A translation compensation method for high maneuvering targets under low signal-to-noise ratio conditions
Through the distance instantaneous Doppler imaging model based on synchronous extraction transform and the SAO-KOA hybrid optimization algorithm, the problem of difficulty in translation compensation of high maneuver targets under low signal-to-noise ratio is solved, high-quality ISAR imaging is achieved, and imaging is improved Resolution and compensation accuracy.
Patent Information
- Application Number
- CN202410613381.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-16
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-05-16
AI Technical Summary
Under low signal-to-noise ratio conditions, the existing cascade translation compensation and combined translation compensation methods cannot effectively compensate for the translation motion of high-motor targets, resulting in a decrease in ISAR imaging quality and making high-resolution imaging impossible.
A distance instantaneous Doppler imaging model based on synchronous extraction transformation is adopted, and a SAO-KOA hybrid optimization algorithm is used to perform translation compensation using the average image entropy to build an accurate translation compensation model to improve imaging resolution.
Accurately compensate for the translational motion of high maneuverable targets under low signal-to-noise ratio conditions, improve ISAR imaging quality, realize high-resolution imaging, and enhance the effectiveness and applicability of the translational compensation algorithm.
Smart Images

Figure CN118501877B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the field of radar signal processing, and particularly relates to a translation compensation method for a maneuvering target in inverse synthetic aperture radar imaging, and more particularly to a translation compensation method for a high maneuvering target under low signal-to-noise ratio conditions. Background Art
[0002] Inverse Synthetic Aperture Radar (ISAR) plays an important role in the civil and military fields due to its ability to perform high-resolution imaging of targets in all weather conditions and at long distances. Since the targets observed by ISAR are generally non-cooperative targets, the movement of the targets cannot be known, so motion compensation is required.
[0003] The motion of the target relative to the radar can be decomposed into two parts: translation and rotation. Rotation is the source of imaging, and translation will have two effects on imaging: one is the misalignment of the range unit, and the signal of the same scattering point is distributed in different units; the other is the introduction of phase error, which makes the ISAR imaging result defocused. Therefore, in order to achieve high-quality ISAR imaging, high-precision translation compensation is required.
[0004] Commonly used translation compensation includes two categories: cascaded translation compensation and joint translation compensation. Cascaded translation compensation is divided into two steps: envelope alignment and self-focusing. Although this method is simple and effective, it has certain requirements for the signal-to-noise ratio of the echo. Usually, ISAR imaging processing requires the signal-to-noise ratio of the echo to be higher than 10dB. When the signal-to-noise ratio is lower than 0dB, that is, it is in a low signal-to-noise ratio condition, the cascade method may fail due to the strong noise in the range image. Joint translation compensation realizes envelope alignment and self-focusing at the same time, and has certain noise resistance. When ISAR performs imaging processing on high-maneuvering targets such as fighters with rapid attitude changes, ships swaying violently with waves, space missiles, and drones, the target rotation speed can no longer be considered constant, the azimuth Doppler will change with time, and the image will be defocused and blurred. At this time, the motion parameters and the changes in image entropy no longer correspond. The target motion parameters obtained by the parameterized translation method may have a large error with the actual parameters, the accuracy of translation compensation is reduced, and high-resolution imaging of the target cannot be achieved. Therefore, for ISAR imaging of highly maneuverable targets under low signal-to-noise ratio conditions, due to the combined effects of low signal-to-noise ratio conditions and high target maneuverability, both cascaded translation compensation and joint translation compensation cannot effectively compensate for the target's translation, and high-quality imaging of the target cannot be achieved, which is not conducive to target feature extraction and recognition. Summary of the invention
[0005] The present invention provides a method for compensating the translation of a highly maneuverable target under low signal-to-noise ratio conditions, which solves the technical problem of the difficulty in compensating the translation of a highly maneuverable target under low signal-to-noise ratio conditions. The purpose of the present invention is to accurately compensate for the translation of a highly maneuverable target, achieve high-resolution ISAR imaging, and improve imaging quality. The present invention fully considers the continuity of target changes and the high maneuverability characteristics, proposes a range instantaneous Doppler imaging model based on synchronized extraction transformation SET, takes the average image entropy of range instantaneous Doppler imaging as the objective function, achieves accurate translation compensation, and improves imaging resolution.
