Motorized platform foresight super-resolution imaging method combining structured sparse and Bayesian frameworks

Abstract

The invention provides a mobile platform foresight super-resolution imaging method combining a structured sparse and Bayesian framework, and relates to the technical field of remote sensing, and the method comprises the steps: building a high-dynamic platform foresight imaging signal processing model; receiving an echo signal, and obtaining a preprocessed signal of the echo signal in a two-dimensional time domain according to the echo signal and the high-dynamic platform forward-looking imaging signal processing model; and constructing a phase compensation factor, and obtaining a multi-domain joint phase correction signal according to the phase compensation factor and the preprocessed signal of the two-dimensional time domain. Expressing the multi-domain joint phase correction signal as a form of an over-complete dictionary and signal scattering coefficient convolution, and determining variable estimation results of the preprocessed signal of the two-dimensional time domain in all range gates; and generating a super-resolution foresight image of the two-dimensional scene according to variable estimation results under all range gates. The high-order motion of the maneuvering platform and the target structure with the block sparse characteristic are considered, and foresight super-resolution imaging of the maneuvering platform on the surface target scene with the structured sparse characteristic is achieved.

Description

Technical Field

[0001] The invention relates to the field of remote sensing technology, and in particular to a mobile platform forward-looking super-resolution imaging method combining structured sparse and Bayesian frameworks. Background Art

[0002] The forward-looking mode is an important imaging mode in the field of remote sensing detection. It can detect and image the area directly in front of the flight direction of the aircraft at all times and at a long distance. Therefore, this imaging mode has broad application prospects in radar imaging equipment for missiles and aircraft. It is worth mentioning that when the radar system uses forward-looking mode imaging, it can be used to detect and identify targets of interest to the user in the area directly in front. Compared with the single-pulse target detection method, the forward-looking mode effectively improves the detection and recognition accuracy, so forward-looking radar imaging has become a key research direction of radar imaging.

[0003] It should be noted that when the aircraft is in motion and the radar system uses forward-looking imaging, the aircraft's motion will affect the slant range (a parameter of radar imaging), which will result in the lack of targets in the radar imaging results. Therefore, in the process of radar imaging by equipment equipped with a radar imaging system, how to effectively achieve super-resolution imaging of the radar system when the equipment has a maneuvering trajectory has become a technical problem that needs to be solved urgently. Summary of the invention

[0004] The present invention provides a forward-looking super-resolution imaging method for a mobile platform combining structured sparsity and a Bayesian framework, so that the high-order motion of the mobile platform and the target structure with block sparsity characteristics are taken into consideration, thereby realizing forward-looking super-resolution imaging of a surface target scene with structured sparsity by the mobile platform.

[0005] To achieve the above technical objectives, the embodiments of the present invention adopt the following technical solutions:

[0006] In a first aspect, the present invention provides a method for forward-looking super-resolution imaging of a mobile platform combining structured sparse and Bayesian frameworks, the method comprising:

[0007] A signal processing model for forward-looking imaging of a high-dynamic platform is established; the echo signal is received, and the preprocessed signal of the echo signal in the two-dimensional time domain is obtained according to the echo signal and the signal processing model for forward-looking imaging of a high-dynamic platform. A phase compensation factor is constructed, and a multi-domain joint phase correction signal is obtained according to the phase compensation factor and the preprocessed signal in the two-dimensional time domain. Among them, the compensation factor is used to compensate for the high-order phase terms and range walk correction introduced by the mobility of the high-dynamic platform. The multi-domain joint phase correction signal is represented as a convolution of an overcomplete dictionary and a signal scattering coefficient, and the variable estimation results of the preprocessed signal in the two-dimensional time domain at all range gates are determined; based on the variable estimation results under all range gates, a super-resolution forward-looking imaging of the two-dimensional scene is generated.

[0008] The embodiment of the present invention provides a forward-looking super-resolution imaging method for a mobile platform using a joint structured sparse and Bayesian framework, which takes into account the phase disturbance introduced by the high-order motion of the high-dynamic platform, constructs a virtual multi-domain joint phase compensation factor, improves the overcomplete dictionary, and accurately builds a linear regression model to characterize the forward-looking imaging geometry of the high-dynamic platform.

[0009] In combination with the first aspect, in a possible implementation, the echo signal includes at least one range gate signal. The above-mentioned method of expressing the multi-domain joint phase correction signal as a form of convolution of an overcomplete dictionary and a signal scattering coefficient, and determining the variable estimation results of the preprocessed signal in the two-dimensional time domain at all distance units includes: expressing the multi-domain joint phase correction signal as a form of convolution of an overcomplete dictionary and a signal scattering coefficient. Perform the following operations on each range gate signal respectively: characterize the convolution form as a vector form, and perform probabilistic modeling on the scattering coefficient and echo vector in vector form. Parameter estimation of the scattering coefficient vector is achieved through variational Bayesian inference and expectation maximization algorithm, and the variable estimation results under the range gate are determined. Traverse all distance units to obtain the variable estimation results at all distances in the two-dimensional scene.

[0010] It can be understood that, since the echo signal includes multiple range gate signals, in order to obtain an ultra-high resolution two-dimensional image, it is necessary to calculate the variable estimation of each range gate.

