Two-dimensional compressed sensing moving target imaging detection method based on scattering coefficient weighting

Abstract

The invention relates to a two-dimensional compressed sensing moving target imaging detection method based on scattering coefficient weighting, and belongs to the technical field of radar SAR imaging. Comprising the following steps: acquiring dual-channel SAR echo data, and carrying out downsampling processing on the dual-channel SAR echo data; constructing a reconstruction matrix based on the SAR echo data of the two channels after downsampling processing; reconstructing a scene scattering coefficient based on a weighted two-dimensional compressed sensing imaging algorithm and the reconstruction matrix; imaging the scene based on the dual-channel scene scattering coefficient to obtain two coherent images; and suppressing a static target in the image after dual-channel clutter cancellation by using a DPCA method to obtain a moving target. According to the method, the reconstruction precision under the condition of low signal-to-noise ratio is improved.

Description

Technical Field

[0001] The invention belongs to the field of radar SAR imaging technology, and in particular relates to a two-dimensional compressed sensing moving target imaging detection method based on scattering coefficient weighting. Background Art

[0002] Since the 1990s, the theory of sparse signal processing has developed rapidly due to the many advantages of sparse signal representation. In 2006, the compressed sensing theory proposed by D. Donoho, E. Candès and others provided a theoretical basis for sparse signal processing. The compressed sensing method uses a measurement matrix to observe a high-dimensional signal that can be sparsely decomposed to obtain a small number of observations, and then uses the sparsity of the signal to reconstruct the original signal by solving a nonlinear optimization problem. The compressed sensing theory believes that the minimum sampling rate required to recover the signal depends on the information rate of the signal, not the signal bandwidth. By making full use of the sparsity or compressibility of the signal, the sampling rate required to recover the signal can be reduced. That is, the compressed sensing theory can achieve accurate reconstruction of the signal at a sampling rate lower than that required by the Nyquist sampling theorem. Once the compressed sensing theory was proposed, it attracted widespread attention in many fields.

[0003] A key to the practical application of compressed sensing theory is to seek an accurate and efficient recovery and reconstruction algorithm. Since the introduction of compressed sensing, the recovery and reconstruction algorithm based on sparse signals has developed rapidly. There are mainly convex optimization algorithms and greedy algorithms. The reconstruction error of convex optimization algorithms is small and the reconstruction effect is good, but the computational complexity of such methods is high and the practicality is poor for large-scale problems. With the increasing maturity of CS theory, domestic and foreign scholars have carried out research on the application of CS theory to radar. In radar SAR imaging, for the entire scene observed by the radar, a small number of strong scattering points contribute most of its energy, and the strong scattering points are very sparse relative to the entire observation scene, so it is possible to apply CS theory to radar. Domestic and foreign scholars have successively proposed radars based on compressed sensing systems, SAR imaging methods under sparsity constraints, SAR imaging methods for complex ground scenes based on CS, and SAR imaging methods based on CS, etc., applying CS theory to radar imaging detection technology, using traditional matched filtering methods in the range direction, and obtaining downsampled echo data through random sparse sampling in the azimuth direction, and then using CS theory to achieve azimuth focusing to obtain SAR images, and then using DPCA methods in the image domain to suppress clutter. For the case where the scene is not sparse, a weighted compressed sensing imaging algorithm is proposed to separate the clutter energy support area from the target support area, and set the weight coefficient of the clutter area to zero, so as to effectively suppress the clutter. However, this method is sensitive to noise and cannot accurately restore the scene image under low signal-to-noise ratio conditions.

[0004] SAR imaging based on CS can effectively reduce the amount of sampled data and alleviate the burden of data storage and transmission. When the sampling rate does not satisfy the Nyquist sampling theorem, it can still achieve unambiguous imaging, which is beneficial for imaging of large surveying strips. Moreover, since there is no need for matched filtering, there are no side lobes, which can effectively improve the radar resolution. Summary of the invention

[0005] The technical problems to be solved by the present invention are:

[0006] Aiming at the problem that background noise has a significant impact on the reconstruction accuracy of traditional compressed sensing methods during ground moving target imaging and detection, the present invention proposes a two-dimensional compressed sensing SAR imaging and moving target detection method based on scattering coefficient weighting.

