CN114140325B - C-ADMN-based structured sparse aperture ISAR imaging method - Google Patents
C-ADMN-based structured sparse aperture ISAR imaging method Download PDFInfo
- Publication number
- CN114140325B CN114140325B CN202111464674.3A CN202111464674A CN114140325B CN 114140325 B CN114140325 B CN 114140325B CN 202111464674 A CN202111464674 A CN 202111464674A CN 114140325 B CN114140325 B CN 114140325B
- Authority
- CN
- China
- Prior art keywords
- representing
- admn
- network
- sparse aperture
- isar
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000003384 imaging method Methods 0.000 title claims abstract description 42
- 238000000034 method Methods 0.000 claims abstract description 36
- 239000011159 matrix material Substances 0.000 claims description 23
- 230000008569 process Effects 0.000 claims description 13
- 238000004422 calculation algorithm Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 11
- 238000009825 accumulation Methods 0.000 claims description 2
- 238000010276 construction Methods 0.000 claims description 2
- 238000005070 sampling Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 6
- 230000009286 beneficial effect Effects 0.000 abstract description 2
- 238000000605 extraction Methods 0.000 abstract description 2
- 238000002592 echocardiography Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000008570 general process Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
- G06T3/4053—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution
- G06T3/4076—Scaling of whole images or parts thereof, e.g. expanding or contracting based on super-resolution, i.e. the output image resolution being higher than the sensor resolution using the original low-resolution images to iteratively correct the high-resolution images
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/22—Source localisation; Inverse modelling
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention belongs to the field of radar imaging, and particularly relates to a structured sparse aperture ISAR imaging method based on C-ADMN, which comprises the following steps: s1, modeling a sparse aperture ISAR echo signal; s2, constructing a C-ADMN forward propagation model; s3, solving a structured sparse aperture ISAR imaging problem by using the C-ADMN. The beneficial effects obtained by the invention are as follows: according to the invention, the target structured sparse aperture ISAR imaging can be realized, the high-resolution ISAR image with complete target structure can be rapidly obtained under the sparse aperture condition, the scattering point distribution characteristic of the ISAR image can be better reflected, and the target structure information can be acquired more clearly. Meanwhile, the invention can obtain better imaging effect without depending on manual parameter setting. The method has important engineering application value for target radar imaging, feature extraction and target identification under the sparse aperture condition.
Description
Technical Field
The invention belongs to the field of radar imaging, and particularly relates to a structured sparse aperture inverse synthetic aperture radar (Inverse synthetic aperture radar, ISAR) imaging method based on a convolution alternate direction multiplier method network (Convolutional alternating direction method of multipliers network, C-ADMN).
Background
ISAR can observe moving targets all the time and all the weather, and reach higher distance resolution and azimuth resolution at the same time, thereby obtaining target resolution two-dimensional images and providing a new technical means for space target identification.
Sparse aperture ISAR imaging refers to imaging a target with incomplete echo data of a radar receiver. Sparse aperture phenomenon is common in actual imaging scenes, in which case, correlation among ISAR echoes is destroyed, imaging is interfered by stronger side lobes and grating lobes, and azimuth resolution is seriously reduced. At this time, a convex optimization model can be generally constructed by using sparsity prior information of ISAR image scattering point distribution, and the image is solved.
In practical application, the target often has a complex shape structure, and the ISAR image scattering point distribution of the target has structural property. Imaging using ISAR image scatter point structural prior information under sparse aperture conditions is commonly referred to as structured sparse aperture ISAR imaging. The complex structure of the object presents difficulties for modeling a priori information, for which a weighting l can be employed 1 Norm minimization method (E.Candes, M.Wakin, and S.Boyd, "Enhancing sparsity by reweighted l) 1 minimization, "in Journal of Fourier Analysis and Applications, vol.14, no.5, pp.877-905,2008). However, the method has weak adaptability to data in different scenes in practical application, the effect depends on manual parameter debugging, the operation efficiency is low, and difficulty is brought to practical application.
Disclosure of Invention
The thinking of the invention is based on weighting l under the sparse aperture condition 1 The structured ISAR imaging algorithm with minimized norm has the problems of low data adaptability, strong parameter sensitivity and low operation efficiency, and a structured sparse aperture ISAR imaging method based on C-ADMN is provided. The method uses depth expansion to base the tradition on weighting l 1 And combining the ADMM algorithm of the norm with a convolutional neural network to construct a C-ADMN depth network model. Aiming at the network model, a complex domain back propagation method is applied to train the network, so as to realize the self-adaptive learning of network parameters. Finally, inputting the sparse aperture ISAR echo of the complex structure target into the trained netAnd the complex model can output and obtain a high-quality ISAR image with a clear target structure.