[0011] In combination with the first aspect, in another possible implementation, the method may further include: transmitting a radar signal, wherein the propagation direction of the radar signal is a forward imaging direction.

[0012] In combination with the first aspect, in another possible implementation manner, the parameter estimation of the scattering coefficient vector is realized by using variational Bayesian inference and expectation maximization algorithm, and the variable estimation result of the range gate is determined, including:

[0013] Based on the Bayesian framework, the forward-looking imaging problem is transformed into a Bayesian posterior probability solution problem, and a structured sparse prior model is constructed for the forward-looking imaging scene and noise to initialize the variables under the range gate. The variational Bayesian inference is used to calculate the posterior functions of the observed initialized variables and related latent variables. The expected maximum method is used to estimate the variable estimation results under the range gate and the update parameters of the related latent variables.

[0014] It is worth mentioning that based on the sparse Bayesian framework, the super-resolution forward-looking imaging problem is transformed into a maximum a posteriori probability problem. Combined with the physical meaning of forward-looking imaging of mobile platforms, reasonable and interpretable prior probability models are established for the sparse scenes to be solved and the observed echoes. Considering the structured sparse characteristics of surface targets, the variational Bayesian inference and the idea of ​​maximizing expectation algorithm are used to derive the analytical estimation of the parameters of each variable in the forward-looking scene, thereby improving the sparse reconstruction of sparse Bayesian learning and obtaining high-quality forward-looking remote sensing images with target detail information.

[0015] In combination with the first aspect, in another possible implementation manner, the multi-domain joint phase correction signal is represented as a convolution of an overcomplete dictionary and a signal scattering coefficient as follows: If the overcomplete dictionary is defined as: The convolutional model is expressed as: y = ψσ + ξ; where the overcomplete dictionary matrix is ​​defined as ψ = P a ⊙L,P a represents the triangular basis phase, and L represents the antenna pattern.

[0016] In a second aspect, an embodiment of the present invention further provides a mobile platform forward-looking super-resolution imaging device with a combined structured sparse and Bayesian framework, comprising: a signal modeling module, a preprocessing module, a matrix construction module, a parameter estimation module and a two-dimensional imaging module.

[0017] The signal modeling module is used to establish a high-dynamic platform forward-looking imaging signal processing model.

[0018] The preprocessing module is used to receive the echo signal and obtain the preprocessing signal of the echo signal in the two-dimensional time domain according to the echo signal and the high dynamic platform forward imaging signal processing model.

[0019] The matrix construction module is used to construct a phase compensation factor, and obtain a multi-domain joint phase correction signal based on the phase compensation factor and the preprocessed signal in the two-dimensional time domain; wherein the compensation factor is used to compensate for the high-order phase terms and range walk correction introduced by the mobility of the high dynamic platform.

[0020] The parameter estimation module is used to represent the multi-domain joint phase correction signal in the form of convolution of an over-complete dictionary and a signal scattering coefficient, and to determine the variable estimation results of the pre-processed signal in the two-dimensional time domain at all range gates.

[0021] The two-dimensional imaging module is used to generate super-resolution forward-looking imaging of the two-dimensional scene based on the variable estimation results of all range gates.

[0022] In conjunction with the second aspect, in a possible implementation manner, the echo signal is a signal including at least one range gate.

[0023] The parameter estimation module is specifically used to represent the multi-domain joint phase correction signal in the form of convolution of an overcomplete dictionary and a signal scattering coefficient; and perform the following operations on each range gate signal respectively:

[0024] The convolution form is represented as a vector form, and the scattering coefficient and echo vector in vector form are probabilistically modeled. The parameter estimation of the scattering coefficient vector is realized through variational Bayesian inference and expectation maximization algorithm, and the variable estimation result of the range gate is determined. All distance units are traversed to obtain the variable estimation results at all distances in the two-dimensional scene.

[0025] In combination with the second aspect, in another possible implementation, the forward-looking imaging device may further include a transmitting module, where the transmitting module is used to transmit a radar signal, and a propagation direction of the radar signal is a forward-looking imaging direction.

[0026] In combination with the second aspect, in another possible implementation, the parameter estimation module is specifically used to: convert the forward-looking imaging problem into a Bayesian posterior probability solution problem based on the Bayesian framework, and perform a structured sparse prior model on the forward-looking imaging scene and noise, and initialize the variables under the range gate. Use variational Bayesian inference to calculate the posterior functions of the observed initialized variables and related latent variables. Use the expected maximum method to estimate the variable estimation results under the range gate and the update parameters of the related latent variables.

[0027] In combination with the second aspect, in another possible implementation manner, the matrix construction module is specifically used to: express the multi-domain joint phase correction signal as a convolution of an overcomplete dictionary and a signal scattering coefficient as follows: If the overcomplete dictionary is defined as: The convolutional model can be expressed as: y = ψσ + ξ; where the overcomplete dictionary matrix is ​​defined as ψ = P a ⊙L,P a represents the triangular basis phase, and L represents the antenna pattern.

