it is necessary to effectively compensate the phase of the RVP term and the envelope tilt term during imaging processing. Multiply equation (5) by exp(-jπf 2 / μ) can simultaneously remove the RVP term and the envelope tilt term, thereby obtaining the target coarse resolution one-dimensional range profile (CRRP)

[0065]

[0066] It can be seen from formula (6) that |S c The peak value of (f,i)| occurs at f = -2μR Δp / c. The range resolution of the target CRRP is determined by the sub-pulse bandwidth, denoted as ρ c =c / 2B1. Usually, when the target spans multiple coarse resolution range units (i.e., the radial length of the target is greater than ρ c ), due to the energy leakage caused by scattering points in different unambiguous range intervals, the SFCS signal high-resolution range image synthesis based on digital signal processing methods will produce false scattering points in the target HRRP. However, the size of space targets is usually small, and the radar transmission signal parameters can be optimized to make the target radial length less than the length of the unambiguous range interval. At this time, the target CRRP appears in the form of a single peak, and the sampling of the corresponding range unit contains all the information of all scattering points on the target.

[0067] make For S c (f,i) is subsampled, we can get

[0068]

[0069] Perform Fourier transform on equation (7) with respect to i, and we have

[0070]

[0071] Among them, Ff is the frequency domain expression of i.

[0072] It can be seen from formula (8) that |S H (F f The peak value of )| appears at F f =-2ΔfR Δp / c, which is the HRRP of the target. The range resolution is determined by the equivalent synthetic bandwidth NΔf, denoted as ρ h =c / 2NΔf.

[0073] The above SFCS signal high-resolution imaging is based on the "stop-go" model, which does not consider the phase modulation effect of target motion on the echo of each sub-pulse in the pulse train. However, the SFCS signal is a Doppler-sensitive signal. When synthesizing the target HRRP, it can no longer be simply assumed that the distance from the scattering point to the reference point remains unchanged within a pulse train time. Instead, the influence of the target radial velocity v needs to be considered. For the i-th sub-pulse, R in equation (6) is replaced by Δp Expressed as

[0074] R Δp (i) = R Δp +iT r v (9)

[0075] Rewrite equation (6) as

[0076]

[0077] Where i = 0, 1, ..., N-1. Usually, within the time of a pulse train, the displacement caused by the radial motion of the target will not exceed one coarse resolution range unit. Therefore, equation (10) can be further expressed as

[0078]

[0079] Where i = 0, 1, ..., N-1, and equation (11) is subsampled and Fourier transformed, and the peak position of the synthesized target HRRP is

[0080]

[0081] The three terms on the right side of equation (12) are analyzed: the first term reflects the real position information of the target scattering point; the second term is the HRRP range movement term caused by the target radial motion. It can be seen that the greater the target motion speed, the more obvious the range movement of the target scattering point in the HRRP; the third term is the range broadening term caused by the target radial motion. Due to the coupling between the speed v and i in this term, when the target CRRP secondary sampling value is Fourier transformed to synthesize the HRRP, the range image peak corresponding to the target scattering point will be broadened. In the following, the range movement effect and the range broadening effect are analyzed respectively.

[0082] First, ignoring the range broadening effect, the peak position of the target scattering point in HRRP can be expressed as

[0083]

[0084] Divide both sides of the above formula by -2Δf / c to obtain the distance coordinates of the target scattering point. for

[0085]

[0086] where f c T p v / B is the distance traveled by the target scattering point, which is related to the target motion speed v and the signal starting carrier frequency f. c , pulse train width T p (T p =NT r ) and inversely proportional to the equivalent synthesis bandwidth B (B = NΔf), and has nothing to do with the number of sub-pulses, sub-pulse parameters, carrier frequency step value and other details of the signal. It should be noted that according to the frequency domain unambiguous interval of Fourier transform, when F f When the value of ′ exceeds the range of [-1 / 2,1 / 2], the peak value of the target scattering point will wrap around. Therefore, the condition for the scattering point not to wrap around is

[0087]

[0088] According to formula (15), the range of target motion speed when the target scattering point does not wrap can be calculated:

[0089]

[0090] When the target moving speed exceeds the above range, the position of the target scattering point in HRRP is

[0091]

[0092] Where mod(a,b) is the remainder operation.

[0093] As for the range stretching effect, since the range stretching term shown in equation (12) includes the target motion speed v, SFCS signal parameters Δf, and T r It is difficult to give a specific mathematical expression for the range broadening due to coupling with i, so only a simple qualitative analysis is performed. The range broadening is caused by the radial motion of the target, and the degree of broadening is determined by the displacement of the target during the pulse train time and the range high resolution ρ h The ratio of vT p / ρ h =2vT p The larger the B / c, the more obvious the range broadening. It can be seen that the range broadening degree is related to the signal starting carrier frequency f c It has nothing to do with the pulse train width T p , is proportional to the equivalent synthesis bandwidth B. p Under certain conditions of A and B, it is independent of the sub-pulse detail parameters.