[0006] To achieve the above object, the present invention proposes a method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions, comprising:
[0007] Step 1: Perform Fourier transform on the initial time domain echo of the target, transform it into the frequency domain to perform residual video phase RVP compensation, and obtain the compensated one-dimensional range image of the target; perform inverse Fourier transform on the compensated one-dimensional range image of the target to obtain the time domain echo of the target;
[0008] Step 2: Obtain the reference distance recorded in the radar system information when performing Dechirp processing on the initial time domain echo of the transmitted pulse, and save all the reference distances as a reference distance set;
[0009] Step 3: Fitting the reference distance set to obtain a polynomial coefficient set of the reference distance set, taking the polynomial coefficients of the reference distance set as initial values of the polynomial coefficients of the radial distance, and saving the initial values of all polynomial coefficients of the radial distance as an initial value set; the radial distance is a high-order polynomial;
[0010] Step 4: construct a translation compensation model through time domain echo, radial distance and reference distance; discretize the translation compensation model to obtain a discretized translation compensation model;
[0011] Step 5: Based on the snow ablation optimization SAO algorithm, the positions of all snow ablation optimized SAO individuals are initialized through the initial value set, the SAO radial distances corresponding to the positions of all SAO individuals are input into the discretized translation compensation model to obtain the SAO time domain echoes corresponding to the positions of all SAO individuals, and the SAO time domain echoes corresponding to the positions of all SAO individuals are imaged by the distance instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain SAO distance-Doppler frequency-slow time three-dimensional stereo data, and a preset number of slice data corresponding to the slow time sequence number are selected from the SAO distance-Doppler frequency-slow time three-dimensional stereo data as the SAO distance instantaneous Doppler image; the SAO average image entropy of the SAO distance instantaneous Doppler image is calculated, and the SAO average image entropy is sorted from small to large to construct a SAO sorting queue; the positions of the SAO individuals corresponding to the top 50% of the SAO average image entropy in the SAO sorting queue are selected as the SAO input queue;
[0012] Based on the Kepler optimized KOA algorithm, the positions of all Kepler optimized KOA individuals are initialized through the SAO input queue, the KOA radial distances corresponding to the positions of all KOA individuals are input into the discretized translation compensation model to obtain the KOA time domain echoes corresponding to the positions of all KOA individuals, and the KOA time domain echoes corresponding to the positions of all KOA individuals are imaged and processed through the distance instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain the KOA distance-Doppler frequency-slow time three-dimensional stereo data, and the slice data corresponding to the slow time serial number of a preset number are selected from the KOA distance-Doppler frequency-slow time three-dimensional stereo data as the KOA distance instantaneous Doppler image; the KOA average image entropy of the KOA distance instantaneous Doppler image is calculated; and the position of the KOA individual is updated according to the KOA average image entropy;
[0013] Based on the SAO average image entropy and the KOA average image entropy, the SAO individual position and the KOA individual position are sorted from small to large, and the SAO individual position and the KOA individual position are merged to construct a SAO-KOA sorting queue. The smallest individual position in the SAO-KOA sorting queue is the global optimal position; the position of the SAO individual is updated, and the global optimal position in the searched SAO-KOA sorting queue is used as the precise motion parameter of the target;
[0014] Step 6: Input the radial distance corresponding to the precise motion parameters into the discretized translation compensation model to obtain the target time domain echo, and perform imaging processing on the target time domain echo through the distance instantaneous Doppler imaging model based on the synchronous extraction transform SET to obtain the target distance-Doppler frequency-slow time three-dimensional stereo data. Select the slice data corresponding to the slow time number corresponding to the task requirements from the target distance-Doppler frequency-slow time three-dimensional stereo data as the target distance instantaneous Doppler imaging result.
[0015] In step 1, compensate the one-dimensional distance image R g (f r ,t m ) and time domain echo s(t r ,t m ) are:
[0016]
[0017] s(t r ,t m )=IFFT(R g (f r ,t m ));
[0018] Among them, s dc (t r ,t m ) is the initial time domain echo; is the RVP compensation item; t r For quick time, T p is the pulse width, n=1,2,…,N, is the distance unit number, N is the number of distance units, t m For slow time, T CPI is the radar imaging processing time; m = 1, 2, ..., M, is the transmission pulse number; M is the number of azimuth units; γ is the modulation frequency of the radar transmission signal; f r is the distance frequency, f s is the sampling rate; FFT(·) is the Fourier transform; IFFT(·) is the inverse Fourier transform.
[0019] In step 2, the reference distance set R ref for:
[0020] R ref =[R ref (t1),…,R ref (t m ),…,R ref (t M )];
[0021] Among them, R ref (t m ) is the reference distance recorded in the acquired radar system information when performing Dechirp processing on the initial time domain echo of the mth transmission pulse; t m is the slow time, m=1,2,…,M, is the transmission pulse number, and M is the total number of transmission pulses.
[0022] In step 3, the initial value set is:
[0023] For the reference distance set R ref Perform a third-order polynomial fit to obtain the polynomial coefficient set of the reference distance set The expression is as follows:
[0024]
[0025] Where X = [1,…,M], is the discrete slow time set, M is the total number of transmitted pulses; polyfit(·) is the polynomial fitting operation; the polynomial coefficient corresponding to the l-order term of the reference distance set is The polynomial coefficient corresponding to the l-order term of the reference distance set The polynomial coefficient a corresponding to the l-order term of the radial distance l The initial value of all polynomial coefficients of radial distance is saved as the initial value set
[0026] In step 3, the radial distance is:
[0027]
[0028] Among them, R T (t m ) is the radial distance, t m is the slow time; m=1,2,Ω,M, is the transmission pulse number, M is the total number of transmission pulses; l is the different order, l=0,1,2,3, a l is the coefficient corresponding to the l-order term; Δt m is the pulse repetition interval.
[0029] In step 4, the translation compensation model is:
[0030]
[0031] in, is the time domain echo after translation compensation, s(t r ,t m ) is the time domain echo; j is the imaginary unit, c is the speed of light, f c is the radar signal carrier frequency, γ is the modulation frequency of the radar transmission signal, R T (t m ) is the radial distance, R ref (t m ) is the reference distance; t r For quick time.
[0032] In step 4, the discretized translation compensation model is:
[0033]
[0034] in, This is the time domain echo after translation compensation processing The discrete form of s(n,m) is the time domain echo s(t r ,t m ) in discrete form; R T (m) is the radial distance R T (t m ), R ref (m) is the reference distance R ref (t m ) in discrete form; n = 1, 2, ..., N, is the distance unit number.