[0028] It can be understood that the device in the second aspect and any possible implementation manner provided above can refer to the beneficial effects in the first aspect and any possible design manner thereof, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A flowchart of an imaging method for a radar system provided by an embodiment of the present invention;

[0030] Figure 2 A high dynamic platform forward-looking super-resolution imaging coordinate provided by an embodiment of the present invention;

[0031] Figure 3 is a flow chart of another imaging method of a radar system provided by an embodiment of the present invention;

[0032] Figure 4 It is a schematic diagram of a distributed point marker in a simulation experiment provided by an embodiment of the present invention;

[0033] Figure 5 This is a reconstruction schematic diagram obtained by simulation under an original solution provided by an embodiment of the present invention;

[0034] Figure 6 It is a schematic diagram of reconstruction obtained by a radar forward-looking imaging method provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to facilitate the understanding of the solutions of the embodiments of the present invention, the special terms appearing in the embodiments of the present invention are first explained.

[0036] Dictionary matrix: can be understood as a matrix, that is, a set of numbers arranged in a rectangular array. In the embodiment of the present invention, the dictionary matrix is ​​used to characterize the target scattering coefficient during radar imaging.

[0037] Single-base Synthetic Aperture Radar (SAR): is a method of imaging based on single-track synthetic aperture radar data.

[0038] Bistatic SAR: It is a method of imaging using two SAR systems with different trajectories, mainly used to solve the limitation that traditional monostatic SAR cannot image the area directly in front.

[0039] Distributed targets: When the imaging area includes multiple independently distributed individuals, these independently distributed individuals become distributed targets.

[0040] Imaging area: During the radar imaging process, pulse signals are emitted, and the area covered by the pulse signals is the imaging area of ​​the radar system.

[0041] Doppler beam sharpening (DBS) method: An airborne pulse Doppler radar uses the Doppler effect to improve azimuth resolution through signal processing.

[0042] Sparse reconstruction theory: If the signal is compressible or sparse in a certain transform domain, then the signal can be sampled at a sampling rate much lower than the Nyquist sampling rate, and the original signal can be reconstructed with high probability by solving an optimization problem.

[0043] Sparsity Adaptive Matching Pursuit (SAMP): It is a signal reconstruction algorithm for compressed sensing. The SAMP method does not need to know the sparsity of the signal in advance. It gradually selects echo signals by setting a fixed step size and selects the echo signal that best matches the reconstructed signal from the candidate set through backtracking.

[0044] The radar imaging process in the embodiment of the present invention is described below.

[0045] As we all know, the imaging method of radar system belongs to the field of microwave remote sensing imaging. In some forward-looking imaging modes, high-resolution capability in the range direction can be achieved through pulse compression technology. In this implementation, the high-resolution performance of radar imaging is determined by the bandwidth of the transmitted pulse signal, and the azimuth resolution is determined by the aperture size of the antenna. Therefore, it is difficult to obtain satisfactory high-resolution images under this imaging method due to the size of the target aperture.

[0046] On the other hand, in order to obtain a two-dimensional high-resolution microwave remote sensing image, a virtual large aperture can be synthesized along the azimuth direction, and further, the SAR, DBS method and bistatic SAR method can be applied in remote sensing applications. It should be understood that for the application of monostatic SAR and DBS methods to radar systems, when the radar system adopts the forward-looking imaging mode, the pulse signal emitted by the radar propagates to the front of the device (i.e., the direction of movement of the device or the direction in front of the device), the Doppler history of the received echo signal is the same, the left and right Doppler blur, and the falling Doppler gradient will cause the virtual aperture to be unable to form, which will affect the resolution in the azimuth direction, thereby affecting the quality of the radar image obtained under forward-looking imaging.

[0047] In addition, for the application of bistatic SAR in radar systems, the transmitter and receiver in the radar system are designed as separate structures. In this case, the introduction of additional Doppler phase can make up for the problem that monostatic SAR cannot synthesize virtual aperture. However, due to the dual-platform structure of this radar system, problems such as time, frequency and space synchronization, as well as communication problems between multiple platforms, the application scenarios of bistatic SAR are limited. It is understandable that single-pulse imaging can achieve higher azimuth focusing capabilities, but the imaging system can only perform well in the case of a single beam and a single target. In the case of multiple targets, its application is limited.

[0048] In summary, among these forward-looking imaging modes, the single-channel system is simple and easy to implement, but its azimuth resolution is limited by the length of the real aperture in azimuth, the ambiguity between the left and right Dopplers, the low Doppler gradient, and the imaging quality in a multi-target imaging environment, making it unsuitable for use in radar imaging systems. Although the azimuth resolution can be improved in the radar system of the bistatic SAR, its engineering application is difficult, making it impossible to use it in actual radar systems and equipment.

[0049] Further, in order to solve the problem of limited lateral resolution in the single-base real beam forward-looking imaging system, in some embodiments, super-resolution imaging and sparse reconstruction theory are introduced into the single-base scanning imaging system. In this way, the forward-looking imaging problem can be converted into a linear regression problem, and then, the overcomplete dictionary matrix in the linear regression can be constructed by beam scanning, so as to realize super-resolution imaging of the forward-looking mode. Moreover, this greedy tracking method (a tracking algorithm name) represented by the sparsity adaptive matching pursuit algorithm (SAMP) has attracted more attention in the field of forward-looking imaging because of its simple configuration and fast tracking ability. However, this greedy tracking method based on greedy tracking has the problem of being easily trapped in local extreme values, and in most cases the greedy tracking method cannot find the global optimal value. In this case, if the imaging area in the forward-looking imaging mode includes distributed targets, the imaging result will affect the reconstruction performance of the distributed targets in the radar image.

