Radar sparse imaging method, system and storage medium for pulse noise environment

Abstract

The application discloses a radar sparse imaging method and system for a pulse noise environment and a storage medium, and is applied to the field of through-wall radar imaging technology. The existing through-wall radar sparse imaging technology is only applicable to image reconstruction under Gaussian noise. In view of non-Gaussian noise, the application proposes a building layout sparse imaging method under a pulse noise environment based on a maximum complex correlation entropy criterion. Firstly, in combination with a block projection matrix, an extended target compressive sensing imaging model under a pulse noise environment is established. Then, based on the maximum complex correlation entropy criterion, an imaging optimization equation for the pulse noise is constructed. Finally, a proximal gradient iteration algorithm is adopted to realize accurate reconstruction of the building layout. The application can reconstruct the building layout under the pulse noise environment and realize high-quality imaging under the condition that the data is partially available.

Description

Technical Field

[0001] This invention belongs to the field of through-wall radar imaging technology, and specifically relates to an imaging technology for pulse noise environments. Background Technology

[0002] In the field of through-wall radar imaging, building layout imaging, which can determine the internal structure and relative distribution of targets in unknown areas in advance, has been widely used in disaster relief, counter-terrorism and stability maintenance, and is particularly beneficial for action planning and command in the military field.

[0003] Generally, in the field of through-the-wall radar imaging, high-resolution imaging results require the use of large signal bandwidth and antenna arrays, which leads to the need to acquire and process a large amount of data. To alleviate the pressure of data acquisition and processing, compressed sensing has emerged, which can provide good image reconstruction results using only a small portion of the total data. In 2014, M. Leigsnering et al. treated walls and targets in the scene as point targets and achieved target and wall imaging in multipath and wall reverberation environments by utilizing group sparsity constraints (M. Leigsnering, F. Ahmad, M. Amin, et al., “Multipath exploitation in through-the-wall radar imaging using sparse reconstruction,” IEEE Trans. Aerosp. Electron. Syst., vol.50, no.2, pp.920-939, Apr.2014). In 2016, VHTang et al., based on the characteristic that walls are extended targets with continuous blocky features, assumed that indoor walls are parallel or perpendicular to the front wall, and constructed a sparse dictionary representing the wall positions as a sparse representation of the scene to achieve sparse imaging of wall-extended targets (VHTang, A. Bouzerdoum, SLPhung and F. HCTivive, "Indoor scene reconstruction for through-the-wall radar imaging using low-rank and sparsity constraints," 2016 IEEE Radar Conference (RadarConf), Philadelphia, PA, USA, 2016, pp. 1-4). However, existing through-wall radar imaging methods based on compressed sensing theory are all based on the assumption of Gaussian noise. In reality, non-Gaussian noise is widespread in practical engineering applications. Therefore, researching a sparse imaging method for building layouts suitable for impulse noise environments has significant practical implications. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a sparse imaging method for building layouts under impulse noise conditions, an electronic device, and a storage medium. By analyzing the blocky characteristics of the walls and introducing the maximum complex correlation entropy, a good image reconstruction effect can be obtained using only a small portion of all the data under impulse noise conditions.

[0005] One of the technical solutions adopted in this invention is: a method for sparse imaging of building layout under impulse noise environment, comprising:

[0006] S1. Establish the echo signal model; specifically, in an unknown region with multiple walls, a monostation through-wall radar system with integrated transceiver antennas moves M times at equal distances along the positive x-direction parallel to the walls, transmitting a stepped-frequency signal containing N frequency points at each location, denoted as z. m,n Let z represent the nth frequency signal received at the mth antenna position, where m = 1, 2, ..., M and n = 1, 2, ..., N. m,n Modeling as

[0007]

[0008] Where R is the number of walls, Let f be the scattering coefficient of the r-th wall surface. n For operating frequency, ε represents the round-trip propagation delay of the signal from the position of the m-th antenna to the surface of the r-th wall. m,n For noise;

[0009] S2. Considering the continuous block characteristics of the wall, a block projection dictionary is introduced, and an imaging model is constructed based on the relationship between the echo measurement vector, the dictionary matrix, and the scene vector.

[0010] S3. Based on the imaging model constructed in step S2, and combined with the mathematical and statistical characteristics of impulse noise, construct the objective function based on the complex correlation entropy criterion.

[0011] S4. Based on the objective function established in step S3, the scene vector is solved using the near-end gradient iteration algorithm to realize the reconstruction of the building layout.