[0035] The specific steps in step five are:
[0036] 8.1, based on the SAO algorithm, according to the initial value set Determine the upper bound Ua and lower bound La of the polynomial coefficients; initialize the total number of SAO individuals Q, q = 1, 2, 3, ..., Q, which is the serial number of the SAO individual; the initial position A of Q SAO individuals can be obtained through Ua and La, and the initial position A is a matrix with Q rows and 4 columns; set the maximum number of iterations of the SAO algorithm to T max ; The number of iterations of the SAO algorithm is t = 1, 2, ..., T max ;
[0037] 8.2, when t = 1, the position of the qth SAO individual The corresponding SAO radial distance R T (m; a q ) is input into the discretized translation compensation model to obtain the SAO time domain echo for:
[0038]
[0039] in, is the coefficient corresponding to the l-th order term of the SAO radial distance corresponding to the position of the q-th SAO individual;
[0040] 8.3, the method of imaging the SAO time domain echoes corresponding to the positions of all SAO individuals by using the range instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain the SAO range-Doppler frequency-slow time three-dimensional stereo data is:
[0041]
[0042] Among them, RID(n,f,m;a q) is the SAO time domain echo input corresponding to the position aq of the qth SAO individual, and the SAO range-Doppler frequency-slow time three-dimensional data obtained in the range instantaneous Doppler imaging model based on the synchronous extraction transformation SET, f is the frequency sequence number, f = 1, 2, ..., F, F is the maximum frequency sequence number, SET (·) represents the synchronous extraction transformation operation;
[0043] 8.4, 1≤v≤M, v is the preset number; select the slice data corresponding to the v-slow time sequence number from the SAO range-Doppler frequency-slow time three-dimensional data, and the obtained SAO range instantaneous Doppler image is RID(n,f,m vc ; a q ), m vc is the slow time sequence number of the selected image, vc=1,…,v;
[0044] 8.5, the SAO average image entropy En(a) of the instantaneous Doppler image at SAO distance v q )for:
[0045]
[0046] Where Ec(·) represents the image entropy calculation operation;
[0047] 8.6, according to the SAO average image entropy En(a q ) Sorting the positions of the SAO individuals from small to large to construct an SAO sorting queue; selecting the positions of the first 50% of the SAO individuals in the SAO sorting queue as the SAO input queue;
[0048] 8.7, based on the KOA algorithm, initialize the positions of all KOA individuals through the SAO input queue; initialize the total number of KOA individuals H, which is the maximum integer not greater than 0.5Q; h = 1, 2, ..., H, is the sequence number of the KOA individual; set the maximum number of iterations of the KOA algorithm to T' max ; The number of iterations of the KOA algorithm is t′=1,2,…,T′ max ;
[0049] 8.8, when t′=1, the position of the hth KOA individual is The corresponding KOA radial distance is input into the discretized translation compensation model to obtain the KOA time domain echo;
[0050] Among them, the KOA radial distance is:
[0051]
[0052] The KOA time domain echo is:
[0053]
[0054] 8.9, the method of imaging the KOA time domain echoes corresponding to the positions of all KOA individuals by using the range instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain the KOA range-Doppler frequency-slow time three-dimensional stereo data is:
[0055]
[0056] Among them, RID(n,f,m;a h ) is the KOA time domain echo input corresponding to the KOA individual position ah, and the KOA distance-Doppler frequency-slow time three-dimensional data obtained in the distance instantaneous Doppler imaging model based on the synchronous extraction transformation SET; f is the frequency sequence number, f = 1, 2, ..., F, F is the maximum frequency sequence number, SET (·) represents the synchronous extraction transformation operation;
[0057] 8.10, 1≤v≤M, v is the preset number; select the slice data corresponding to the v-slow time sequence number from the KOA range-Doppler frequency-slow time three-dimensional data, and the obtained KOA range instantaneous Doppler image is RID(n,f n ,m vc ; a h ), m vc is the slow time sequence number of the selected image, vc=1,…,v;
[0058] 8.11, the KOA average image entropy of the instantaneous Doppler image of KOA distance v is:
[0059]
[0060] Where Ec(·) represents the image entropy calculation operation;
[0061] 8.12, increase t′ by 1. If the number of iterations of the KOA algorithm reaches the maximum value T′ max , terminate the search process, and sort the SAO individual position and the KOA individual position from small to large based on the SAO average image entropy and the KOA average image entropy, merge the SAO individual position with the KOA individual position, and build a SAO-KOA sorting queue. The smallest individual position in the SAO-KOA sorting queue is the global optimal position. Otherwise, update the position of the KOA individual according to the KOA average image entropy, and repeat 8.8 to 8.12;
[0062] 8.13, increase t by 1. If the number of iterations of the SAO algorithm reaches the maximum value T max , terminate the search process, otherwise, update the position of the SAO individual, repeat 8.2 to 8.13, and finally output the position of the smallest individual in the SAO-KOA sorting queue as the precise motion parameter a fine .
[0063] The specific method in step six includes:
[0064] Precise motion parametersa fine =[a′0,a1′,a′2,a3′]corresponding radial distance R T (m; a fine )for:
[0065]
[0066] The target time domain echo is:
[0067]
[0068] The target distance-Doppler frequency-slow time three-dimensional data is:
[0069]
[0070] According to the task requirements, select RID (n, f, m; a fine ) is used as the target distance instantaneous Doppler imaging result RID(n,f,m ch ; a fine ), m ch is the slow time sequence number of the selected image, m ch =1,2,…,M.
[0071] Beneficial effects of the present invention:
[0072] 1. The present invention constructs a translation compensation model based on the minimum average image entropy, and jointly compensates the translation motion of the target according to the average image entropy, thereby overcoming the influence of low signal-to-noise ratio and high maneuverability on translation compensation, accurately compensating the translation motion of the target, and realizing high-quality ISAR imaging of the target.
[0073] 2. In the imaging process, the present invention proposes a distance instantaneous Doppler imaging model based on synchronous extraction transformation SET, and uses synchronous extraction transformation to perform distance instantaneous Doppler imaging processing. The synchronous extraction transformation has a high time-frequency resolution and improves the quality of target distance instantaneous Doppler imaging.
[0074] 3. The present invention uses the SAO-KOA hybrid optimization algorithm to estimate the target motion parameters, fully combining the advantages of the SAO algorithm and the KOA algorithm, which can effectively avoid the parameter optimization from falling into the local optimal value and realize the global accurate optimization solution of the motion parameters. The algorithm has the advantages of fast convergence speed, strong local search ability and accurate parameter estimation.
[0075] 4. In the process of optimizing target motion parameters, the present invention uses the average image entropy of multiple distance instantaneous Doppler images as the objective function, which effectively reduces the impact of random noise on the translation compensation algorithm and increases the effectiveness and applicability of the translation compensation algorithm. BRIEF DESCRIPTION OF THE DRAWINGS
[0076] Figure 1 It is the target model used in the simulation experiment.
[0077] Figure 2 It is the range image of the target with translation error when SNR = -4dB.
[0078] Figure 3 This is the ideal range Doppler imaging result of the target when SNR = -4dB.