[0007] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0008] A two-dimensional compressed sensing moving target imaging detection method based on scattering coefficient weighting, characterized by comprising:

[0009] Acquire SAR echo data of the first channel and the second channel, and perform downsampling processing on the SAR echo data of the first channel and the second channel;

[0010] Based on the SAR echo data of the first channel after downsampling processing, a first channel reconstruction matrix is ​​constructed; based on the SAR echo data of the second channel after downsampling processing, a second channel reconstruction matrix is ​​constructed;

[0011] The scene scattering coefficient of the first channel is reconstructed based on the weighted two-dimensional compressed sensing imaging algorithm and the first channel reconstruction matrix; the scene scattering coefficient of the second channel is reconstructed based on the weighted two-dimensional compressed sensing imaging algorithm and the second channel reconstruction matrix;

[0012] Compensating the scene scattering coefficient of the second channel with a fixed phase difference value to obtain a compensated scene scattering coefficient of the second channel;

[0013] Imaging the scene based on the scene scattering coefficient of the first channel, and imaging the scene based on the scene scattering coefficient of the second channel; obtaining two coherent images;

[0014] The DPCA method is used to suppress the stationary target in the image after dual-channel clutter cancellation and obtain the moving target.

[0015] A further technical solution of the present invention is: based on a weighted two-dimensional compressed sensing imaging algorithm, the scattering coefficient of the scene is reconstructed by solving the following optimization problem:

[0016]

[0017] in, is the scene scattering coefficient of the i-th iteration, W is the weight coefficient, is the actual echo signal vector, is the reconstruction matrix of the actual scene; the weight w of the next step i+1 It is the scene scattering coefficient generated by the previous iteration Calculated, that is ε is a very small positive number. The maximum number of iterations is selected according to the reconstruction accuracy. When the maximum number of iterations is met, the scattering coefficient of the scene is obtained.

[0018] A further technical solution of the present invention is as follows: the fixed phase difference value of the scene scattering coefficient compensation for the second channel is specifically:

[0019] Multiply the scene scattering coefficient of the second channel by a fixed phase difference

[0020] A computer system, characterized in that it comprises: one or more processors, and a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned method.

[0021] A computer-readable storage medium, characterized in that it stores computer-executable instructions, which are used to implement the above method when executed.

[0022] A computer program product, characterized by comprising computer executable instructions, wherein the instructions are used to implement the above method when executed.

[0023] The beneficial effects of the present invention are:

[0024] The invention provides a two-dimensional compressed sensing moving target imaging detection method based on scattering coefficient weighting, which realizes two-dimensional SAR imaging by downsampling in the range and azimuth directions and combining CS theory. The algorithm is closer to the original image by weighting during reconstruction. norm model, thereby improving the reconstruction accuracy under low signal-to-noise ratio conditions.

[0025] The present invention enables better moving target detection performance to be achieved when the amount of sampling data is small, and has high engineering application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like components throughout the drawings.

[0027] Figure 1It is a geometric relationship diagram of the dual-channel SAR to which the present invention is applicable;

[0028] Figure 2 The present invention is a process flow chart of the method. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0030] The present invention provides a two-dimensional compressed sensing moving target imaging detection method based on scattering coefficient weighting, comprising the following steps:

[0031] (1) Establish a dual-channel radar echo signal model;

[0032] (2) Two-dimensional compressed sensing imaging processing;

[0033] (3) Two-dimensional compressed sensing processing based on weighted optimization.

[0034] in:

[0035] Step (1) mainly includes the following steps:

[0036] a) Establish a dual-channel radar imaging geometric model;

[0037] b) Establish analytical expressions for the echo signals after dual-channel demodulation respectively.

[0038] Step (2) mainly includes the following steps:

[0039] a) Performing distance-azimuth two-dimensional division on the image scene;

[0040] b) downsampling the two-channel radar echo data;

[0041] c) The orthogonal matching pursuit (OMP) algorithm is used to reconstruct the two-channel scene scattering coefficients;

[0042] d) Fixed phase difference compensation.

[0043] Step (3) mainly includes the following steps:

[0044] a) calculating the weighted value using the scene scattering coefficient obtained iteratively;

[0045] b) reconstructing the dual-channel scene scattering coefficient by weighting the scattering coefficient;

[0046] c) After compensating the fixed phase difference, the DPCA method is used to suppress moving targets.

[0047] (1) Establishing a dual-channel radar echo signal model

[0048] like Figure 1 This is a geometric diagram of a dual-channel SAR. Channel 1 randomly transmits a linear frequency modulation signal along the flight direction. The time interval between its transmitted pulses, i.e., the PRF, is not fixed. Both channels receive echoes at the same time. m represents the azimuth slow time, assuming that the beam center is t m =0, d represents the distance between the two channels, the position of the point target P is x0, t m At the time point, the target P moves to P′, v r and v x They represent the range velocity and azimuth velocity of the point target, assuming that the acceleration effect is not considered; Δt i is the interval of the i+1th sampling in azimuth; R c Indicates the minimum slant distance between the target and the platform; R(t m ) represents the instantaneous slant distance between the target and the platform; v a The carrier speed.