[0012] The second technical solution adopted in this invention is: a sparse imaging system for building layout under impulse noise environment, comprising: a monostatic through-wall radar with integrated transceiver antenna, a signal modeling module, a sparse dictionary, an imaging model construction module, an objective function construction module, and a building layout reconstruction module; the monostatic through-wall radar with integrated transceiver antenna is used to move M times at equal distances along the positive x-direction parallel to the wall, and transmits a stepped frequency signal containing N frequency points at each position; the signal modeling module is used to model the frequency signals received by the monostatic through-wall radar with integrated transceiver antenna as:

[0013]

[0014] Among them, z m,n This represents the nth frequency signal received at the mth antenna position, where R is the number of wall surfaces. Let f be the scattering coefficient of the r-th wall surface. n For operating frequency, ε represents the round-trip propagation delay of the signal from the position of the m-th antenna to the surface of the r-th wall. m,n For noise;

[0015] The imaging model building module constructs an imaging model based on the modeled frequency signal and sparse dictionary;

[0016] The objective function construction module constructs the objective function based on the imaging model and the complex correlation entropy criterion.

[0017] The building layout reconstruction module uses a near-end gradient iteration algorithm to solve the objective function and realize the building layout reconstruction.

[0018] The sparse dictionary is used to represent the location of the wall.

[0019] The third technical solution adopted by the present invention is: an electronic device, including: a processor, a memory and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is running, the processor communicates with the memory through the bus, and when the machine-readable instructions are executed by the processor, the steps of a method for sparse imaging of building layout under impulse noise environment are performed.

[0020] The fourth technical solution adopted in this invention is: a computer-readable storage medium storing a computer program, which, when run by a processor, executes the steps of a method for sparse imaging of building layout under impulse noise environment.

[0021] The beneficial effects of this invention are as follows: This invention provides a sparse imaging method for building layouts under impulse noise conditions, enabling high-accuracy reconstruction of unknown scenes under both Gaussian and impulse noise. Specifically, firstly, based on the expansion characteristics of walls, a compressed sensing sparse imaging model for expanded targets is established by constructing a sparse dictionary representing wall positions; then, the maximum complex correlation entropy is introduced to establish an imaging optimization problem for impulse noise; finally, to solve this optimization problem, the proximal gradient descent method is introduced. The method of this invention has the following advantages:

[0022] 1. The method of the present invention can achieve superior reconstruction performance when the data portion is available;

[0023] 2. The method of the present invention effectively solves the problem of building layout reconstruction under impulse noise environment;

[0024] 3. This invention can be applied to fields such as surveying and mapping, disaster relief, etc. Attached Figure Description

[0025] Figure 1 The processing flow of the proposed method is described below.

[0026] Figure 2 This is a schematic diagram of through-wall radar imaging data acquisition.

[0027] Figure 3 This is for constructing the projection matrix R.

[0028] Figure 4 This is a schematic diagram of a simulation scenario.

[0029] Figure 5 This is a diagram showing the reconstruction results of the building layout under Gaussian noise.

[0030] in, Figure 5 (a) is the reconstruction result of the orthogonal matching pursuit method. Figure 5 (b) is the reconstruction result of the block sparse method. Figure 5 (c) is the reconstruction result of the proposed method.

[0031] Figure 6 This is a diagram showing the reconstruction results of the building layout under impulse noise.

[0032] in, Figure 6 (a) is the reconstruction result of the orthogonal matching pursuit method. Figure 6 (b) is the reconstruction result of the block sparse method. Figure 6 (c) is the reconstruction result of the proposed method.

[0033] Figure 7 This is a schematic diagram of a sparse imaging system for building layouts under impulse noise conditions. Detailed Implementation

[0034] To facilitate understanding of the technical content of this invention by those skilled in the art, the following description, in conjunction with the accompanying drawings, further illustrates the invention.

[0035] Example 1

[0036] This invention provides a method for sparse imaging of building layouts under impulse noise conditions, such as... Figure 1 As shown, it includes the following steps:

[0037] Step 1: Signal Model Construction

[0038] Consider as Figure 2 The diagram shows an unknown region with multiple walls. A monostatic through-wall radar system with an integrated transceiver antenna moves M times along a sampling path parallel to the walls in the horizontal direction, transmitting a stepped-frequency signal containing N frequency points at each location. Let z... m,n Let z represent the nth (n=1,2,...,N) frequency signal received at the mth (m=1,2,...,M) antenna position. m,n It can be modeled as

[0039]

[0040] Where R is the number of walls, Let f be the scattering coefficient of the r-th wall surface. n For operating frequency, ε represents the round-trip propagation delay of the signal from the position of the m-th antenna to the surface of the r-th wall. m,n It is noise.