[0079] Figure 4 It is the range Doppler imaging result obtained by the MERA+MEPA translation compensation method when SNR=-4dB.
[0080] Figure 5 It is the distance instantaneous Doppler imaging result obtained based on the STFT translation compensation method when SNR=-4dB.
[0081] Figure 6 It is the distance instantaneous Doppler imaging result obtained based on the translation compensation method of the present invention when SNR=-4dB.
[0082] Figure 7 It is the range image of the target with translation error when SNR=-8dB.
[0083] Figure 8 This is the ideal range Doppler imaging result of the target when SNR=-8dB.
[0084] Fig. 9 It is the range Doppler imaging result obtained by the MERA+MEPA translation compensation method when SNR=-8dB.
[0085] Fig.10 It is the distance instantaneous Doppler imaging result obtained based on the STFT translation compensation method when SNR=-8dB.
[0086] Fig.11 It is the distance instantaneous Doppler imaging result obtained based on the translation compensation method of the present invention when SNR=-8dB. DETAILED DESCRIPTION
[0087] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.
[0088] An embodiment of the present invention provides a method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions, comprising:
[0089] Step 1: Perform Fourier transform on the initial time domain echo of the target, transform it into the frequency domain to perform residual video phase RVP compensation, and obtain the compensated one-dimensional range image of the target; perform inverse Fourier transform on the compensated one-dimensional range image of the target to obtain the time domain echo of the target;
[0090] Step 2: Obtain the reference distance recorded in the radar system information when performing Dechirp processing on the initial time domain echo of the transmitted pulse, and save all the reference distances as a reference distance set;
[0091] Step 3: Fitting the reference distance set to obtain a polynomial coefficient set of the reference distance set, taking the polynomial coefficients of the reference distance set as initial values of the polynomial coefficients of the radial distance, and saving the initial values of all polynomial coefficients of the radial distance as an initial value set; the radial distance is a high-order polynomial;
[0092] Step 4: construct a translation compensation model through time domain echo, radial distance and reference distance; discretize the translation compensation model to obtain a discretized translation compensation model;
[0093] Step 5: Based on the Snow Ablation Optimizer (SAO) algorithm, the positions of all snow ablation optimized SAO individuals are initialized through an initial value set, the SAO radial distances corresponding to the positions of all SAO individuals are input into a discretized translation compensation model to obtain the SAO time domain echoes corresponding to the positions of all SAO individuals, and the SAO time domain echoes corresponding to the positions of all SAO individuals are imaged by a distance instantaneous Doppler imaging model based on a synchronous extraction transform (SET) to obtain SAO distance-Doppler frequency-slow time three-dimensional stereo data, and a preset number of slice data corresponding to the slow time sequence number are selected from the SAO distance-Doppler frequency-slow time three-dimensional stereo data as the SAO distance instantaneous Doppler image; the SAO average image entropy of the SAO distance instantaneous Doppler image is calculated, and the SAO average image entropy is sorted from small to large to construct a SAO sorting queue; the positions of the SAO individuals corresponding to the top 50% of the SAO average image entropy in the SAO sorting queue are selected as the SAO input queue;
[0094] Based on the Kepler Optimization Algorithm (KOA), the positions of all Kepler optimized KOA individuals are initialized through the SAO input queue, the KOA radial distances corresponding to the positions of all KOA individuals are input into the discretized translation compensation model to obtain the KOA time domain echoes corresponding to the positions of all KOA individuals, the KOA time domain echoes corresponding to the positions of all KOA individuals are imaged through the distance instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain the KOA distance-Doppler frequency-slow time three-dimensional stereo data, and the slice data corresponding to the slow time serial number of a preset number are selected from the KOA distance-Doppler frequency-slow time three-dimensional stereo data as the KOA distance instantaneous Doppler image; the KOA average image entropy of the KOA distance instantaneous Doppler image is calculated; and the position of the KOA individual is updated according to the KOA average image entropy;
[0095] Based on the SAO average image entropy and the KOA average image entropy, the SAO individual position and the KOA individual position are sorted from small to large, the SAO individual position is merged with the KOA individual position, and a SAO-KOA sorting queue is constructed. The smallest individual position in the SAO-KOA sorting queue is the global optimal position; the position of the SAO individual is updated, and the global optimal position in the searched SAO-KOA sorting queue is used as the precise motion parameter of the target; the specific method steps disclosed in step five are referred to as the SAO-KOA hybrid optimization algorithm.
[0096] Step 5 of the embodiment of the present invention uses the SAO-KOA hybrid optimization algorithm to optimize and solve the target precise motion parameters, which can effectively avoid the parameter optimization from falling into the local optimal value and realize the global precise optimization solution of the motion parameters. The algorithm has the advantages of fast convergence speed, strong local search ability and accurate parameter estimation.
[0097] Step 6: Input the radial distance corresponding to the precise motion parameters into the discretized translation compensation model to obtain the target time domain echo, and perform imaging processing on the target time domain echo through the distance instantaneous Doppler imaging model based on the synchronous extraction transform SET to obtain the target distance-Doppler frequency-slow time three-dimensional stereo data. Select the slice data corresponding to the slow time number corresponding to the task requirements from the target distance-Doppler frequency-slow time three-dimensional stereo data as the target distance instantaneous Doppler imaging result.
[0098] In step 1, compensate the one-dimensional distance image R g (f r ,t m ) and time domain echo s(t r ,t m ) are:
[0099]
[0100] s(t r ,t m )=IFFT(R g (f r ,t m ));
[0101] Among them, s dc (t r ,t m ) is the initial time domain echo; is the RVP compensation item; t r For quick time, T p is the pulse width, n=1,2,…,N, is the distance unit number, N is the number of distance units, t m For slow time, T CPI is the radar imaging processing time; m = 1, 2, ..., M, is the transmission pulse number; M is the number of azimuth units; γ is the modulation frequency of the radar transmission signal; f r is the distance frequency, f s is the sampling rate; FFT(·) is the Fourier transform; IFFT(·) is the inverse Fourier transform.