[0049] Depend on Figure 1 It can be obtained that the instantaneous slant distances from the point target to the two channels are:

[0050]

[0051]

[0052] After the echo data of the first channel is demodulated, the echo signal of a single target can be written as follows:

[0053]

[0054] Where σ is the target scattering coefficient, t is the distance fast time, γ is the emission modulation frequency, c is the speed of light, λ is the carrier wavelength, and w r (t) is the window function of the LFM signal, w a (t m ) is the azimuth window function. Similarly, the demodulated echo of the second channel can be obtained as:

[0055]

[0056] From the above analysis, it can be seen that the echo data under random sparse sampling conditions have the same form as the echo data under traditional sampling conditions, so the matched filtering method can still be used in the range direction, and in the azimuth direction, the scene SAR image can be reconstructed using the CS algorithm through a small number of downsampled measurement values.

[0057] (2) 2D Compressed Sensing Imaging Processing

[0058] The actual scene is divided into M points in the range direction and N points in the azimuth direction according to the range resolution and azimuth resolution, that is, the scene is divided into M×N grids. The targets in the scene are divided into different grids. The scattering coefficient of these M×N points is written as:

[0059]

[0060] The scattering coefficient in the scene is σ mn The instantaneous slant ranges from the point target to the two channels of the radar are:

[0061]

[0062] Among them, R min is the shortest distance from the radar to the scene, X min is the starting point of the scene orientation, v r is the distance velocity, v x is the azimuthal velocity, ρ r is the range resolution, ρ a is the azimuth resolution; d is the distance between the first channel and the second channel.

[0063] Therefore, the radar echo data of the first channel and the second channel are:

[0064]

[0065] The radar echo data s1(t,t m ) and s2(t,t m ) for downsampling, N r and N a Respectively represent the number of sampling points in the range and azimuth directions, then the echo of the kth point in the range and the lth point in the azimuth direction can be expressed in the following form:

[0066]

[0067] Among them,

[0068]

[0069] The φ1 obtained in formula (12) mn Transformed into N r N a ×1-dimensional column vector, and obtain the first channel reconstruction matrix Φ1:

[0070] Φ1=[φ1 11 …φ1 1N …φ1MN ] (14)

[0071] The φ2 obtained in formula (13) mn Transformed into N r N a ×1-dimensional column vector, and obtain the second channel reconstruction matrix Φ2:

[0072] Φ2=[φ2 11 …φ2 1N …φ2 MN ] (15)

[0073] According to equations (10) and (14), the echo data vector of the first channel can be written as:

[0074] S1=Φ1σ1 (16)

[0075] According to equations (11) and (15), the echo data vector of the second channel can be written as:

[0076] S2=Φ2σ2 (17)

[0077] According to CS theory, when the signal satisfies sparsity under a certain basis, the original signal can be restored with a high probability from a small number of measurements (much less than the number of samples required by the Nyquist sampling theorem). Since the sampling method is random sampling, the reconstruction matrix in equations (14) and (15) satisfies the restricted isometry property (RIP), while the strong scattering points in the scene meet the sparsity, and the weak scattering points have little effect on the detection of moving targets. Therefore, equations (16) and (17) meet the conditions for compressed sensing applications, which can be solved by solving the following minimum Norm, reconstruct the scene orientation image:

[0078]

[0079] Through equations (18) and (19), the orthogonal matching pursuit (OMP) algorithm is used to reconstruct the scene scattering coefficient of the first channel and the scene scattering coefficient of the second channel The scene scattering coefficient of the second channel is obtained Compensate for fixed phase differences caused by channel spacing Get the scene scattering coefficient of the second channel after compensation After compensation, the images of the two channels have good coherence. The DPCA method can be used to suppress the stationary targets in the scene image, while most of the moving target images are retained, thereby achieving the purpose of moving target detection.

[0080] When using the above OMP algorithm, when the spatial sparsity of the scattering points in the scene is not enough, less sampling data will lead to a large reconstruction error. In addition, the OMP algorithm is sensitive to noise. If the noise factor is taken into account, the randomness of the noise will also lead to a larger reconstruction error. A large reconstruction error will make the coherence of the two channel images worse, and the clutter suppression performance will decrease. At this time, the above method cannot accurately detect the moving target information. A robust reconstruction algorithm must be used to enable the scene scattering coefficient to be reconstructed more accurately when the scene is not sparse enough and there is noise, thereby improving the clutter suppression and moving target detection performance. Therefore, considering optimizing and improving the existing algorithm, a weighted compressed sensing imaging method is proposed. This method can reduce the reconstruction error to a certain extent when the scene is not sparse enough and there is noise, thereby improving the clutter suppression and moving target detection performance.