[0041] The values ​​of M and N depend on the size of the detection scene and the radar system, respectively. In the specific implementation, M is 41 times and N is 301 frequency points.

[0042] Step 2: Imaging Model Construction

[0043] If we disregard the expansion characteristics of the walls, we can divide the imaging scene into a rectangular grid consisting of Q pixels, and arrange the image pixels into a vector. in Represents the complex field. The q-th pixel s in s. q The value of can be viewed as a weighted index function, representing the target reflectance coefficient, defined as follows:

[0044]

[0045] Based on the echo model, the wall imaging model can be constructed as follows:

[0046] z = Ψs + ε

[0047] in, zm =[z m,1 ,z m,2 ,...,z m,N ] T , Given an N×Q matrix and τ m,q Let ε be the round-trip propagation delay from the m-th antenna to the q-th grid, and let ε be the noise vector.

[0048] The value of Q is related to the specific application scenario and radar resolution. In actual implementation, the value of Q is 400.

[0049] However, when focusing on building layout imaging, walls, as an extended target, appear dense in traditional compressed sensing models. If the solution method is limited by the degree of freedom parameters, the reconstructed image can only include a portion of the walls and cannot recover the correct building layout structure. Therefore, the above model is improved as follows:

[0050]

[0051] in:

[0052] 1. If the dictionary matrix only considers the contribution of pixels located in front of each antenna, then Ψ is improved to... This represents the improved dictionary matrix corresponding to the m-th channel, where the dictionary matrix corresponding to the n frequency points of the m-th channel is represented as follows. The definition is as follows:

[0053]

[0054] 2. The dictionary R is a block projection matrix, constructed as follows: Figure 3 As shown. Each column of dictionary R represents a dictionary of length l. x The image of the wall consists of pixels, located within a specific azimuth and distance range in the image. Consider dividing the azimuth range into N... b Non-overlapping l x Pixel block (N) x =l x ×N b The distance is divided into N. y There are N blocks. b By l x The value of l determines x This represents the minimum desired wall length in the scenario. The dimension of R is N. x N y ×N b N y The product N b N yThis indicates the number of possible wall locations.

[0055] N b l x N y The value of N depends on the size of the scene and the resolution of the radar. In specific implementation, N b The value of l is 40. x The value of N is 5. y The value is 200.

[0056] 3. g is the wall image after block projection, determined by the following formula:

[0057]

[0058] Where B[b] represents the b-th azimuth block, and b = 1,...,N b .

[0059] The sampled data is represented as follows:

[0060]

[0061] Where y is the sampled echo measurement value, and z is the echo measurement value. For the sampling matrix, Let Φ be the real number field, K be the amount of data after downsampling, and each row of the sampling matrix Φ has only one non-zero element, representing the selected frequency of a specific antenna. Let Rg be a dictionary matrix, and Rg be the building layout image. It is impulse noise.

[0062] Step 3: Construct the objective function

[0063] As a nonlinear similarity measure, the complex correlation entropy is defined as:

[0064]

[0065] Where X and Y are complex random variables, κ σ (·) denotes a kernel function with a kernel width of σ, E[·] is the expectation equation, and X i ,Y i Let be the i-th element of X and Y.

[0066] The kernel function is defined as:

[0067]

[0068] in, *As the conjugate sign, the negative exponent term and kernel width involved in the complex correlation entropy can reduce the impact of large errors on the correlation entropy value. Therefore, the method based on the maximum complex correlation entropy has strong robustness to impulse noise or outliers.

[0069] To address the building layout reconstruction problem under impulse noise, the following objective function is established:

[0070]

[0071] Where ||·||1 is the L1 norm, and ||·|| F Let y represent the F-norm. i For the i-th row of y, f i Let θ be the i-th row, and λ be the regularization factor. The coefficients are omitted in the objective function above. It will not affect the subsequent solution.

[0072] Step 4: Inversion Imaging

[0073] Because the first term of the objective function Since the expression is smooth and differentiable, the second term λ||g||1 is non-smooth. Therefore, a proximal gradient descent algorithm is introduced to solve the above optimization problem. Let g... k Let represent the estimated value of the wall image in the k-th iteration. Then the estimated value in the (k+1)-th iteration can be solved using the following formula:

[0074]

[0075] in, α is the step size, and α includes the constant term in the derivative. Right now The original expression was This embodiment will Consider it as a whole as α.