[0050] It is worth mentioning that the sparse Bayesian learning method is a classic method in sparse reconstruction theory. Specifically, this method is based on the Bayesian framework, which equates the forward-looking imaging problem to the probability model parameter estimation problem, and models the forward-looking imaging scene based on the sparse Bayesian prior, giving the forward-looking reconstruction model a physically interpretable meaning. This unique processing idea can achieve better reconstruction performance than other recovery methods. However, for the forward-looking imaging system with trajectory maneuvers, the introduction of high-order motion will affect the slant range history, thereby destroying the traditional dictionary matrix, and then destroying the sparse recovery process, seriously affecting the reconstruction performance. Traditional sparse Bayesian learning is based on the assumption of "sparse pixels". In the reconstruction results of surface targets with block structures in the forward-looking scene, the image shows discontinuity and missing target details, which cannot reflect the real ground object information, seriously affecting the subsequent target recognition application.

[0051] It is understandable that when the radar system detects and images in the forward-looking mode, the lateral resolution within the imaging area of ​​the forward-looking scene is related to the real aperture length of the radar antenna in azimuth. During the geometric construction process of the forward-looking mode, when the emission direction of the radar beam is the same as the movement direction of the device carrying the radar system, the left and right sides of the imaging scene will have the same spatial cone angle, causing the left and right Doppler ambiguity problem in the forward-looking mode.

[0052] In some implementations, super-resolution imaging technology is based on the real beam scanning method of a single-base radar, making full use of the target area information contained in the scanned echo sequence, inverting the echo sequence from the data domain to the target domain, and then obtaining an angular resolution that exceeds the real aperture beam width. In forward-looking radar imaging, super-resolution technology has become an important research direction.

[0053] In essence, the azimuth scanning echo of the moving platform can be converted into a convolution model of the target scattering rate azimuth distribution function and the antenna pattern. Theoretically, from the perspective of imaging mechanism, it is feasible to use the convolution inversion method to achieve azimuth high-resolution imaging. Specifically, the forward-looking radar imaging model can be organized into a linear regression model. Under the premise of sparsity assumption, relying on the sparse microwave imaging processing framework, super-resolution sparse reconstruction of forward-looking imaging can be achieved. In addition, it can be well compatible with existing real beam scanning radar systems and working modes, with low cost and high efficiency.

[0054] In some embodiments, sparse Bayesian learning with sparse pixels can be used to realize forward imaging of the target area based on the carrier platform. Specifically, first, a geometric model and a signal processing model of the airborne platform scanning imaging are established, and distance processing is realized by pulse compression. Further, the preprocessed signal is represented as a convolution form of the target discrete points and the antenna pattern. For example, the known observation value is y, and the reconstruction problem x of the forward-looking scene is represented in the form of a linear regression model. The convolution form is expressed as: y=Ax+n, to complete the construction of the measurement dictionary matrix. Furthermore, based on the Bayesian framework, the forward-looking scene is appropriately modeled a priori using Gaussian distribution, and the parameters of the relevant variables are estimated based on the maximum a posteriori method to restore the target to be reconstructed.

[0055] It should be noted that when considering the forward imaging of the target area of ​​the airborne platform, the influence of acceleration does not need to be considered. When constructing the Doppler overcomplete dictionary, the influence of three-dimensional acceleration is not considered. The constructed measurement matrix cannot accurately describe the forward imaging scene of the mobile platform, which leads to the mismatch of the sparse reconstruction kernel. In addition, the current forward imaging method based on sparse Bayesian learning is based on the assumption of pixel sparsity. The reconstruction result is discontinuous, which will lead to the loss of target detail information. Therefore, when there is a surface target in the scene, the sparse reconstruction effect of the traditional sparse Bayesian learning based on pixel sparsity will be greatly reduced, affecting subsequent detection applications.

[0056] Therefore, an embodiment of the present invention provides a forward-looking super-resolution imaging method for a mobile platform that combines structured sparse and Bayesian frameworks, considers the phase disturbance introduced by the high-order motion of the high-dynamic platform, constructs a virtual multi-domain joint phase compensation factor, improves the overcomplete dictionary, and accurately builds a linear regression model to characterize the forward-looking imaging geometry of the high-dynamic platform. Based on the sparse Bayesian framework, the super-resolution forward-looking imaging problem is transformed into a maximum a posteriori probability solution problem, and reasonable and interpretable prior probability models are established for the sparse scene to be solved and the observed echo in combination with the physical meaning of the forward-looking imaging of the mobile platform. Considering the structured sparse characteristics of the surface target, the variational Bayesian inference and the idea of ​​maximizing the expectation algorithm are used to derive the analytical estimation of the parameters of each variable in the forward-looking scene, thereby improving the sparse reconstruction of the sparse Bayesian learning and obtaining a high-quality forward-looking remote sensing image with target detail information.

[0057] The following is a detailed description of the forward-looking super-resolution imaging method for a mobile platform using a combined structured sparse and Bayesian framework provided in an embodiment of the present invention.