[0081] (3) Two-dimensional compressed sensing processing based on weighted optimization

[0082] Considering that in the compressed sensing algorithm, due to solving The norm is an NP-hard problem, so when the perception matrix satisfies the RIP property, it can be used Norm Replacement Norm. Based on The optimization model of the norm imposes different penalties on large and small coefficients. Large coefficients have a greater impact on the optimization reconstruction, so the penalty imposed is larger. Small coefficients have a smaller impact on the optimization reconstruction, so the penalty imposed is smaller. In order to balance the impact of each coefficient on the optimization reconstruction, Candes et al. proposed a weighted Norm model, weighted The norm model can better approximate The weighted two-dimensional compressed sensing imaging algorithm (WCS algorithm) reconstructs the scattering coefficient of the scene by solving the following optimization problem:

[0083]

[0084] in, is the scene scattering coefficient of the i-th iteration, W is the weight coefficient, is the actual echo signal vector, is the reconstruction matrix of the actual scene. The weight w of the next step i+1 It is the scene scattering coefficient generated by the previous iteration Calculated, that is ε is a very small positive number. The maximum number of iterations is selected according to the reconstruction accuracy. When the maximum number of iterations is met, the scattering coefficient of the scene is obtained. The specific process is as follows Figure 2 .

[0085] The scene scattering coefficient of the first channel is reconstructed using the weighted reconstruction algorithm described above. and the scene scattering coefficient of the second channel This algorithm can reduce the reconstruction error to a certain extent when the scene is not sparse enough and there is noise, so the scene scattering coefficient of the second channel is Compensate for fixed phase differences caused by channel spacing After that, the images of the two channels have good coherence. This improves the clutter suppression performance. The offset phase center antenna (DPCA) method can suppress the static targets in the scene image, while most of the moving target images are retained.

[0086] The above description is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should be included in the protection scope of the present invention.

Claims

1. A two-dimensional compressed sensing moving target imaging detection method based on scattering coefficient weighting, characterized in that: include: Acquire SAR echo data of the first channel and the second channel, and perform downsampling processing on the SAR echo data of the first channel and the second channel; Based on the SAR echo data of the first channel after downsampling processing, a first channel reconstruction matrix is ​​constructed; based on the SAR echo data of the second channel after downsampling processing, a second channel reconstruction matrix is ​​constructed; The scene scattering coefficient of the first channel is reconstructed based on the weighted two-dimensional compressed sensing imaging algorithm and the first channel reconstruction matrix; the scene scattering coefficient of the second channel is reconstructed based on the weighted two-dimensional compressed sensing imaging algorithm and the second channel reconstruction matrix; Compensating the scene scattering coefficient of the second channel with a fixed phase difference value to obtain a compensated scene scattering coefficient of the second channel; Imaging the scene based on the scene scattering coefficient of the first channel, and imaging the scene based on the scene scattering coefficient of the second channel; Two coherent images are obtained; The DPCA method is used to suppress the stationary target in the image after dual-channel clutter cancellation and obtain the moving target.

2. According to claim 1, a two-dimensional compressed sensing moving target imaging detection method based on scattering coefficient weighting is characterized in that: Based on the weighted two-dimensional compressed sensing imaging algorithm, the scattering coefficient of the scene is reconstructed by solving the following optimization problem: in, is the scene scattering coefficient of the i-th iteration, W is the weight coefficient, is the actual echo signal vector, is the reconstruction matrix of the actual scene; the weight w of the next step i+1 It is the scene scattering coefficient generated by the previous iteration Calculated, that is ε is a very small positive number. The maximum number of iterations is selected according to the reconstruction accuracy. When the maximum number of iterations is met, the scattering coefficient of the scene is obtained.

3. The two-dimensional compressed sensing moving target imaging detection method based on scattering coefficient weighting according to claim 1 is characterized in that: The fixed phase difference value is compensated for the scene scattering coefficient of the second channel, specifically: multiplying the scene scattering coefficient of the second channel by the fixed phase difference 4. A computer system, characterized in that include: One or more processors, and a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method of claim 1.

5. A computer-readable storage medium, characterized in that Computer executable instructions are stored, and when the instructions are executed, they are used to implement the method of claim 1.

6. A computer program product, characterized in that The invention comprises computer executable instructions, which are used to implement the method of claim 1 when being executed.

Images