The technical scheme adopted for solving the technical problems is as follows: a structured sparse aperture ISAR imaging method based on C-ADMN comprises the following steps:
s1, modeling a sparse aperture ISAR echo signal:
the ISAR echo can be imaged after translational compensation, and because the technical route is relatively mature, the invention assumes that the translational compensation is completed, and directly processes the one-dimensional range profile sequence after the translational compensation (guarantor, xing Mengdao, wang Tong. Radar imaging technology [ M ]. Beijing: electronic industry Press, 2005). The two-dimensional echo received by the radar system after demodulation can be expressed as:
wherein sigma i Representing the reflection coefficient of the ith scattering center of the target, R i (t m ) Representing the instantaneous rotational distance of the ith scattering center of the target relative to the radar,t m respectively express fast time and slow time, f c C and gamma respectively represent the center frequency of a radar signal, the vacuum light speed and the signal modulation frequency; m=1, 2, …, M represents the number of pulses contained in the full aperture radar echo.
Instantaneous rotation distance R of ith scattering center of target relative to radar i (t m ) Can be expressed as:
R i (t m )=x i sin(ωt m )+y i cos(ωt m )≈x i ωt m +y i (2)
wherein, (x) i ,y i ) Represents the coordinate of the ith scattering center of the target in a reference coordinate system, omega represents the rotation angular speed of the target, and the target moves in the imaging time due to the short ISAR imaging accumulation time and stable target movement stateCan be regarded as constant rotation. Because the pulse repetition time is short, the target rotation angle is small, so there is sin (ωt m )≈ωt m 、cos(ωt m )≈1。
Substituting equation (2) into equation (1) can obtain radar approximate two-dimensional echoWill->Digital about fast time->Performing fast Fourier transform (Fast Fourier transform, FFT) to obtain a target one-dimensional range profile sequence s' (f, t) m )。
Radar signal at slow time t m The internal contains M pulses, the fast timeThe sparse aperture signal comprises L pulses, L < M. For a one-dimensional range profile sequence s' (f, t) under sparse aperture conditions m ) The following downsampling model is established:
Y=AX+W=PFX+W (3)
wherein the method comprises the steps ofRepresenting the downsampled one-dimensional range profile sequence, < >>Representing the observation matrix->Representing ISAR image matrix, < >>Representing a gaussian white noise matrix; />Representing a fourier transform matrix;representing a downsampling matrix consisting of 0 and 1, let ∈>The distance image index representing the sampled distance image isFor the m-column element P of the first row of the matrix P l,m When the vector s is the first element s l When m, there is P l,m =1, otherwise P l,m =0,l=1,2,…,L。/>Representing an L N-dimensional complex matrix, ">Representing an L x M-dimensional real matrix, ">Representing the L-dimensional real vector, and so on.
S2, constructing a C-ADMN forward propagation model:
the C-ADMN is formed by cascading K-level sub-networks, K K =1, 2, …, and the K-1 level sub-network is formed by three network layers in sequence, and is respectively a reconstruction layer, a noise reduction layer and a multiplier updating layer; while the K-th sub-network contains only one reconstruction layer. In the kth level of subnetwork, the reconstruction layer outputs X (k) Representing the reconstructed ISAR image, the noise reduction layer outputs Z (k) Representation of X (k) The result obtained after noise reduction is output B by the multiplier updating layer (k) Representing the updated lagrangian multiplier. The final output of the network is X (K) 。
The specific construction process is as follows:
s2.1, constructing a reconstruction layer forward propagation model:
in a u (u=1, 2,., K) level subnetwork, the reconstruction layer forward propagation procedure can be represented as equation (4):
X (u) =F(P H P+μ (u) I M ) -1 (P H Y-FB (u-1) -μ (u) FZ (u-1) ) (4)
wherein I is M Represents M×M dimension identity matrix, μ (u) A penalty parameter representing the class u subnetwork to be trained is typically initialized to 1. When u=1, B (0) =0 M×N ,Z (0) =0 M×N Wherein 0 is M×N Representing an all 0 matrix in M x N dimensions. The penalty parameter training process in the reconstruction layer is given at S3.1.