[0058] Please refer to Figure 1 , which is a flow chart of a method for forward-looking super-resolution imaging of a mobile platform using a combined structured sparse and Bayesian framework provided in an embodiment of the present invention, such as Figure 1 As shown, the method includes steps 101 to 105.

[0059] Step 101: Establish a high dynamic platform forward imaging signal processing model.

[0060] Specifically, the middle time and position of the high dynamic platform in the acquisition of echo data are used as reference, the projection point of the high dynamic platform position on the ground is used as the coordinate center, the horizontal equivalent velocity direction of the high dynamic platform is the y-axis, and the upward direction is the z-axis, to establish the high dynamic platform forward super-resolution imaging coordinates. Please refer to Figure 2 , is the high dynamic platform forward super-resolution imaging coordinate established according to the above coordinate center and direction. Figure 2 As shown, represents the maneuvering trajectory of the platform flight, the yellow shaded area represents the forward scanning imaging area, Q represents any point on the curved trajectory, and QG represents the projection point of Q on the ground plane, G represents the intersection of the center of the mobile platform beam and the ground illumination scene, and R s represents the scene center slant distance at the imaging center moment, H represents the flight height of the mobile platform at the imaging center moment, ω v represents the beam scanning angular velocity, represents the three-dimensional velocity of the high dynamic platform, Denotes the three-dimensional acceleration of the high dynamic platform, θ(t) = ω v t represents the instantaneous beam azimuth. R(t) represents the true slant range between the target and the maneuvering platform with the introduction of three-dimensional acceleration and the time-varying azimuth, which can be expressed as a vector The length of

[0061] As shown in formula 1:

[0062]

[0063] Among them, |·| represents the modulus operation on the vector, Indicates that the sum operation is performed in sequence for subscript n = 0, 1, 2, 3, 4, n represents the sum variable of the sum operation, k n They represent the n-order Maclaurin expansion coefficients of the precise true slant range history R(t) at t=0, which can be expressed by the following formula 2:

[0064]

[0065] Among them, n! represents the factorial from 1 to n, It means that the function takes the nth derivative with respect to the variable.

[0066] In order to conveniently express the McLaughlin expansion coefficients, three spatial angle information are constructed, and the spatial angle information is expressed by the following formula 3:

[0067]

[0068] At this time, the Maclaurin expansion coefficient can be expressed by the following formula 4:

[0069]

[0070] Step 102: receiving the echo signal, and obtaining a preprocessing signal of the echo signal in the two-dimensional time domain according to the echo signal and the high dynamic platform forward imaging signal processing model.

[0071] For example, assuming that the radar transmits a linear frequency modulated pulse signal at a preset pulse repetition frequency, the pulse signal can be expressed by the following formula 5:

[0072]

[0073] Among them, t r represents the distance-to-fast time variable, T p represents the pulse width, f0 represents the radar carrier frequency, κ a represents the frequency modulation slope, λ represents the wavelength, rect(·) represents the rectangular window function, exp{·} represents the complex exponential function, π represents pi, and c represents the speed of light.

[0074] Specifically, the radar receiver receives the echo signal and performs carrier frequency removal processing to generate a demodulated baseband echo signal, and obtains a pulse pressure signal after matched filtering in the range direction. The pulse pressure signal can be expressed in the two-dimensional time domain by the following formula 6:

[0075]

[0076] Where l(·) represents the antenna pattern function, σ represents the backscatter coefficient, j represents the imaginary unit, θ(t) represents the instantaneous scanning angle between the target and the platform, and B represents the bandwidth.

[0077] Step 103: construct a phase compensation factor, and obtain a multi-domain joint phase correction signal according to the phase compensation factor and the preprocessed signal in the two-dimensional time domain.

[0078] Among them, the compensation factor is used to compensate for the high-order phase terms and range walk correction introduced by the high dynamic platform maneuverability.

[0079] First, a high-order motion phase compensation function is constructed in the two-dimensional time domain. The compensation function can be expressed as follows:

[0080] Formula 7 indicates:

[0081]

[0082] Among them, k 2a ,k 3a ,k 4a It means that the coefficients of the McLaughlin expansion contain high-order terms of acceleration, which can be specifically expressed by the following formula 8:

[0083]

[0084] At this time, after compensating for the high-order phase terms, the phases in the range frequency domain and slow time domain can be expressed as follows:

[0085] Formula 9 is expressed as:

[0086]

[0087] Among them, k 2_rest ,k 3_rest ,k 4_restBoth indicate that the coefficients of the Maclaurin expansion do not include the coefficient term of acceleration.

[0088] The distance movement compensation factor is constructed in the distance frequency domain and slow time domain. The distance compensation factor is expressed as follows:

[0089] Formula 10 represents:

[0090]

[0091] After completing high-order phase compensation and range movement correction in multi-domain, the preprocessed signal in the two-dimensional time domain is expressed as follows:

[0092]

[0093] Among them, R rw (t) = R S +k 2_rest t 2 +k 3_rest t 3 +k 4_rest t 4 , R pa (t) = R S +k1t+k 2_rest t 2 +k 3_rest t 3 +k 4_rest t 4 They respectively represent the slope history in the envelope and phase terms after multi-domain joint phase compensation.

[0094] Step 104: The multi-domain joint phase correction signal is represented as a convolution of an over-complete dictionary and a signal scattering coefficient, and variable estimation results of the two-dimensional time domain preprocessed signal at all range gates are determined.