[0102] In step 2, the reference distance set R ref for:
[0103] R ref =[R ref (t1),…,R ref (t m ),…,R ref (t M )];
[0104] Among them, R ref (t m ) is the reference distance recorded in the acquired radar system information when performing Dechirp processing on the initial time domain echo of the mth transmission pulse; t m is the slow time, m=1,2,…,M, is the transmission pulse number, and M is the total number of transmission pulses.
[0105] In step 3, the initial value set is:
[0106] For the reference distance set R ref Perform a third-order polynomial fit to obtain the polynomial coefficient set of the reference distance set The expression is as follows:
[0107]
[0108] Where X = [1,…,M] is the discrete slow time set, M is the total number of transmitted pulses; polyfit(·) is the polynomial fitting operation; the polynomial coefficient corresponding to the l-order term of the reference distance set is The polynomial coefficient corresponding to the l-order term of the reference distance set The polynomial coefficient a corresponding to the l-order term of the radial distance l The initial value of all polynomial coefficients of radial distance is saved as the initial value set
[0109] In step 3, the radial distance is:
[0110]
[0111] Among them, R T (t m ) is the radial distance, t m is the slow time; m=1,2,…,M, is the transmission pulse number, M is the total number of transmission pulses; l is the different orders, l=0,1,2,3, a l is the coefficient corresponding to the l-order term; Δt m is the pulse repetition interval. Considering that the imaging time is generally short and the target motion is relatively stable, a third-order polynomial is sufficient to describe the target motion. Therefore, the target motion is modeled as a third-order polynomial.
[0112] In step 4, the translation compensation model is:
[0113]
[0114] in, is the time domain echo after translation compensation, s(t r ,t m ) is the time domain echo; j is the imaginary unit, c is the speed of light, f c is the radar signal carrier frequency, γ is the modulation frequency of the radar transmission signal, R T (t m ) is the radial distance, R ref (t m ) is the reference distance; t r For quick time.
[0115] In step 4, the discretized translation compensation model is:
[0116]
[0117] in, This is the time domain echo after translation compensation processing The discrete form of s(n,m) is the time domain echo s(t r ,tm ) in discrete form; R T (m) is the radial distance R T (t m ), R ref (m) is the reference distance R ref (t m ) in discrete form; n = 1, 2, ..., N, is the distance unit number.
[0118] The specific steps in step five are:
[0119] 8.1, based on the SAO algorithm, according to the initial value set Determine the upper bound Ua and lower bound La of the polynomial coefficients; initialize the total number of SAO individuals Q, q = 1, 2, 3, ..., Q, which is the serial number of the SAO individual; the initial position A of Q SAO individuals can be obtained through Ua and La, and the initial position A is a matrix with Q rows and 4 columns; set the maximum number of iterations of the SAO algorithm to T max ; The number of iterations of the SAO algorithm is t = 1, 2, ..., T max ;
[0120] 8.2, when t = 1, the position of the qth SAO individual The corresponding SAO radial distance R T (m; a q ) is input into the discretized translation compensation model to obtain the SAO time domain echo for:
[0121]
[0122] in, is the coefficient corresponding to the l-th order term of the SAO radial distance corresponding to the position of the q-th SAO individual;
[0123] 8.3, the method of imaging the SAO time domain echoes corresponding to the positions of all SAO individuals by using the range instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain the SAO range-Doppler frequency-slow time three-dimensional stereo data is:
[0124]
[0125] Among them, RID(n,f,m;a q ) is the position a of the qth SAO individual qThe corresponding SAO time domain echo input is based on the SAO range-Doppler frequency-slow time three-dimensional data obtained in the range instantaneous Doppler imaging model of the synchronous extraction transformation SET, where f is the frequency sequence number, f=1,2,…,F, F is the maximum frequency sequence number, and SET(·) represents the synchronous extraction transformation operation;
[0126] 8.4, 1≤v≤M, v is the preset number; select the slice data corresponding to the v-slow time sequence number from the SAO range-Doppler frequency-slow time three-dimensional data, and the obtained SAO range instantaneous Doppler image is RID(n,f,m vc ; a q ), m vc is the slow time sequence number of the selected image, vc=1,…,v;
[0127] 8.5, the SAO average image entropy En(a) of the instantaneous Doppler image at SAO distance v q )for:
[0128]
[0129] Where Ec(·) represents the image entropy calculation operation;
[0130] 8.6, according to the SAO average image entropy En(a q ) Sort the positions of SAO individuals from small to large to construct SAO
[0131] Sorting queue; selecting the positions of the first 50% of SAO individuals in the SAO sorting queue as the SAO input queue;
[0132] 8.7, based on the KOA algorithm, initialize the positions of all KOA individuals through the SAO input queue; initialize the total number of KOA individuals H, which is the maximum integer not greater than 0.5Q; h = 1, 2, ..., H, is the sequence number of the KOA individual; set the maximum number of iterations of the KOA algorithm to T m ' ax ; The number of iterations of the KOA algorithm is t′=1,2,…,T′ max ;
[0133] 8.8, when t′=1, the position of the hth KOA individual is The corresponding KOA radial distance is input into the discretized translation compensation model to obtain the KOA time domain echo;
[0134] Among them, the KOA radial distance is:
[0135]
[0136] The KOA time domain echo is:
[0137]
[0138] 8.9, the method of imaging the KOA time domain echoes corresponding to the positions of all KOA individuals by using the range instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain the KOA range-Doppler frequency-slow time three-dimensional stereo data is:
[0139]
[0140] Among them, RID(n,f,m;a h ) is the KOA individual position a h The corresponding KOA time domain echo input is based on the KOA range-Doppler frequency-slow time three-dimensional data obtained in the range instantaneous Doppler imaging model of the synchronous extraction transformation SET; f is the frequency sequence number, f = 1, 2, ..., F, F is the maximum frequency sequence number, SET (·) represents the synchronous extraction transformation operation;
[0141] 8.10, 1≤v≤M, v is the preset number; select the slice data corresponding to the v-slow time sequence number from the KOA range-Doppler frequency-slow time three-dimensional data, and the obtained KOA range instantaneous Doppler image is RID(n,f n ,m vc ; a h ), m vc is the slow time sequence number of the selected image, vc=1,…,v;
[0142] 8.11, the KOA average image entropy of the instantaneous Doppler image of KOA distance v is:
[0143]
[0144] Where Ec(·) represents the image entropy calculation operation;
[0145] 8.12, increase t′ by 1. If the number of iterations of the KOA algorithm reaches the maximum value T m ' ax , terminate the search process, and sort the SAO individual position and the KOA individual position from small to large based on the SAO average image entropy and the KOA average image entropy, merge the SAO individual position with the KOA individual position, and build a SAO-KOA sorting queue. The smallest individual position in the SAO-KOA sorting queue is the global optimal position. Otherwise, update the position of the KOA individual according to the KOA average image entropy, and repeat 8.8 to 8.12;
[0146] 8.13, increase t by 1. If the number of iterations of the SAO algorithm reaches the maximum value T max, terminate the search process, otherwise, update the position of the SAO individual, repeat 8.2 to 8.13, and finally output the position of the smallest individual in the SAO-KOA sorting queue as the precise motion parameter a fine .