[0094] Step 2: From the analysis of the first step, it can be seen that although the range walk, range wrap and range stretching effects will affect the target imaging performance, they are caused by target motion and can be used to obtain target motion information.

[0095] make By resampling the target CRRP shown in formula (11), we can get

[0096]

[0097] Considering the incoherence of the signal and adding the influence of the random unknown signal phase θ, equation (11) can be re-expressed as

[0098]

[0099] The target echo observation vector is denoted as S c =[S c (0),...,S c (i),...,S c (N-1)] T , where T represents the transpose operation. h As unit increment, the radar observation area is discretized into Q distance units along the distance direction, denoted as R = [R1, ..., R q ,...,R Q ], where R q =R1+(q-1)·ρ h ,q=1,2,...,Q. Formula (19) can be expressed as

[0100] S c =D(v,θ)σ+E,σ=[σ1,...,σq ,...,σ Q ] T (20)

[0101] in

[0102] D(v,θ)=[d1,...,d q ,...,d Q ]

[0103]

[0104] E is the noise, σ is the target HRRP, v q is the velocity of the scattering point in the qth distance unit, v = [v1, ..., v q ,...,v Q ] represents the velocity vector composed of the movement speed of the scattering points in each distance unit.

[0105] Usually, most of the energy of the target echo is contributed by only a few scattering centers, so σ is sparse and can be reconstructed by solving an optimization problem:

[0106] σ=argmin||σ||0 stS c =D(v,θ)σ (22)

[0107] Among them, ||·||0 means taking the zero norm.

[0108] The above formula is a typical compressed sensing signal reconstruction model, and the orthogonal matching pursuit (OMP) algorithm can be used to achieve signal reconstruction. Obviously, only when the velocity vector v and phase θ in the parameterized sparse representation dictionary D(v,θ) are the same as the true value, can the dictionary D(v,θ) and the echo observation vector S be consistent. c The high degree of match between them can avoid the range movement and range broadening effect, and reconstruct σ with good focusing. When the motion velocity vector v and phase θ in D(v,θ) do not match the true value, it will cause D(v,θ) and S c Therefore, when v and θ are unknown, it is necessary to comprehensively consider the estimation of v, θ and the reconstruction of σ to establish a joint optimization model.

[0109] {v,θ,σ}=argmin||σ||0 stS c =D(v,θ)σ (23) Step 3: For the above joint optimization problem, v, θ and σ can be iteratively updated and solved. The specific method is as follows:

[0110] Step 1) Given the initial target velocity vector estimate v0 and phase estimate θ0, initialize the number of iterations k = 0, and set the iteration convergence threshold η;

[0111] Step 2) Based on the current v k and θ k , according to formula (21), a parameterized sparse representation dictionary D(v k ,θ k ), where v k and θ k is the estimated value of the target motion velocity vector and phase at the kth iteration;

[0112] Step 3) Use the OMP algorithm to solve equation (22) and obtain the reconstruction result σ of the target HRRP at the kth iteration: k ;

[0113] Step 4) Based on the σ obtained in the previous step k , update the target motion velocity vector and phase estimation value, and record the current velocity vector and phase joint estimation value as p k =[v k ,θ k ], define the joint estimate update value of velocity and phase in the kth iteration as p k+1 =[v k+1 ,θ k+1 ], then

[0114] p k+1 =[v k+1 ,θ k+1 ]=argmin||S c -D(v k ,θ k )σ k || (24)

[0115] Step 5) Calculate the estimated parameter increment Δp k =p k+1 -p k , increase the number of iterations k = k + 1, if the iteration termination condition |Δp is met k |<η, the iteration stops and the final estimated value of the target motion velocity vector, the final estimated value of the phase and the final reconstruction result of the target HRRP are obtained; otherwise, go to Step 2) and continue to iteratively update v, θ and σ.

[0116] Next, the specific solution process of the above Step 4) is given.

[0117] (i) For D(v,θ) in (v k ,θ k ) is Taylor expansion

[0118]

[0119] Where Δv q represents the velocity increment of the scattering point in the qth distance unit, Δθ is the phase increment, and v q,k is the velocity of the scattering point in the qth distance unit at the kth iteration. and It can be calculated as

[0120]

[0121]

[0122]

[0123]

[0124] (ii) k As a known quantity, substitute equation (26) into equation (24) and transform the parameter estimation problem into

[0125]

[0126] Where Δv k and Δθ k Represents the target velocity vector increment and signal phase increment at the kth iteration.