S2.2, constructing a noise reduction layer forward propagation model:
in a K (k=1, 2., K-1) level subnetwork, the noise reduction layer forward propagation process can be expressed as:
wherein S is (·) (. Cndot.) represents a soft threshold function, with respect to any complex scalar x and real threshold tS is provided for any complex vector x and real number threshold t t (x)=[S t (x 1 ),S t (x 2 ),...] T Wherein x is i Representing the ith element of the complex vector x. />The initial value of the convolution kernel parameter representing the kth level sub-network to be trained is 1 3×3 ,1 3×3 Representing a 3 x 3 dimensional full 1 matrix; symbol represents a two-dimensional convolution operation; lambda (lambda) (k) Representing regularization parameters to be trained of the kth level sub-network, wherein the initial value is 0.1; epsilon is a sufficiently small constant to avoid singular values, typically let epsilon=10 -6 . The training process of the convolution kernel parameters and regularization parameters in the noise reduction layer is given in S3.1.
S2.3, updating a layer forward propagation model by a multiplier:
in a K (k=1, 2., K-1) level subnetwork, the multiplier update layer forward propagation expression is as shown in equation (6):
B (k) =B (k-1) +μ (k) (X (k) -Z (k) ) (6)
s3, solving a structured sparse aperture ISAR imaging problem by using the C-ADMN:
s3.1 training C-ADMN network:
S3.1.1A data set (electromagnetic calculation, software simulation, darkroom measurement, etc., are used, Q.Liu, X.Zhang, Y.Liu, K.Huo, W.Jiang and X.Li, "Multi-Polarization Fusion Few-Shot HRRP Target Recognition Based on Meta-Learning Framework," in IEEE Sensors Journal, vol.21, no.16, pp.18085-18100, aug, 2021) similar to the actual application scene is constructed, and the data set contains Q groups of sparse aperture range profile data Y q Q=1, 2,. -%, Q; each group of sparse aperture range profile data corresponds to a group of label image data
S3.1.2 initialize network parameters. For any network parameter μ (k) 、λ (k) 、c (k) (k=1,2,...,K-1)、μ (K) Initialize it asWherein->
S3.1.3 for a constructed dataset, the following root mean square error loss function is defined:
wherein,representing initialized C-ADMN network utilization Y q As a result of the imaging performed, the image data,representing the initialized set of network parameters. For the loss function of equation (7), the loss function can be calculated with respect to any parameter in the network by applying a complex domain back propagation algorithm (G.M. Georgiou and C.Koutsouglas, "Complex domain backpropagation," in IEEE Transactions on Circuits and Systems II: analog and Digital Signal Processing, vol.39, no.5, pp.330-334, may 1992)>Gradient of->After solving to obtain the gradient, adopting a gradient descent algorithm, and carrying out parameter updating by using the formula (8):
where η represents the learning rate, typically taking η=10 -3 . Parameters (parameters)Respectively represent pairs of And performing parameter values after the gradient descent. After gradient descent, an updated network parameter set +.>And further calculate the loss function after parameter update using equation (7)>Repeatedly using the formulas (7) and (8) to update the parameters in an iterative manner, ending the training process when the termination condition is met, and setting the iterative step as J at the time to obtain a final network parameter value +.>The iteration termination condition is->And->And->And->
S3.2, reconstructing an ISAR image by using the trained C-ADMN network:
and acquiring actual observation sparse aperture radar echoes, and obtaining a sparse aperture one-dimensional range profile sequence through fast time FFT. And carrying out translational rough compensation on the one-dimensional range profile sequence by using a cross-correlation method to obtain a one-dimensional range profile matrix (roll-keeping, xing Mengdao, wang Tong. Radar imaging technology [ M ]. Beijing: electronic industry Press, 2005) after rough compensation.
And inputting the coarsely compensated one-dimensional distance image matrix into a trained C-ADMN network, wherein the output of the C-ADMN network is the ISAR image to be solved.
The beneficial effects obtained by the invention are as follows: according to the invention, the target structured sparse aperture ISAR imaging can be realized, the high-resolution ISAR image with complete target structure can be rapidly obtained under the sparse aperture condition, the scattering point distribution characteristic of the ISAR image can be better reflected, and the target structure information can be acquired more clearly. Meanwhile, the invention can obtain better imaging effect without depending on manual parameter setting. The method has important engineering application value for target radar imaging, feature extraction and target identification under the sparse aperture condition.