[0147] The specific method in step six includes:
[0148] Precise motion parametersa fine =[a′0,a1′,a′2,a3′]corresponding radial distance R T (m; a fine )for:
[0149]
[0150] The target time domain echo is:
[0151]
[0152] The target distance-Doppler frequency-slow time three-dimensional data is:
[0153]
[0154] According to the task requirements, select RID (n, f, m; a fine ) is used as the target distance instantaneous Doppler imaging result RID(n,f,m ch ; a fine ), m ch is the slow time sequence number of the selected image, m ch =1,2,…,M.
[0155] In order to facilitate those skilled in the art to understand the present invention, the present invention provides a specific example.
[0156] In the present invention, the target model used in the simulation experiment is as follows: Figure 1 The radar imaging parameters are shown in Table 1. The radial motion parameters of the target are: [a0, a1, a2, a3] = [8000, 80, 30, 5]. The maneuvering parameters of the target are set as follows: the rotation speed is 0.05 rad / s, and the rotation acceleration is 0.023 rad / s 2 , the rotational acceleration is 0.001rad / s 3 The total number of SAO individuals is Q = 40, and the maximum number of iterations of the SAO algorithm is T max =120, preset number v=4, total number of KOA individuals H=20, maximum number of iterations of KOA algorithm T m ' ax =30.
[0157] Table 1
[0158]
[0159] In order to prove the effectiveness and superiority of the method of the present invention, comparative simulation experiments were carried out using different translation compensation methods. The first comparative method is the minimum entropy envelope alignment, minimum entropy self-focusing cascade motion compensation method, called: MERA+MEPA translation compensation method. The second method is the translation compensation method of the present invention, which adopts the distance instantaneous Doppler imaging model based on synchronous extraction transform (Synchroextracting Transform, SET), called: the translation compensation method of the present invention. The third method is the STFT translation compensation method, which replaces the distance instantaneous Doppler imaging model based on synchronous extraction transform (Synchroextracting Transform, SET) in the distance instantaneous Doppler imaging model of the present invention with the distance instantaneous Doppler imaging model based on short-time Fourier transform (STFT), called: STFI translation compensation method. Under the signal-to-noise ratio conditions of SNR=-4dB and SNR=-8dB, the simulation experimental results under different signal-to-noise ratio conditions are as follows. Figure 2-11 As shown. Figure 2 As shown, due to the translation error in the echo, the range image of the target is misplaced. Figure 3 The ideal range Doppler image of the target when SNR=-4dB is given. Since the target has strong maneuvers during the imaging time, the Doppler is time-varying, resulting in severe defocusing of the image. Figure 4 The imaging results of the MERA+MEPA translation compensation method for translation compensation when SNR=-4dB are given. Due to the low signal-to-noise ratio and poor compensation accuracy, the image is very blurred, with defocus and artifacts, which seriously affects the recognition of the target. Figure 5 and Figure 6 The instantaneous Doppler imaging results of the distance after the STFT translation compensation method and the translation compensation method of the present invention are respectively given when SNR = -4dB. Figure 5 and Figure 6 It can be seen that compared with the STFT translation compensation method, the distance instantaneous Doppler imaging quality after translation compensation by the translation compensation method of the present invention is higher. Figure 7 The target range image is given when SNR = -8dB. Due to the translation error in the echo, the target range image is misplaced. Figure 8 The ideal range Doppler image of the target when SNR=-8dB is given. Since the target has strong maneuvers during the imaging time, the Doppler is time-varying, resulting in severe defocusing of the image. Fig. 9The imaging results of the MERA+MEPA translation compensation method for translation compensation when SNR=-8dB are given. Due to the low signal-to-noise ratio, the compensation accuracy is poor, the image is defocused, and imaging fails. Fig.10 and Fig.11 The instantaneous Doppler imaging results of the distance after the STFT translation compensation method and the translation compensation method of the present invention are respectively given when SNR = -8dB. Fig.10 and Fig.11 It can be seen that compared with the STFT translation compensation method, the distance instantaneous Doppler imaging quality after translation compensation by the translation compensation method of the present invention is higher.
[0160] From the above simulation experiments, it can be concluded that: compared with the MERA+MEPA translation compensation method, the translation compensation method of the present invention can accurately compensate for the translation of a highly maneuverable target under low signal-to-noise ratio conditions. Compared with the STFT translation compensation method, the translation compensation method of the present invention has a higher time-frequency resolution and achieves high-quality imaging of the target. The translation compensation method of the present invention solves the problem of accurate translation compensation of highly maneuverable targets under low signal-to-noise ratio conditions, obtains high-quality distance instantaneous Doppler imaging results of the target, and is beneficial to the feature extraction and recognition of the target. Therefore, the present invention has great application value.