[0127] (iii) Definition

[0128]

[0129] The solution of equation (27) is

[0130]

[0131] in,(·) -1 Represents a matrix inversion operation.

[0132] (iv) Calculate the updated value of the joint estimate of velocity vector and phase

[0133] p k+1 =p k +Δp k (30)

[0134] By iteratively updating v, θ, and σ, accurate estimation of v and θ and accurate reconstruction of σ can be achieved. So far, we have obtained the velocity information of each scattering point. Using the above method, at each slow time t m The velocity information of the scattering points in each distance unit of the target can be obtained at any time, which is recorded as v(t m)=[v1(t m ),...,v q (t m ),...,v Q (t m )], where v q (t m ) is t m The velocity of the scattering point within the qth distance unit at time.

[0135] Step 4: According to the target movement mode, the scattering point velocity information is fitted by least squares, so as to realize the extraction and imaging of target micro-motion characteristic parameters.

[0136] Assume that the radar is located at the origin O of the coordinate system OXYZ. The target translation velocity is expressed as v G =[v X ,v Y ,v Z ] T , where v X 、v Y and v Z are the speeds of the target in the X-axis, Y-axis and Z-axis directions respectively. At the same time, the target moves at an angular velocity of ω = [ω X ,ω Y ,ω Z ] T Spin around the target center, where ω X ,ω Y and ω Z are the angular velocities of the target in the X, Y and Z directions respectively. When the observation starts, the target center is located at (X o ,Y o ,Z o ), slow time t m At time q, the velocity of the qth scattering point can be expressed as

[0137]

[0138] Among them, Ω=||ω||, ε is the angle between the radar line of sight and ω, is the distance from the qth scattering point to the target center, is the initial phase, and ||·|| represents the modulo operation.

[0139] According to the above expression of scattering point velocity, the least square method is used to fit the scattering point velocity information to obtain Ω,r,d, of which is the rotation radius, which means that the target micro-motion characteristic parameter extraction is realized, and the target scattering distribution information is obtained at the same time, and then the target high-resolution imaging result is obtained.

[0140] like Figure 1 As shown, after the above four steps, the non-coherent signal radar target micro-motion feature extraction and imaging can be achieved.

[0141] Example: Non-coherent signal radar target micro-motion feature extraction and imaging processing

[0142] Experimental parameters: Radar transmits linear frequency modulation step signal: f c =35GHz, T r =93.75μs, Δf=4.6875MHz, N=64, B=300MHz, B1=4.6875MHz, T1=2.93μs, T p = 6ms. The target center is located in the OXYZ coordinate system (X o ,Y o ,Z o )=(3,4,5)km, the target translation speed is v G =[300,200,400] T m / s, angular velocity ω=(π,2π,π) T rad / s. There are three scattering points on the target, and their coordinates relative to the center of the target are (1,0,0), (3,1.5,1.5) and (-3,-1.5,-1.5), respectively, in meters. Gaussian white noise with SNR = 5dB is added to the echo, and the target is detected non-coherently under different slow time conditions (the slow time interval is 0.05s), and the movement speed of each scattering point is estimated. The results are as follows: Figure 2 As shown in the figure, it can be seen that the estimated values ​​of the motion speed of the three scattering points on the target are very consistent with the true values, and the mean square error is only 0.017%. On this basis, the speed of each scattering point is fitted according to the target motion mode, and the micro-motion characteristic parameters of the target scattering point can be obtained, such as Figure 3 As shown, the target imaging result is obtained, such as Figure 4 shown.

[0143] The present invention first analyzes the influence of the Doppler sensitivity of the linear frequency modulation step signal on the high-resolution imaging of the target, that is, the range movement and range broadening effects caused by the target speed. On this basis, using the range movement and range broadening effects, under the condition of signal incoherence, a parameterized sparse representation dictionary containing the unknown velocity vectors of multiple scattering points of the target and the unknown phase of the signal is constructed, and a joint optimization model of the target velocity, signal phase and target HRRP is established and solved to achieve accurate estimation of the motion velocity of the target scattering points, and finally obtain the target micro-motion feature extraction results and high-resolution imaging results through parameter fitting.

Claims

1. A method for extracting and imaging micro-motion features of non-coherent signal radar targets, characterized in that: The following steps are involved: S1, analyze the impact of the linear frequency modulation step signal SFCS Doppler sensitivity on the target high-resolution imaging, and determine the range movement and range broadening effects caused by the target speed; S2, using the range walk and range broadening effects, under the condition of signal incoherence, a parameterized sparse representation dictionary containing the unknown velocity vectors of multiple scattering points of the target and the unknown phase of the signal is constructed, and a joint optimization model of the target velocity, signal phase and target high-resolution one-dimensional range profile HRRP is established; S3, solve the joint optimization model of target velocity, signal phase and target high-resolution one-dimensional range profile HRRP to estimate the moving velocity of the target scattering point; S4, derive the mathematical expression of the scattering point velocity according to the target movement mode, fit the target scattering point velocity, and obtain the target micro-motion feature extraction result and high-resolution imaging result.