Drawings
FIG. 1 is a flow chart of an embodiment of the present invention;
FIG. 2C-ADMN overall block diagram;
fig. 3 Yak-42 aircraft in full aperture condition: (a) a one-dimensional range profile sequence; (b) an ISAR image;
fig. 4 under sparse aperture conditions with a sparsity of 50%: (a) a target one-dimensional range profile sequence; (b) range-doppler method resulting target ISAR image; (c) target ISAR image obtained by ADMM method; (d) ISAR images obtained by the present invention;
fig. 5 under sparse aperture condition with 25% sparsity: (a) a target one-dimensional range profile sequence; (b) range-doppler method resulting target ISAR image; (c) target ISAR image obtained by ADMM method; (d) ISAR images obtained by the present invention;
Detailed Description
The invention is further described below with reference to the accompanying drawings:
FIG. 1 is a general process flow of the present invention. The invention discloses a structured sparse aperture ISAR imaging method based on C-ADMN, which comprises the following steps:
s1, modeling a sparse aperture ISAR echo signal;
s2, constructing a C-ADMN forward propagation model;
s3, solving a structured sparse aperture ISAR imaging problem by using the C-ADMN;
FIG. 2 is a C-ADMN overall structure, in which the connection relation of the reconstruction layer, the noise reduction layer and the multiplier update layer is marked, and the specific structure of each layer is also provided.
Fig. 3 (a) and 3 (b) are respectively a Yak-42 aircraft one-dimensional range profile sequence and an ISAR image under the full aperture condition. The radar transmit signal parameters are as follows: the center frequency is 5.52GHz, the bandwidth is 400MHz, the pulse width is 25.6 μs, the full aperture contains 256 pulses in a range-wise sequence, each pulse containing 256 range cells.
128 pulses are randomly extracted from the full-aperture data, sparse aperture data with the sparsity of 50% are obtained, and at this time, a target one-dimensional range profile sequence is shown in fig. 4 (a). The sparse aperture data was further ISAR imaged using conventional range-Doppler (RD) methods, conventional ADMM algorithms, and the present invention, and the resulting ISAR images were shown in FIG. 4 (b), FIG. 4 (c), FIG. 4 (d), respectively. As can be seen from fig. 4 (b), due to the sparse aperture effect, the ISAR image is interfered by serious side lobes and grating lobes, the resolution of the azimuth dimension is lower, and the image quality is poor. The methods shown in fig. 4 (c) and fig. 4 (d) can obtain images with higher quality. Compared with the traditional ADMM algorithm, the method provided by the invention has the advantages that the target structure characteristic is imaged more clearly, and the result is closer to full-aperture label data.
Further, 64 pulses were randomly extracted from the full aperture data to simulate sparse aperture data with a sparsity of 25%. Under this condition, the target one-dimensional range profile sequence is shown in fig. 5 (a), the ISAR image obtained by adopting the RD method is shown in fig. 5 (b), and the ISAR images obtained by adopting the traditional ADMM algorithm and the invention are respectively shown in fig. 5 (c) and 5 (d). At this time, the image quality is extremely poor because the sparsity is further reduced and the azimuth resolution of fig. 5 (b) is further reduced. While the image shown in fig. 5 (c) is clearer, the image is thinner, and the imaging of the target main structure is unclear. From fig. 5 (d), it can be seen that the present invention can still clearly image the main body structure of the target fuselage under the condition of lower sparsity. Further, the invention is verified to have better imaging effect on the target with the block sparse structure under the sparse aperture condition.
In conclusion, the invention can effectively realize the target structured ISAR imaging function under the sparse aperture condition, has good effects on the sparse aperture data with 50% sparsity and 25% sparsity, and can obtain the target structure shape more clearly than the traditional ADMM algorithm. The invention avoids the complex parameter setting process in the traditional method, enhances the data adaptability, improves the algorithm operation efficiency and has higher engineering application value.