[0095] The echo signal includes at least one range gate signal. In order to generate a high-resolution radar image, each range gate signal is processed separately to obtain a corresponding variable estimation result, so as to finally generate a two-dimensional high-resolution radar image.

[0096] The multi-domain joint phase correction signal is represented as the convolution of an overcomplete dictionary and the signal scattering coefficient. The following operations are performed on each range gate signal: the convolution form is represented as a vector form, and the vector scattering coefficient and echo vector are probabilistically modeled. The parameter estimation of the scattering coefficient vector is realized through variational Bayesian inference and expectation maximization algorithm, and the variable estimation result under the range gate is determined. All range cells are traversed to obtain the variable estimation results at all distances in the two-dimensional scene.

[0097] Specifically, after multi-domain joint phase correction, the signal y pre(τ, t) can be expressed as the convolution of the overcomplete dictionary and the signal scattering coefficient, as shown in the following formula 12:

[0098]

[0099] Among them, the overcomplete dictionary factor is defined as

[0100] Considering the influence of noise, the above convolution model can be represented in vector form:

[0101] y=ψσ+ξ Formula 13

[0102] Among them, the overcomplete dictionary matrix is ​​defined as ψ=P a ⊙L,P a represents the triangular basis phase, L represents the antenna pattern, and the two expressions are:

[0103]

[0104] Among them, R g (θ k ,t m ) represents the extension of the slant range in the azimuth direction, l h Represents the spatial sampling points of the antenna pattern vector, H represents the effective length of each antenna pattern vector, K represents the number of scattering points in the azimuth dimension, and M represents the number of sampling points in the azimuth dimension.

[0105] Specifically, for each range gate of the forward-looking scene scattering coefficient, a Bayesian posterior probability problem of the forward-looking scene scattering coefficient is constructed; using the student-t distribution idea, according to the forward-looking scene scattering coefficient observation value and the overcomplete dictionary matrix, the Bayesian posterior probability problem is transformed into the maximum posterior probability problem based on the hierarchical student-t distribution; using variational Bayesian inference and expectation maximization idea to solve the parameters of related variables, super-resolution imaging of the current range gate is realized.

[0106] For example, the above-mentioned implementation method of estimating variables can be further divided into multiple implementation steps. Figure 3 , is a flow chart of another forward-looking imaging method of a radar system provided by an embodiment of the present invention. Figure 3 As shown, it includes steps 301 to 303.

[0107] S301. Based on the Bayesian framework, the forward-looking imaging problem is converted into a Bayesian posterior probability solution problem, and a structured sparse prior model is constructed for the forward-looking imaging scene and noise, and the variables under the range gate are initialized.

[0108] Specifically, the embodiment of the present invention takes the backscattering coefficient of each range gate in the area to be reconstructed as the random variable to be reconstructed, takes the preprocessed signal as the observation variable to construct a Bayesian posterior probability model, and performs structured sparse prior modeling on the forward imaging scene and noise, and initializes the relevant parameters.

[0109] First, the noise is sparsely modeled as a Student-t distribution,

[0110]

[0111] Among them, α represents the noise accuracy, the accuracy is the inverse of the noise, η represents the gamma degree of freedom, It means that the random variable ζ follows a gamma distribution with scale parameter a and shape parameter b. represents the gamma function, represents a Gaussian distribution.

[0112] At this time, the distribution function of the observed variable is expressed as:

[0113]

[0114] Where Λ=diag(λ1,…,λ m ,…λ M ) is an implicit function used to assist in improving the accuracy of each observed variable, diag(·) represents a diagonal matrix, and the accuracy α follows a gamma distribution and is expressed as:

[0115] p(α)=Gam(α|a1,a2) Formula 19

[0116] The degrees of freedom η are also sparsely modeled as a gamma distribution:

[0117] p(η)=Gam(η|c1,c2) ​​Formula 20

[0118] The sparse modeling of the backscattering coefficient variables to be reconstructed is expressed as:

[0119]

[0120] in, represents the kth element in Ξ, represents the structural factor; and the latent variable β is modeled as a gamma distribution, expressed as:

[0121]

[0122] S302. Calculate the posterior functions of observed variables and related latent variables using variational Bayesian inference.

[0123] The posterior function of all variables is estimated by minimizing the relative entropy Kullback-Leibler (KL) divergence between the true posterior distribution and the approximate posterior function. The KL divergence is defined as:

[0124]

[0125] The posterior function is approximated by:

[0126]

[0127]

[0128] Among them, Δ={σ,α,β,Λ,η} represents the set of random variables and hidden variables.

[0129] Compute the posterior of α:

[0130]

[0131] in, It means to find the expectation of f(ε) for the variables in Δ except ζ, ζ is one of the variables in the parameter set Δ, and const means the variable Irrelevant constants, Express expectations

[0132] The obtained posterior probability follows the gamma distribution and is expressed as:

[0133]

[0134] Calculating λ m The posterior:

[0135]

[0136] The obtained posterior probability follows the gamma distribution and is expressed as:

[0137]

[0138] Compute the posterior of η:

[0139]

[0140] The obtained posterior probability follows the gamma distribution and is expressed as:

[0141]

[0142] S303: Estimate the update parameters of the scattering coefficient and related latent variables using the expectation maximization method.