2. The method for extracting and imaging target micro-motion features of a non-coherent signal radar according to claim 1, characterized in that: In S1, in the inverse synthetic aperture radar ISAR imaging, the radar transmission signal adopts a linear frequency modulation step signal composed of several linear frequency modulation sub-pulses; Perform dechirp processing on the target echo signal; Effectively compensate the phases of the residual video phase RVP item and the echo envelope tilt item, remove the residual video phase RVP item and the echo envelope tilt item, and obtain the target coarse resolution one-dimensional range image CRRP; Sub-sampling and Fourier transform are performed on the target coarse resolution one-dimensional range image CRRP to determine the peak position of the target high resolution one-dimensional range image HRRP; The distance walking effect and distance broadening effect are analyzed separately.

3. The method for extracting and imaging target micro-motion features of a non-coherent signal radar according to claim 2, characterized in that: After dechirp processing of the target echo signal, analysis of the three phase terms yields: The first term is the Doppler term; the second term is the residual video phase term unique to the dechirp process; and the third term is the echo envelope tilt term.

4. The method for extracting and imaging target micro-motion features of a non-coherent signal radar according to claim 2, characterized in that: After performing secondary sampling and Fourier transformation on the target coarse resolution one-dimensional range image CRRP and determining the peak position of the target high resolution one-dimensional range image HRRP, the three mathematical expressions that affect the peak of the high resolution one-dimensional range image HRRP are analyzed and obtained: The first item is used to reflect the real position information of the target scattering point; the second item is the high-resolution one-dimensional range image HRRP range movement item caused by the radial motion of the target; and the third item is the range broadening item caused by the radial motion of the target.

5. The method for extracting and imaging target micro-motion features of non-coherent signal radar according to claim 2, characterized in that: The specific methods for analyzing the distance walking effect and the distance broadening effect are as follows: Ignoring the range broadening effect, the peak position of the target scattering point in the high-resolution one-dimensional range image HRRP is obtained; Determine the distance coordinates and distance travel of the target scattering point; Determine the conditions under which no winding occurs at the scattering point; Determine the range of target motion speed when the target scattering point does not wrap around; When the target moving speed exceeds the value range, the position of the target scattering point in the high-resolution one-dimensional range image HRRP is determined; Qualitatively analyze the range-broadening effect and determine the signal parameters that affect the degree of range-broadening.

6. The method for extracting and imaging target micro-motion features of non-coherent signal radar according to claim 1, characterized in that: The specific method of S2 is as follows: The radar observation area is discretized into a number of distance units along the distance direction with the preset distance resolution as the unit increment; Combined with the representation of the target echo observation vector, the unknown velocity vector and the unknown phase of the signal at each scattering point are taken as the parameters to be optimized, and the mathematical expression of the parameterized sparse representation dictionary is derived; Under the theoretical framework of compressed sensing, a joint optimization model of target velocity, signal phase and target high-resolution one-dimensional range profile HRRP is established.

7. The method for extracting and imaging target micro-motion features of non-coherent signal radar according to claim 1, characterized in that: The specific methods of S3 are as follows: S31, given the initial target speed estimate and phase estimate, initialize the number of iterations, and set the iteration convergence threshold; S32, constructing a parameterized sparse representation dictionary based on the current target speed estimate and phase estimate; S33, using the orthogonal matching pursuit (OMP) algorithm, the reconstruction result of the target high-resolution one-dimensional range image HRRP is obtained; S34, using the least square method to update the target motion speed and phase estimation value; S35, calculating the estimated parameter increment.

8. The method for extracting and imaging target micro-motion features of non-coherent signal radar according to claim 7, characterized in that: If the iteration termination condition |Δp is met k |<η,Δp k is the estimated parameter increment, η is the iterative convergence threshold, the iteration stops, and the final estimated value of the target motion speed, the final estimated value of the target phase and the final reconstruction result of the target high-resolution one-dimensional range image HRRP are obtained; otherwise, go to S32 to continue iteratively solving the motion speed estimation value, the phase estimation value and the target high-resolution one-dimensional range image HRRP reconstruction result.

9. The method for extracting and imaging target micro-motion features of non-coherent signal radar according to claim 1, characterized in that: The specific method of S4 is as follows: Assume that the radar is located at the origin O of the coordinate system OXYZ; according to the target motion mode, determine the relationship between the velocity of each scattering point and the slow time; The least squares method is used to fit the velocity information of the scattering points to obtain the target micro-motion feature extraction results and high-resolution imaging results.

Images