[0076] The above equation represents a typical Lasso problem, which can be solved using a contraction operator:

[0077] g k+1 =shrink(u k ,αλ) (7)

[0078] Among them, shrink(y,x)=sgn(y)max(|y|-x,0).

[0079] The iteration stops when the maximum number of iterations (100) is reached or the numerical difference between adjacent iterations is less than 0.01, at which point the building layout reconstruction result is obtained; otherwise, the iteration continues. Furthermore, when σ→∞, the proposed algorithm transforms into a traditional compressed sensing solution model, i.e., it becomes a problem of reconstructing building layouts under Gaussian noise.

[0080] The following is a detailed implementation of the present invention based on a gprMax simulation example.

[0081] Configure gprMax electromagnetic simulation; see [link to gprMax electromagnetic simulation settings] for specific simulation scene size. Figure 4 The wall is composed of uniformly materialed walls, each with a relative permittivity of 6, a conductivity of 0.01 S / m, and a thickness of 0.1 m. The transmitted signal uses a stepped-frequency signal with a starting frequency of 1 GHz and an ending frequency of 3 GHz, offering 301 frequency points. A self-transmitting and self-receiving antenna performs synthetic aperture detection along a path parallel to the front wall, with 41 detection positions set along each path. The moving path is 1 m from the wall, and the moving interval is 0.1 m.

[0082] First, Gaussian noise with a signal-to-noise ratio of 10dB is added to the echo signal. The imaging result under Gaussian noise is as follows. Figure 5 As shown, Figure 5 (a) refers to the method proposed in "Greed is good: Algorithmic results for sparse approximation, IEEE Trans. Inform. Theory, vol. 50, no. 10, pp. 2231-2242, 2004". Figure 5 (b) The method proposed in "Determining building interior structures using compressive sensing, J.Electron.Imag., vol.22, no.2, pp.021003-021003, 2013.", Figure 5 (c) is the method proposed in this invention. Further, 10 dB of impulse noise is added to the echo signal; the imaging result under impulse noise is as follows. Figure 6 As shown in the figure. Through comparative analysis, it can be seen that the sparse imaging method for building layout under impulse noise environment proposed in this paper can not only effectively overcome the influence of Gaussian noise, but also adapt to the interference of impulse noise, thereby achieving high-quality building layout reconstruction and ensuring the accuracy and reliability of imaging results.

[0083] Example 2

[0084] like Figure 7As shown, this embodiment provides a sparse imaging system for building layout under impulse noise environment, including: a monostatic through-wall radar with integrated transceiver antenna, a signal modeling module, a sparse dictionary, an imaging model construction module, an objective function construction module, and a building layout reconstruction module; the monostatic through-wall radar with integrated transceiver antenna is used to move M times at equal distances along the positive x-direction parallel to the wall, and transmits a stepped frequency signal containing N frequency points at each position; the signal modeling module is used to model the frequency signals received by the monostatic through-wall radar with integrated transceiver antenna as:

[0085]

[0086] Among them, z m,n This represents the nth frequency signal received at the mth antenna position, where R is the number of wall surfaces. Let f be the scattering coefficient of the r-th wall surface. n For operating frequency, ε represents the round-trip propagation delay of the signal from the position of the m-th antenna to the surface of the r-th wall. m,n For noise;

[0087] The imaging model building module constructs an imaging model based on the modeled frequency signal and sparse dictionary;

[0088] The objective function construction module constructs the objective function based on the imaging model and the complex correlation entropy criterion.

[0089] The building layout reconstruction module uses a near-end gradient iteration algorithm to solve the objective function and realize the building layout reconstruction.

[0090] The sparse dictionary is used to represent the location of the wall.

[0091] Example 3

[0092] This embodiment provides a computer device, including a processor and a memory for storing processor-executable programs. When the processor of the computer executes the program stored in the memory, it realizes the task of sparse imaging of building layout under impulse noise environment as described in Embodiment 1 above.

[0093] Example 4

[0094] This embodiment provides a storage medium storing a program, which, when executed by a processor, performs the task of sparse imaging of building layout under impulse noise environment as described in Embodiment 1 above.

[0095] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the invention should be included within the scope of the claims of the invention.