[0143] Compute the posterior of σ:

[0144]

[0145] The obtained posterior probability follows a Gaussian distribution and is expressed as:

[0146]

[0147] Among them, the variance Σ σ and mean μ σ Respectively expressed as:

[0148] Σ σ -1 =<α>ψ T <Λ>ψ+<Ξ>,μ σ =Σ σ <α>ψ T <Λ>y Formula 33

[0149] Because k Is β k With β k+j , j = ± 1, the related variables cannot be directly obtained using variational Bayesian to obtain their posterior probabilities. Therefore, the maximum expectation method is used to estimate the parameters. First, the E-step calculates the Q equation about β:

[0150]

[0151] In the M-step, the parameter β is updated by maximizing the Q equation. Therefore, the Q equation is calculated with respect to the parameter β k The first derivative of:

[0152]

[0153] At this time, update β k The suboptimal solution is:

[0154]

[0155] Step 105: Generate a super-resolution forward-looking image of the two-dimensional scene based on the variable estimation results under all range gates.

[0156] The forward-looking imaging payload is mounted on a mobile platform, and the motion trajectory of the platform is a curved trajectory, which is specifically manifested by the presence of three-dimensional acceleration in the motion parameters. The scene contains targets with block-sparse characteristics. The forward-looking imaging system works in single-base scanning mode, and the main radar system parameters are shown in Table 1.

[0157] Table 1 Main radar parameters

[0158] parameter Numeric parameter Numeric Carrier frequency X-Band Velocity Vector [0,200,7]m / s Azimuth real aperture 0.55m Acceleration vector <![CDATA[[0.3,0.7,0.5]m / s 2 ]]> Transmit bandwidth 60MHz Beam scanning angular velocity 60degree / s Distance sampling points 2048 Beam scanning range -10～10° Reference distance 12km Pulse repetition frequency 1000Hz

[0159] like Figure 2As shown, in the front area (i.e. Figure 2 In the yellow shaded area on the ground in the figure, a surface target array with block sparse characteristics is arranged along the x-axis and y-axis directions, respectively, to verify the correctness and effectiveness of the forward-looking super-resolution imaging method based on a joint structured sparse and Bayesian framework mentioned in the present invention.

[0160] For example, in some simulation experiments, the parameters in Table 1 are used as an example for simulation experiments. Figure 4 The figure shows the original front. Figure 4 It can be seen that the imaging scene of the radar system includes a distributed dot matrix. In some implementations, it is reconstructed using the original technology, and the following can be obtained: Figure 5 The reconstruction result is shown in Figure 5 As shown, the restoration of the distributed dot matrix is ​​low, the position of the distributed dot matrix cannot be seen, and the resolution of the image is low. When the forward imaging mode provided by the embodiment of the present invention is adopted, the following can be obtained: Figure 6 The image shown, Figure 6 As shown, the structure of the distributed dot matrix can be clearly seen, and the image has high resolution.

[0161] In summary, although the original algorithm can achieve the reconstruction of the front view scene, the reconstruction effect of the surface target with block sparse characteristics is poor, the surface target structure is discontinuous, and the detail information is missing. The proposed algorithm can achieve super-resolution reconstruction of the surface target with block sparse characteristics and retain the structural information of the target.

[0162] The embodiment of the present invention also provides a mobile platform forward-looking super-resolution imaging device with a combined structured sparse and Bayesian framework, including: a signal modeling module, a matrix building module, a parameter estimation module and a two-dimensional imaging module.

[0163] The signal modeling module is used to establish a high-dynamic platform forward-looking imaging signal processing model. The matrix construction module is used to construct a phase compensation factor and an over-complete dictionary. The compensation factor is used to compensate for the high-order phase terms and distance movement correction introduced by the mobility of the high-dynamic platform. The parameter estimation module is used to perform structured sparse probability modeling of the forward-looking scene based on the Bayesian framework, estimate the posterior probability distribution of each variable according to the variational Bayesian derivation and expectation maximization algorithm, and determine the sparse reconstruction of the forward-looking scene. The two-dimensional imaging module is used to traverse all distance units, obtain the variable estimation results of the two-dimensional scene, and realize super-resolution forward-looking imaging of the two-dimensional scene.

[0164] In some implementations, the parameter estimation module is also used to: convert the forward-looking imaging problem into a Bayesian posterior probability solution problem based on a Bayesian framework, and perform a structured sparse prior model on the forward-looking imaging scene and noise; use variational Bayesian inference to calculate the posterior functions of the observed variables and related latent variables; and use the expectation maximization method to estimate the scattering coefficient and the update parameters of the related latent variables.

[0165] In another possible implementation manner, the matrix construction module is specifically used for: after performing multi-domain joint phase correction, the signal is represented as a convolution of an overcomplete dictionary and a signal scattering coefficient as follows: If the overcomplete dictionary is defined as: The convolutional model can be expressed as: y = ψσ + ξ; where the overcomplete dictionary matrix is ​​defined as ψ = P a ⊙L,P a represents the triangular basis phase, and L represents the antenna pattern.