Claims (5)
1. A structured sparse aperture ISAR imaging method based on C-admn, the method comprising the steps of:
s1, modeling a sparse aperture ISAR echo signal:
the two-dimensional echo received by the radar system is expressed as:
wherein sigma i Representing the reflection coefficient of the ith scattering center of the target, R i (t m ) Representing the instantaneous rotational distance of the ith scattering center of the target relative to the radar,t m respectively express fast time and slow time, f c C and gamma respectively represent the center frequency of a radar signal, the vacuum light speed and the signal modulation frequency; m=1, 2, …, M representing the number of pulses comprised by the full aperture radar echo;
instantaneous rotation distance R of ith scattering center of target relative to radar i (t m ) Expressed as:
R i (t m )=x i sin(ωt m )+y i cos(ωt m )≈x i ωt m +y i (2)
wherein, (x) i ,y i ) Representing the coordinate of the ith scattering center of the target under a reference coordinate system, wherein omega represents the rotation angular speed of the target, and the rotation component of the target moving in the imaging time can be regarded as uniform rotation because the ISAR imaging accumulation time is short and the target moving state is stable; because the pulse repetition time is short, the target rotation angle is small, so there is sin (ωt m )≈ωt m 、cos(ωt m )≈1;
Substituting equation (2) into equation (1) can obtain radar approximate two-dimensional echoWill->Digital about fast time->Performing fast Fourier transform to obtain a target one-dimensional range profile sequence s' (f, t) m );
Radar signal at slow time t m The internal contains M pulses, the fast timeThe inside of the device comprises N sampling points, the sparse aperture signal comprises L pulses, and L is less than M; for a one-dimensional range profile sequence s' (f, t) under sparse aperture conditions m ) The following downsampling model is established:
Y=AX+W=PFX+W (3)
wherein the method comprises the steps ofRepresenting the downsampled one-dimensional range profile sequence, < >>Representing the observation matrix->Representing ISAR image matrix, < >>Representing a gaussian white noise matrix; />Representing a fourier transform matrix; />Representing a downsampling matrix consisting of 0 and 1, let ∈>Representing the sampled range profile index, there is +.>For the m-column element P of the first row of the matrix P l,m When the vector s is the first element s l When m, there is P l,m =1, otherwise P l,m =0,l=1,2,…,L;/>Representing an L N-dimensional complex matrix, ">Representing an L x M-dimensional real matrix, ">Representing an L-dimensional real vector;
s2, constructing a C-ADMN forward propagation model:
the C-ADMN is formed by cascading K-level sub-networks, the K-level sub-network is formed by three network layers in sequence, and k=1, 2, … and K1 are respectively a reconstruction layer, a noise reduction layer and a multiplier updating layer; and the K-level sub-network only comprises a reconstruction layer; in the kth level of subnetwork, the reconstruction layer outputs X (k) Representing the reconstructed ISAR image, the noise reduction layer outputs Z (k) Representation of X (k) The result obtained after noise reduction is output B by the multiplier updating layer (k) Representing the updated Lagrangian multiplier, and the final output of the network is X (K) ;
The specific construction process is as follows:
s2.1, constructing a reconstruction layer forward propagation model:
in the level u subnetwork, u=1, 2,..k, the reconstructed layer forward propagation process can be expressed as equation (4):
X (u) =F(P H P+μ (u) I M ) -1 (P H Y-FB (u-1) -μ (u) FZ (u-1) ) (4)
wherein I is M Represents M×M dimension identity matrix, μ (u) Representing punishment parameters to be trained of the sub-network of the nth level; when u=1, B (0) =0 M×N ,Z (0) =0 M×N Wherein 0 is M×N Representation ofAn all 0 matrix of m×n dimensions; the penalty parameter training process in the reconstruction layer is given in S3.1;
s2.2, constructing a noise reduction layer forward propagation model:
in the kth level subnetwork, the noise reduction layer forward propagation process can be expressed as:
wherein S is (·) (. Cndot.) represents a soft threshold function, with respect to any complex scalar x and real threshold tS is provided for any complex vector x and real number threshold t t (x)=[S t (x 1 ),S t (x 2 ),...] T Wherein x is i An i-th element representing a complex vector x; />Representing a convolution kernel parameter to be trained of a kth level sub-network; symbol represents a two-dimensional convolution operation; lambda (lambda) (k) Representing regularization parameters to be trained of the kth level sub-network; epsilon is a sufficiently small constant to avoid generating singular values; the training process of the convolution kernel parameters and regularization parameters in the noise reduction layer is given in S3.1;