Claims

1. A method for sparse imaging of building layouts under impulse noise environment, characterized in that, include: S1. Establish the echo signal model; specifically, in an unknown area with multiple walls, a monostation through-wall radar system with integrated transceiver antennas operates along a path parallel to the walls. Move equidistantly in the positive direction Each time, a transmission containing [the following] is launched at each location. The step frequency signal at each frequency point is denoted as... Indicates the first The first antenna position received the first... A frequency signal, , Then the signal Modeling as ； in, Number of walls For the first The scattering coefficient of a wall surface. For operating frequency, Indicates the first The position of the root antenna to the first The round-trip propagation delay of a signal on the surface of a wall For noise; S2. Considering the continuous block characteristics of the wall, a block projection dictionary is introduced to obtain the dictionary matrix. An imaging model is then constructed based on the relationship between the echo measurement vector, the dictionary matrix, and the scene vector. S3. Based on the imaging model constructed in step S2, and combined with the mathematical and statistical characteristics of impulse noise, construct the objective function based on the complex correlation entropy criterion. S4. Based on the objective function established in step S3, the scene vector is solved using the near-end gradient iteration algorithm to realize the reconstruction of the building layout.

2. The method for sparse imaging of building layout under impulse noise environment according to claim 1, characterized in that, The echo measurement value vector is represented as follows: , The superscript T indicates transpose.

3. The method for sparse imaging of building layout under impulse noise environment according to claim 2, characterized in that, The imaging model for step S2 is represented as follows: ； in, This is the echo measurement value. It is a dictionary matrix. For architectural layout images, For block projection matrix, This is a wall image after segmented projection. This is the noise vector.

4. The method for sparse imaging of building layout under impulse noise environment according to claim 3, characterized in that, The specific implementation process of step S3 is as follows: The echo measurement values ​​were sampled to obtain: ； in, These are the sampled echo measurements. For the sampling matrix, , It is impulse noise; Based on the definition of complex correlation entropy, construct the objective function: ； Where σ represents the kernel width, ||•||1 is the L1 norm, and ||•|| F Denotes the F-norm, for The OK, for The OK, This is the regularization factor.

5. The method for sparse imaging of building layout under impulse noise environment according to claim 4, characterized in that, The specific implementation process of step S4 is as follows: S41, Order Indicates the first The estimated value of the wall image in the nth iteration, then the nth The estimate for the next iteration is obtained by solving the following formula: ； in, , Step size; S42, This is a typical Lasso problem, solved using a shrinkage operator: ； in, ; S43. When the maximum number of iterations is reached or the numerical difference between adjacent iterations is less than 0.01, the iteration stops, thus obtaining the final building layout reconstruction result; otherwise, return to step S41.

6. A sparse imaging system for building layouts under impulse noise conditions, characterized in that, include: The system includes a monostatic through-wall radar with integrated transceiver antennas, a signal modeling module, a sparse dictionary, an imaging model construction module, an objective function construction module, and a building layout reconstruction module. The monostatic through-wall radar with integrated transceiver antennas is used for imaging along walls parallel to the ground. Move equidistantly in the positive direction Once, and at each location, it contains The signal modeling module is used to model the frequency signals received by the monostation through-wall radar with integrated transceiver antenna as follows: ； in, Indicates the first The first antenna position received the first... A frequency signal, Number of walls For the first The scattering coefficient of a wall surface. For operating frequency, Indicates the first The position of the root antenna to the first The round-trip propagation delay of a signal on the surface of a wall For noise; Considering the continuous blocky characteristics of the wall, a block projection dictionary is introduced to obtain the dictionary matrix. The imaging model construction module constructs the imaging model based on the relationship between the echo measurement vector, the dictionary matrix, and the scene vector. The objective function construction module constructs the objective function based on the imaging model and the complex correlation entropy criterion. The building layout reconstruction module uses a near-end gradient iteration algorithm to solve the objective function and realize the building layout reconstruction.

7. The building layout sparse imaging system under impulse noise environment according to claim 6, characterized in that, The sparse dictionary is used to represent the location of the wall.

8. An electronic device, characterized in that, include: The device includes a processor, a memory, and a bus, wherein the memory stores machine-readable instructions executable by the processor, and when the electronic device is in operation, the processor communicates with the memory via the bus, and the machine-readable instructions, when executed by the processor, perform the steps of the method according to any one of claims 1-5.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the method according to any one of claims 1-5.

Images