[0166] The above contents are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions within the technical scope disclosed by the present invention 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 forward-looking super-resolution imaging of a mobile platform combining structured sparse and Bayesian frameworks, characterized in that: include: Establish a high dynamic platform forward imaging signal processing model; Receiving an echo signal, and obtaining a preprocessing signal of the echo signal in a two-dimensional time domain according to the echo signal and the high dynamic platform forward imaging signal processing model; Constructing a phase compensation factor, and obtaining a multi-domain joint phase correction signal according to the phase compensation factor and the preprocessed signal in the two-dimensional time domain; wherein the compensation factor is used to compensate for the high-order phase term and range walk correction introduced by the mobility of the high dynamic platform; The multi-domain joint phase correction signal is expressed as a convolution of an overcomplete dictionary and a signal scattering coefficient, and variable estimation results of the two-dimensional time domain preprocessed signal at all range gates are determined; A super-resolution forward-looking image of a two-dimensional scene is generated based on the variable estimation results of all the range gates.

2. The method according to claim 1, characterized in that The echo signal includes at least one range gate signal; The step of expressing the multi-domain joint phase correction signal in the form of convolution of an overcomplete dictionary and a signal scattering coefficient, and determining the variable estimation results of the preprocessed signal in the two-dimensional time domain in all distance units includes: The multi-domain joint phase correction signal is represented as the convolution of an overcomplete dictionary and the signal scattering coefficient; Perform the following operations for each range gate signal: The convolution form is represented as a vector form, and the scattering coefficient and echo vector in the vector form are probabilistically modeled; Implementing parameter estimation of the scattering coefficient vector through variational Bayesian inference and expectation maximization algorithm to determine the variable estimation result under the range gate; Iterate over all distance cells to obtain the variable estimation results at all distances in the two-dimensional scene.

3. The method according to claim 1 or 2, characterized in that: The method further comprises: A radar signal is transmitted, wherein the propagation direction of the radar signal is the forward imaging direction.

4. The method according to claim 3, characterized in that The method of implementing parameter estimation of the scattering coefficient vector by using variational Bayesian inference and expectation maximization algorithm and determining the variable estimation result of the range gate includes: Based on the Bayesian framework, the forward-looking imaging problem is converted into a Bayesian posterior probability solution problem, and a structured sparse prior model is constructed for the forward-looking imaging scene and noise, and the variables under the range gate are initialized; Calculating the posterior function of the observed initialized variables and related latent variables using variational Bayesian inference; The variable estimation result under the range gate and the update parameters of the related latent variables are estimated using the expectation maximization method.

5. The method according to any one of claims 1 to 2 and 4, characterized in that: The multi-domain joint phase correction signal is expressed as a convolution of an overcomplete dictionary and a signal scattering coefficient as follows: If the overcomplete dictionary is defined as: The convolution model is expressed as: y = ψσ + ξ; Among them, the overcomplete dictionary matrix is ​​defined as ψ=P a ⊙L,P a represents the triangular basis phase, and L represents the antenna pattern.

6. A mobile platform forward-looking super-resolution imaging device combining structured sparse and Bayesian frameworks, characterized in that: include: Signal modeling module, preprocessing module, matrix building module, parameter estimation module and two-dimensional imaging module; The signal modeling module is used to establish a high dynamic platform forward imaging signal processing model; The preprocessing module is used to receive the echo signal, and obtain a preprocessing signal of the echo signal in the two-dimensional time domain according to the echo signal and the high dynamic platform forward imaging signal processing model; The matrix construction module is used to construct a phase compensation factor, and obtain a multi-domain joint phase correction signal according to the phase compensation factor and the preprocessed signal of the two-dimensional time domain; wherein the compensation factor is used to compensate for the high-order phase term and range walk correction introduced by the mobility of the high dynamic platform; The parameter estimation module is used to represent the multi-domain joint phase correction signal in the form of convolution of an overcomplete dictionary and a signal scattering coefficient, and determine the variable estimation results of the preprocessed signal in the two-dimensional time domain at all range gates; The two-dimensional imaging module is used to generate super-resolution forward-looking imaging of a two-dimensional scene according to the variable estimation results of all the range gates.

7. The mobile platform forward-looking super-resolution imaging device of the combined structured sparse and Bayesian framework according to claim 6, characterized in that: The echo signal is a signal including at least one range gate; The parameter estimation module is specifically used to represent the multi-domain joint phase correction signal in the form of convolution of an overcomplete dictionary and a signal scattering coefficient; Perform the following operations for each range gate signal: The convolution form is represented as a vector form, and the scattering coefficient and echo vector in vector form are probabilistically modeled; the parameter estimation of the scattering coefficient vector is realized through variational Bayesian inference and expectation maximization algorithm, and the variable estimation result of the range gate is determined; all distance units are traversed to obtain the variable estimation results at all distances in the two-dimensional scene.

8. The mobile platform forward-looking super-resolution imaging device of the combined structured sparse and Bayesian framework according to claim 6 or 7, characterized in that: The matrix building module is specifically used for: The multi-domain joint phase correction signal is expressed as a convolution of an overcomplete dictionary and a signal scattering coefficient as follows: If the overcomplete dictionary is defined as: The convolution model is expressed as: y = ψσ + ξ; Among them, the overcomplete dictionary matrix is ​​defined as ψ=P a ⊙L,P a represents the triangular basis phase, and L represents the antenna pattern.

Images