s2.3, updating a layer forward propagation model by a multiplier:
in the kth level subnetwork, the multiplier update layer forward propagation expression is as shown in equation (6):
B (k) =B (k-1) +μ (k) (X (k) -Z (k) ) (6)
s3, solving a structured sparse aperture ISAR imaging problem by using the C-ADMN:
s3.1 training C-ADMN network:
s3.1.1 firstly, constructing a data set similar to the actual application scene, wherein the data set comprises Q groups of sparse aperture range profile data Y q Q=1, 2,. -%, Q; each set of sparsity Kong JingjuThe image-separating data corresponds to a group of tag image data
S3.1.2 initializing network parameters;
s3.1.3 for a constructed dataset, the following root mean square error loss function is defined:
wherein,representing initialized C-ADMN network utilization Y q As a result of the imaging performed, the image data,representing the initialized network parameter set; for the loss function of (7), calculating the loss function about any parameter in the network by applying a complex domain back propagation algorithm>Gradient of->After solving to obtain the gradient, adopting a gradient descent algorithm, and carrying out parameter updating by using the formula (8):
wherein η represents a learning rate; parameters (parameters)Respectively represent pair->Carrying out parameter values after primary gradient descent; after gradient descent, updated network parameter set can be obtainedAnd further calculate the loss function after parameter update by using equation (7)Repeatedly using the formulas (7) and (8) to update the parameters in an iterative manner, ending the training process when the termination condition is met, and setting the iterative step as J at the time to obtain a final network parameter value +.>
S3.2, reconstructing an ISAR image by using the trained C-ADMN network:
acquiring an actual observation sparse aperture radar echo, and acquiring a sparse aperture one-dimensional range profile sequence through fast time FFT; performing translational rough compensation on the one-dimensional distance image sequence by using a cross-correlation method to obtain a one-dimensional distance image matrix after rough compensation;
and inputting the coarsely compensated one-dimensional distance image matrix into a trained C-ADMN network, wherein the output of the C-ADMN network is the ISAR image to be solved.
2. A C-admn-based structured sparse aperture ISAR imaging method according to claim 1, wherein: s2.2, let ε=10 -6 。
3. A C-admn-based structured sparse aperture ISAR imaging method according to claim 1, wherein: s3.1.2 mu, mu (k) 、λ (k) 、c (k) 、μ (K) Respectively initialized toWherein->
4. A C-admn-based structured sparse aperture ISAR imaging method according to claim 1, wherein: s3.1.3, η=10 -3 。
5. A C-admn-based structured sparse aperture ISAR imaging method according to claim 1, wherein: s3.1.3 the iteration termination condition isAnd->And->And->
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111464674.3A CN114140325B (en) | 2021-12-02 | 2021-12-02 | C-ADMN-based structured sparse aperture ISAR imaging method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111464674.3A CN114140325B (en) | 2021-12-02 | 2021-12-02 | C-ADMN-based structured sparse aperture ISAR imaging method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114140325A CN114140325A (en) | 2022-03-04 |
CN114140325B true CN114140325B (en) | 2024-04-09 |
Family
ID=80387476
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111464674.3A Active CN114140325B (en) | 2021-12-02 | 2021-12-02 | C-ADMN-based structured sparse aperture ISAR imaging method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114140325B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20190036325A (en) * | 2017-09-27 | 2019-04-04 | 포항공과대학교 산학협력단 | Apparatus for autofocusing and cross range scaling of isar image using compressive sensing and method thereof |
CN110703249A (en) * | 2019-11-05 | 2020-01-17 | 中国民航大学 | Robust and efficient synthetic aperture radar multi-feature enhanced imaging method |
CN111610522A (en) * | 2020-06-04 | 2020-09-01 | 中国人民解放军国防科技大学 | SA-ISAR imaging method for target with micro-motion component based on low-rank and sparse combined constraint |
CN112099008A (en) * | 2020-09-16 | 2020-12-18 | 中国人民解放军国防科技大学 | SA-ISAR imaging and self-focusing method based on CV-ADMMN |