[0161] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.
Claims
1. A method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions, characterized in that: include: Step 1: Perform Fourier transform on the target's initial time domain echo, transform it into the frequency domain to perform residual video phase RVP compensation, and obtain the compensated one-dimensional range image of the target; Perform inverse Fourier transform on the compensated one-dimensional range image of the target to obtain the time domain echo of the target; Step 2: Obtain the reference distance recorded in the radar system information when performing Dechirp processing on the initial time domain echo of the transmitted pulse, and save all the reference distances as a reference distance set; Step 3: Fitting the reference distance set to obtain a polynomial coefficient set of the reference distance set, taking the polynomial coefficients of the reference distance set as initial values of the polynomial coefficients of the radial distance, and saving the initial values of all polynomial coefficients of the radial distance as an initial value set; the radial distance is a high-order polynomial; Step 4: construct a translation compensation model through time domain echo, radial distance and reference distance; discretize the translation compensation model to obtain a discretized translation compensation model; Step 5: Based on the snow ablation optimization SAO algorithm, the positions of all snow ablation optimized SAO individuals are initialized through the initial value set, the SAO radial distances corresponding to the positions of all SAO individuals are input into the discretized translation compensation model to obtain the SAO time domain echoes corresponding to the positions of all SAO individuals, and the SAO time domain echoes corresponding to the positions of all SAO individuals are imaged through the distance instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain SAO distance-Doppler frequency-slow time three-dimensional stereo data, and a preset number of slice data corresponding to the slow time sequence number are selected from the SAO distance-Doppler frequency-slow time three-dimensional stereo data as the SAO distance instantaneous Doppler image; the SAO average image entropy of the SAO distance instantaneous Doppler image is calculated, and the SAO average image entropy is sorted from small to large to construct a SAO sorting queue; Select the positions of the SAO individuals corresponding to the top 50% of the SAO average image entropy in the SAO sorting queue as the SAO input queue; Based on the Kepler optimized KOA algorithm, the positions of all Kepler optimized KOA individuals are initialized through the SAO input queue, the KOA radial distances corresponding to the positions of all KOA individuals are input into the discretized translation compensation model to obtain the KOA time domain echoes corresponding to the positions of all KOA individuals, and the KOA time domain echoes corresponding to the positions of all KOA individuals are imaged and processed through the distance instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain the KOA distance-Doppler frequency-slow time three-dimensional stereo data, and the slice data corresponding to the slow time serial number of a preset number are selected from the KOA distance-Doppler frequency-slow time three-dimensional stereo data as the KOA distance instantaneous Doppler image; the KOA average image entropy of the KOA distance instantaneous Doppler image is calculated; and the position of the KOA individual is updated according to the KOA average image entropy; Based on the SAO average image entropy and the KOA average image entropy, the SAO individual position and the KOA individual position are sorted from small to large, and the SAO individual position and the KOA individual position are merged to construct a SAO-KOA sorting queue. The smallest individual position in the SAO-KOA sorting queue is the global optimal position; the position of the SAO individual is updated, and the global optimal position in the searched SAO-KOA sorting queue is used as the precise motion parameter of the target; Step 6: Input the radial distance corresponding to the precise motion parameters into the discretized translation compensation model to obtain the target time domain echo, and perform imaging processing on the target time domain echo through the distance instantaneous Doppler imaging model based on the synchronous extraction transform SET to obtain the target distance-Doppler frequency-slow time three-dimensional stereo data. Select the slice data corresponding to the slow time number corresponding to the task requirements from the target distance-Doppler frequency-slow time three-dimensional stereo data as the target distance instantaneous Doppler imaging result.
2. The method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions as claimed in claim 1, characterized in that: In step 1, compensate the one-dimensional distance image R g (f r ,t m ) and time domain echo s(t r ,t m ) are: s(t r ,t m )=IFFT(R g (f r ,t m )); Among them, s dc (t r ,t m ) is the initial time domain echo; is the RVP compensation item; t r For quick time, T p is the pulse width, n=1,2,…,N, is the distance unit number, N is the number of distance units, t m For slow time, T CPI is the radar imaging processing time; m = 1, 2, ..., M, is the transmission pulse number; M is the number of azimuth units; γ is the modulation frequency of the radar transmission signal; f r is the distance frequency, f s is the sampling rate; FFT(·) is the Fourier transform; IFFT(·) is the inverse Fourier transform.
3. The method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions as claimed in claim 1, characterized in that: In step 2, the reference distance set R ref for: R ref =[R ref (t1),…,R ref (t m ),…,R ref (t M )]; Among them, R ref (t m ) is the reference distance recorded in the acquired radar system information when performing Dechirp processing on the initial time domain echo of the mth transmission pulse; t m is the slow time, m=1,2,…,M, is the transmission pulse number, and M is the total number of transmission pulses.
4. The method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions as claimed in claim 3, characterized in that: In step 3, the initial value set is: For the reference distance set R ref Perform a third-order polynomial fit to obtain the polynomial coefficient set of the reference distance set The expression is as follows: Where X = [1,…,M] is the discrete slow time set, M is the total number of transmitted pulses; polyfit(·) is the polynomial fitting operation; the polynomial coefficient corresponding to the l-order term of the reference distance set is The polynomial coefficient corresponding to the l-order term of the reference distance set The polynomial coefficient a corresponding to the l-order term of the radial distance l The initial value of all polynomial coefficients of radial distance is saved as the initial value set 5. The method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions as claimed in claim 4, characterized in that: In step 3, the radial distance is: Among them, R T (t m ) is the radial distance, t m is the slow time; m=1,2,…,M, is the transmission pulse number, M is the total number of transmission pulses; l is the different orders, l=0,1,2,3, a l is the coefficient corresponding to the l-order term; Δt m is the pulse repetition interval.
6. A method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions as claimed in any one of claims 1 to 5, characterized in that: In step 4, the translation compensation model is: in, is the time domain echo after translation compensation, s(t r ,t m ) is the time domain echo; j is the imaginary unit, c is the speed of light, f c is the radar signal carrier frequency, γ is the modulation frequency of the radar transmission signal, R T (t m ) is the radial distance, R ref (t m ) is the reference distance; t r For quick time.
7. The method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions as claimed in claim 6, characterized in that: In step 4, the discretized translation compensation model is: in, This is the time domain echo after translation compensation processing The discrete form of s(n,m) is the time domain echo s(t r ,t m ) in discrete form; R T (m) is the radial distance R T (t m ), R ref (m) is the reference distance R ref (t m ) in discrete form; n = 1, 2, ..., N, is the distance unit number.
8. The method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions according to claim 7, wherein the specific steps in step 5 are: 8.1, based on the SAO algorithm, according to the initial value set Determine the upper bound Ua and lower bound La of the polynomial coefficients; initialize the total number of SAO individuals Q, q = 1, 2, 3, ..., Q, which is the serial number of the SAO individual; the initial position A of Q SAO individuals can be obtained through Ua and La, and the initial position A is a matrix with Q rows and 4 columns; set the maximum number of iterations of the SAO algorithm to T max ; The number of iterations of the SAO algorithm is t = 1, 2, ..., T max ; 8.2, when t = 1, the position of the qth SAO individual The corresponding SAO radial distance R T (m; a q ) is input into the discretized translation compensation model to obtain the SAO time domain echo for: in, is the coefficient corresponding to the l-th order term of the SAO radial distance corresponding to the position of the q-th SAO individual; 8.3, the method of imaging the SAO time domain echoes corresponding to the positions of all SAO individuals by using the range instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain the SAO range-Doppler frequency-slow time three-dimensional stereo data is: Among them, RID(n,f,m;a q ) is the position a of the qth SAO individual q The corresponding SAO time domain echo input is based on the SAO range-Doppler frequency-slow time three-dimensional data obtained in the range instantaneous Doppler imaging model of the synchronous extraction transformation SET, where f is the frequency sequence number, f=1,2,…,F, F is the maximum frequency sequence number, and SET(·) represents the synchronous extraction transformation operation; 8.4, 1≤v≤M, v is the preset number; select the slice data corresponding to the v-slow time sequence number from the SAO range-Doppler frequency-slow time three-dimensional data, and the obtained SAO range instantaneous Doppler image is RID(n,f,m vc ; a q ), m vc is the slow time sequence number of the selected image, vc=1,…,v; 8.5, the SAO average image entropy En(a) of the instantaneous Doppler image at SAO distance v q )for: Where Ec(·) represents the image entropy calculation operation; 8.6, according to the SAO average image entropy En(a q ) Sorting the positions of the SAO individuals from small to large to construct an SAO sorting queue; selecting the positions of the first 50% of the SAO individuals in the SAO sorting queue as the SAO input queue; 8.7, based on the KOA algorithm, initialize the positions of all KOA individuals through the SAO input queue; initialize the total number of KOA individuals H, which is the maximum integer not greater than 0.5Q; h = 1, 2, ..., H, is the sequence number of the KOA individual; set the maximum number of iterations of the KOA algorithm to T' max ; The number of iterations of the KOA algorithm is t′=1,2,…,T′ max ; 8.8, when t′=1, the position of the hth KOA individual is The corresponding KOA radial distance is input into the discretized translation compensation model to obtain the KOA time domain echo; Among them, the KOA radial distance is: The KOA time domain echo is: 8.9, the method of imaging the KOA time domain echoes corresponding to the positions of all KOA individuals by using the range instantaneous Doppler imaging model based on the synchronous extraction transformation SET to obtain the KOA range-Doppler frequency-slow time three-dimensional stereo data is: in, RID(n,f,m;a h ) is the KOA individual position a h The corresponding KOA time domain echo input is based on the KOA range-Doppler frequency-slow time three-dimensional data obtained in the range instantaneous Doppler imaging model of the synchronous extraction transformation SET; f is the frequency sequence number, f = 1, 2, ..., F, F is the maximum frequency sequence number, SET (·) represents the synchronous extraction transformation operation; 8.10, 1≤v≤M, v is the preset number; select the slice data corresponding to the v-slow time sequence number from the KOA range-Doppler frequency-slow time three-dimensional data, and the obtained KOA range instantaneous Doppler image is RID(n,f n ,m vc ; a h ), m vc is the slow time sequence number of the selected image, vc=1,…,v; 8.11, the KOA average image entropy of the instantaneous Doppler image of KOA distance v is: Where Ec(·) represents the image entropy calculation operation; 8.12, increase t′ by 1. If the number of iterations of the KOA algorithm reaches the maximum value T′ max , terminate the search process, and sort the SAO individual position and the KOA individual position from small to large based on the SAO average image entropy and the KOA average image entropy, merge the SAO individual position with the KOA individual position, and build a SAO-KOA sorting queue. The smallest individual position in the SAO-KOA sorting queue is the global optimal position. Otherwise, update the position of the KOA individual according to the KOA average image entropy, and repeat 8.8 to 8.12; 8.13, increase t by 1. If the number of iterations of the SAO algorithm reaches the maximum value T max , terminate the search process, otherwise, update the position of the SAO individual, repeat 8.2 to 8.13, and finally output the position of the smallest individual in the SAO-KOA sorting queue as the precise motion parameter a fine .
9. The method for compensating translation of a high-maneuvering target under low signal-to-noise ratio conditions as claimed in claim 8, characterized in that: The specific method in step six includes: Precise motion parametersa fine =[a′0,a′1,a′2,a′3]corresponding radial distance R T (m; a fine )for: The target time domain echo is: The target distance-Doppler frequency-slow time three-dimensional data is: According to the task requirements, select RID (n, f, m; a fine ) The slow time sequence number is m ch The slice data is used as the target distance instantaneous Doppler imaging result RID(n,f,m ch ; a fine ), m ch is the slow time sequence number of the selected image, m ch =1,2,…,M.