CN112965064A (en) * | 2021-01-19 | 2021-06-15 | 中国人民解放军国防科技大学 | PCSBL-GAMP-Net-based block sparse aperture ISAR imaging method |
Family Cites Families (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9702971B2 (en) * | 2014-03-17 | 2017-07-11 | Raytheon Company | High-availability ISAR image formation |
US10042046B2 (en) * | 2015-07-07 | 2018-08-07 | Mitsubishi Electric Research Laboratories, Inc. | System and method for radar imaging using distributed arrays and compressive sensing |
CN117192548A (en) * | 2023-08-28 | 2023-12-08 | 西安电子科技大学 | Sparse ISAR high-resolution imaging method based on depth expansion |
-
2021
- 2021-12-02 CN CN202111464674.3A patent/CN114140325B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR20190036325A (en) * | 2017-09-27 | 2019-04-04 | 포항공과대학교 산학협력단 | Apparatus for autofocusing and cross range scaling of isar image using compressive sensing and method thereof |
CN110703249A (en) * | 2019-11-05 | 2020-01-17 | 中国民航大学 | Robust and efficient synthetic aperture radar multi-feature enhanced imaging method |
CN111610522A (en) * | 2020-06-04 | 2020-09-01 | 中国人民解放军国防科技大学 | SA-ISAR imaging method for target with micro-motion component based on low-rank and sparse combined constraint |
CN112099008A (en) * | 2020-09-16 | 2020-12-18 | 中国人民解放军国防科技大学 | SA-ISAR imaging and self-focusing method based on CV-ADMMN |
CN112965064A (en) * | 2021-01-19 | 2021-06-15 | 中国人民解放军国防科技大学 | PCSBL-GAMP-Net-based block sparse aperture ISAR imaging method |
Non-Patent Citations (3)
Title |
---|
Deep Learning Approach for Sparse Aperture ISAR Imaging and Autofocusing Based on Complex-Valued ADMM-Net;Ruize Li, et al.;IEEE SENSORS JOURNAL;20210201;第21卷(第3期);第3437-3451页 * |
一种基于多维交替方向乘子法的多输入多输出逆合成孔径雷达成像方法;邓理康等;雷达学报;20210630;第10卷(第3期);第416-431页 * |
一种高分辨的稀疏孔径ISAR成像方法;李军;邢孟道;张磊;吴顺君;;西安电子科技大学学报;20100620(第03期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN114140325A (en) | 2022-03-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112099008B (en) | SA-ISAR imaging and self-focusing method based on CV-ADMMN | |
CN110244303B (en) | SBL-ADMM-based sparse aperture ISAR imaging method | |
CN110068805B (en) | High-speed target HRRP reconstruction method based on variational Bayesian inference | |
CN111077523A (en) | Inverse synthetic aperture radar imaging method based on generation countermeasure network | |
CN112990334A (en) | Small sample SAR image target identification method based on improved prototype network | |
CN112782695B (en) | Satellite attitude and size estimation method based on ISAR image and parameter optimization | |
CN112859074B (en) | Multi-band multi-view ISAR fusion imaging method | |
CN110780298A (en) | Multi-base ISAR fusion imaging method based on variational Bayes learning | |
Zhu et al. | Shadow compensation for synthetic aperture radar target classification by dual parallel generative adversarial network | |
CN111948652B (en) | SAR intelligent parameterized super-resolution imaging method based on deep learning | |
CN114442092B (en) | SAR deep learning three-dimensional imaging method for distributed unmanned aerial vehicle | |
CN110969212A (en) | ISAR image classification method based on spatial transformation three-channel convolution | |
CN112147608A (en) | Rapid Gaussian gridding non-uniform FFT through-wall imaging radar BP method | |
CN110275163B (en) | Millimeter wave radar detection target imaging method based on neural network | |
Mao et al. | Target fast reconstruction of real aperture radar using data extrapolation-based parallel iterative adaptive approach | |
CN105759264A (en) | Micro-motion target defect echo high-resolution imaging method based on time-frequency dictionary | |
CN114140325B (en) | C-ADMN-based structured sparse aperture ISAR imaging method | |
CN109946696B (en) | Radar staring correlated imaging method based on target prior information | |
CN108931770B (en) | ISAR imaging method based on multi-dimensional beta process linear regression | |
CN112965064B (en) | PCSBL-GAMP-Net-based block sparse aperture ISAR imaging method | |
CN112946644B (en) | Based on minimizing the convolution weight l1Norm sparse aperture ISAR imaging method | |
CN114624646B (en) | DOA estimation method based on model driven complex neural network | |
CN114519384B (en) | Target classification method based on sparse SAR amplitude-phase image dataset | |
Song et al. | Efficient Image Reconstruction Methods Based on Structured Sparsity for Short-Range Radar | |
CN110082765B (en) | Spatial target attitude extrapolation method based on three-dimensional reconstruction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |