Millimeter wave scattering aperture sparse three-dimensional imaging method and system

Abstract

The invention discloses a millimeter wave scattering aperture sparse three-dimensional imaging method and a millimeter wave scattering aperture sparse three-dimensional imaging system, and aims to realize high-resolution three-dimensional reconstruction through a sparse reconstruction algorithm aiming at a non-vision field target imaging problem under a rough scattering surface condition. The method comprises the following steps: firstly, improving the spectral resolution of an echo signal by using a time domain zero padding technology, accurately positioning a scattering aperture position based on spectrum peak estimation and completing distance compensation, converting a non-vision field echo into an equivalent vision field echo, and retaining complete information of a target; secondly, in combination with target space scattering point sparsity and scattering aperture sparse sampling characteristics, a virtual array and a measurement matrix are constructed, an imaging problem is converted into a sparse reconstruction task, and a target scattering coefficient is efficiently solved by adopting a generalized orthogonal matching pursuit algorithm; the dependence on a mirror reflection condition is broken through, and a non-vision target can be accurately reconstructed in a diffuse reflection scene; compared with an existing scattering aperture algorithm, the method has the advantage that the imaging resolution and the anti-noise capability are remarkably improved through double sparsity constraints. The method has important application value.

Description

Technical Field

[0001] The present invention relates to the field of radar imaging technology, and in particular to a millimeter wave scattering aperture sparse three-dimensional imaging method and system. Background Art

[0002] Non-line-of-sight imaging technology exploits the multipath effect of electromagnetic waves to image areas beyond the reach of direct signals, significantly extending the sensor's detection range. Research in millimeter-wave non-line-of-sight sensing can be divided into two categories: research based on through-wall signals and research based on the three-reflection model of scattering surfaces.

[0003] Jinshan Wei first proposed non-line-of-sight imaging for millimeter-wave radars in a two-dimensional scanning system. This research modeled the radar signal propagation path, assuming the scattering surface is an ideal mirror. He then proposed an improved backscattering imaging algorithm (a mirror-symmetric backscattering algorithm), which successfully reconstructed non-line-of-sight scenes. A subsequent mirror-projection autofocusing algorithm based on minimum entropy was proposed based on this algorithm. In addition to traditional imaging algorithms, researchers have also introduced compressed sensing into radar non-line-of-sight imaging, enabling three-dimensional reconstruction of non-line-of-sight scenes. This includes the Bayesian optimization-based non-line-of-sight imaging algorithm proposed by Xiang Cai et al.

[0004] Existing non-line-of-sight imaging algorithms assume that the relay surface is smooth and exploit mirror symmetry to transform the non-line-of-sight problem into a line-of-sight imaging task. While this approach simplifies the problem, it only works on smooth surfaces. However, on rough scattering surfaces, radar signals are primarily diffusely reflected, with the beam covering multiple scattering points, forming a virtual scattering aperture array. This results in different incident and reflection angles of the radar signal on the target surface, and multipath scattering occurs at multiple scattering points. This significantly degrades the performance of traditional non-line-of-sight imaging algorithms that rely on the mirror reflection assumption.

[0005] Therefore, a scattering aperture three-dimensional sparse imaging method and system applicable to rough scattering surfaces is needed. Summary of the Invention

[0006] In view of this, the object of the present invention is to provide a millimeter-wave scattering aperture sparse three-dimensional imaging method and system, which carries out a millimeter-wave non-line-of-sight sparse imaging method based on a scattering surface model.

[0007] In order to achieve the above object, the present invention provides the following technical solutions:

[0008] The millimeter wave scattering aperture sparse three-dimensional imaging method provided by the present invention comprises the following steps:

[0009] S1. Obtain a three-dimensional millimeter-wave radar echo signal, fill the three-dimensional millimeter-wave radar echo signal with zeros in the distance upward, and perform range pulse compression to obtain a range signal;

[0010] S2. Obtain the equivalent scattering aperture position by calculating the spectrum peak position based on the distance signal after pulse compression;

[0011] S3. Perform distance compensation in the time domain based on the estimated scattering aperture position to convert the non-line-of-sight signal into a line-of-sight signal;

[0012] S4. Sparse sampling of the slow time dimension of the distance-compensated three-dimensional echo;

[0013] S5. Calculate the virtual array spacing of the scattering aperture using the scan angles between the radar and the scattering surface in altitude and azimuth;

[0014] S6. Select the corresponding range slice, rearrange the echo signals, divide the imaging grid, and construct a measurement matrix corresponding to the distance unit;

[0015] S7. A sparse reconstruction model for scattering aperture imaging is established, and the generalized orthogonal matching pursuit (gOMP) algorithm is used to iterate the residuals and reconstruct non-viewing targets.

[0016] Furthermore, the process of obtaining the range direction signal by using the range pulse pressure in step S1 is specifically performed according to the following steps:

[0017] Step S11: Establish a three-reflection path model and define the round-trip delay from the radar through the scattering aperture to the non-line-of-sight target: the propagation delay between the radar and the scattering aperture is t1; the propagation delay between the scattering aperture and the non-line-of-sight target is t2; the phase function of the echo signal is decomposed into the line-of-sight phase φ1(t r ,t1) and the non-viewing area phase φ2(t r ,t2);

[0018] Step S12: Perform zero-filling operation on the radar echo signal in the time domain in the range direction: the original number of sampling points in the range direction is N, and the signal length after zero filling is N1 (N1>N); the spectrum resolution is increased to Δf'=f sr / N1, the range resolution is optimized to Δr'=c / (2B r )×N / N1; N is the number of sampling points in the range direction; N1 is the length of the signal after zero padding.

[0019] Furthermore, the spectrum peak position in step S2 is specifically determined in the following manner:

[0020] Step S21: Perform distance Fourier transform on the zero-padded signal to obtain the spectrum expression:

[0021]

[0022] Among them, K r is the range modulation frequency, n is the range unit index; the first sin c(·) function represents the position of the non-line-of-sight target focus; the second sin c(·) function represents the position of the scattering aperture;

[0023] Step S22: Determine the index position m of the scattering aperture by spectrum peak detection: Search for the maximum peak in the distance slice interval [n1, n2] close to the scattering surface to obtain the index m; calculate the actual distance between the radar and the scattering aperture, and derive the round-trip delay;

[0024] Step S23: Construct a range compensation filter to eliminate the phase φ1 (t r ,t1), retain the non-viewing area phase φ2(t r ,t2).

[0025] Furthermore, the time domain zero-padding operation in step S12 satisfies:

[0026] The length N1 of the signal after zero padding is an integer power of 2, where N1 is the length of the signal after zero padding;

[0027] The zero-padding position is the tail of the echo signal's time domain waveform.

[0028] Furthermore, the method for determining the distance slice interval [n1, n2] in step S22 is:

[0029] According to the prior position information of the scattering surface, the distance index range is selected;

[0030] The position of the reflection peak of the scattering surface is calibrated by actual measurement, and n1 and n2 are adjusted dynamically.

[0031] Furthermore, the measurement matrix in step S6 is determined in the following manner:

[0032] The echo signal matrix and the target scattering coefficient matrix are arranged into a one-dimensional vector form s = Ψe; the specific expression is as follows:

[0033]

[0034] Where s is the echo signal vector of size LM×1, Ψ is the measurement matrix of size LM×PQ, and e is the scattering coefficient vector.

[0035] Furthermore, the specific process of iterating the residual using the generalized orthogonal matching pursuit gOMP algorithm in step S7 is as follows:

[0036] Initialized residual vector r0=s, initialize index sequence number set Initialize the column collection Iteration number i = 1;

[0037] Calculate the inner product of each column of the observation matrix Ψ and the residual vector, select the K values ​​with the largest inner product, and form the column numbers corresponding to these values ​​into a set J0;

[0038] Update the corresponding index number set and column set, Λ i =Λ i-1 ∪J0,Ψ i =Ψ i-1 ∪a j (a j ∈Ψ);

[0039] Calculate the least squares solution ε i =(Ψ i T Ψ i ) -1 Ψ i T s, and update the residual vector r i =s-Ψ i ε i ;

[0040] Let i = i + 1. When the number of iterations is equal to the target sparsity estimate or the residual is less than the preset error value, stop the iteration and output ε = ε i ,r=r i , otherwise continue to iterate;

[0041] The compressed sensing model is s=Ψe, where s is a two-dimensional echo signal rearranged into a one-dimensional vector form, Ψ is a measurement matrix, and e is a scattering coefficient vector.

[0042] The millimeter-wave scattering aperture sparse three-dimensional imaging system provided by the present invention includes a signal transmitting module, a receiving module and a signal processing unit; the signal transmitting module is used to transmit a linear frequency-modulated continuous wave signal of 76 GHz to 81 GHz to provide an excitation source for target detection; the receiving module is used to receive the direct wave from the scattering surface and the echo signal of the scattering aperture three times reflection; the signal processing unit is used to perform high-speed sampling, low-noise amplification, mixing and filtering on the echo signal, and down-convert the high-frequency analog signal to an intermediate-frequency digital signal.

[0043] The millimeter wave scattering aperture sparse three-dimensional imaging system provided by the present invention includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the above method is implemented.

[0044] The beneficial effects of the present invention are:

[0045] The present invention discloses a millimeter-wave scattering aperture sparse three-dimensional imaging method and system. For the imaging problem of non-line-of-sight targets under rough scattering surface conditions, high-resolution three-dimensional reconstruction is achieved through a sparse reconstruction algorithm. The method first uses time-domain zero-filling technology to improve the spectral resolution of the echo signal, accurately locates the scattering aperture position based on spectrum peak estimation and completes distance compensation, converts the non-line-of-sight echo into an equivalent line-of-sight echo, and retains the complete information of the target; secondly, combining the sparsity of the target space scattering points and the sparse sampling characteristics of the scattering aperture, constructs a virtual array and measurement matrix, converts the imaging problem into a sparse reconstruction task, and adopts the generalized orthogonal matching pursuit (gOMP) algorithm to efficiently solve the target scattering coefficient. Compared with the traditional specular reflection model, the present invention breaks through the dependence on specular reflection conditions and can accurately reconstruct non-line-of-sight targets in diffuse reflection scenes; compared with the existing scattering aperture algorithm, the imaging resolution and noise resistance are significantly improved through the dual sparsity (target space + sampling dimension) constraints. This method has important application value.

[0046] Other advantages, objects, and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art upon examination of the following description or may be learned from practice of the present invention. The objects and other advantages of the present invention may be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to make the purpose, technical solutions and beneficial effects of the present invention more clear, the present invention provides the following drawings for illustration.

[0048] Figure 1 Flowchart of scattering aperture sparse 3D imaging.

[0049] Figure 2 Schematic diagram of the scattering aperture imaging model.

[0050] Figure 3 is the relationship between the radar sampling interval and the scattering aperture sampling interval.

[0051] Figure 4 is the non-line-of-sight range pulse pressure signal with artificial turf as the scattering surface.

[0052] Figure 5 is the scattering aperture position estimation result.

[0053] Figure 6 It is the distance pulse pressure signal after distance compensation.

[0054] Figure 7 This is the gOMP algorithm imaging result of the scattering aperture of the slice at a distance of 0.53m.

[0055] Figure 8This is the RMA algorithm imaging result of the slice at a distance of 0.53m. DETAILED DESCRIPTION

[0056] The present invention will be further described below with reference to the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it. However, the embodiments are not intended to limit the present invention.

[0057] like Figure 1 As shown, Figure 1 This is a flow chart of scattering aperture sparse 3D imaging. The millimeter wave scattering aperture sparse 3D imaging method provided in this embodiment includes the following steps:

[0058] S1. Obtain a three-dimensional millimeter-wave radar echo signal, fill the three-dimensional millimeter-wave radar echo signal with zeros in the distance upward to make the distance resolution reach the millimeter level, and then perform distance pulse compression;

[0059] S2. Obtain the equivalent phase center position of the scattering aperture by calculating the spectrum peak position based on the distance signal after pulse compression;

[0060] S3. Perform distance compensation in the time domain based on the estimated scattering aperture position to convert the non-line-of-sight signal into a line-of-sight signal;

[0061] S4. Sparse sampling of the slow time dimension of the distance-compensated three-dimensional echo;

[0062] S5. Recalculate the virtual array spacing of the scattering aperture using the scanning angles between the radar and the scattering surface in altitude and azimuth;

[0063] S6. Select the corresponding range slice, rearrange the echo signals, divide the imaging grid, and construct a measurement matrix corresponding to the distance unit;

[0064] S7. A sparse reconstruction model for scattering aperture imaging is established, and the generalized orthogonal matching pursuit (gOMP) algorithm is used to iterate the residuals and reconstruct non-viewing targets.

[0065] The method provided in this embodiment is based on a spectrum peak estimation algorithm with time-domain zero padding. The scattering aperture position is calculated and distance compensation is performed, thereby converting the non-line-of-sight imaging problem into a line-of-sight imaging problem. The sparsity of target scattering points in the imaging space and the sparse sampling characteristics of the scattering aperture are utilized to convert the non-line-of-sight imaging problem into a sparse reconstruction task, which is solved using the generalized orthogonal matching pursuit (gOMP) algorithm.

[0066] In this embodiment, after acquiring the three-dimensional echo signal of the millimeter-wave radar, the signal is zero-padded in the range direction to improve the range resolution to the millimeter level, and then range pulse compression is performed. The range pulse compression is a Fourier transform of the three-dimensional echo signal after frequency modulation is removed in the range direction.

[0067] The scattering aperture signal processing algorithm in this embodiment uses Fourier transform to perform range pulse compression on the echo signal, and uses time domain zero padding to improve the range resolution to accurately estimate the scattering aperture equivalent phase center position. Finally, a compensation filter is constructed to compensate the view echo signal.

[0068] In this embodiment, the spectral peak position is calculated based on the range signal after pulse compression to obtain the position of the equivalent phase center point of the scattering aperture. That is, based on the echo signal after the range pulse compression, the range index corresponding to the spectral peak is obtained, and the actual scattering aperture equivalent phase center point distance is calculated based on the improved range resolution.

[0069] In this embodiment, a measurement matrix is ​​constructed, that is, according to the actual position of the non-viewing area target, the corresponding echo range index is calculated, the corresponding range slice is taken, and the echo signal is rearranged to divide the imaging grid and construct a measurement matrix corresponding to the range unit.

[0070] In this embodiment, the spacing of the virtual array of scattering apertures is determined by recalculating the scanning angles between the radar and the scattering surface in the altitude and azimuth directions; that is, the array spacing of the scattering apertures is recalculated based on the actual antenna scanning spacing in the altitude and azimuth directions and the angle between the antenna motion trajectory and the normal vector of the scattering surface.

[0071] This embodiment generates sparse sampling data of a three-dimensional echo matrix by constructing a virtual scattering aperture array. The construction of the virtual scattering aperture array involves projecting the actual antenna array position onto the scattering surface in a Cartesian coordinate system, and constructing one-dimensional virtual scattering aperture array vectors in the elevation and azimuth directions based on the scattering aperture array spacing.

[0072] In this embodiment, the residual is iterated by the generalized orthogonal matching pursuit (gOMP) algorithm, and the specific process is as follows:

[0073] This embodiment is based on the compressed sensing model s = Ψe. Where s is the two-dimensional echo signal rearranged into a one-dimensional vector form, Ψ is the measurement matrix, and e is the scattering coefficient vector. The gOMP steps are as follows:

[0074] Initialized residual vector r0=s, initialize index sequence number set Initialize the column collection Iteration number i = 1;

[0075] Calculate the inner product of each column of the observation matrix Ψ and the residual vector, select the K values ​​with the largest inner product, and form the column numbers corresponding to these values ​​into a set J0;

[0076] Update the corresponding index number set and column set, Λ i =Λ i―1 ∪J0,Ψ i =Ψi―1 ∪a j (a j ∈Ψ);

[0077] Calculate the least squares solution ε i =(Ψ i T Ψ i ) ―1 Ψ i T s, and update the residual vector r i =s―Ψ i ε i ;

[0078] i=i+1, when the number of iterations is equal to the target sparsity estimate or the residual is less than the preset error value, the iteration stops and the output is ε=ε i ,r=r i , otherwise continue iterating.

[0079] Among them, Λ i Represents the index number set of the i-th iteration; a j represents the jth column in the observation matrix; ε i represents the least squares solution of the i-th iteration; Ψ i represents the observation matrix of the i-th iteration; r i represents the residual vector of the i-th iteration;

[0080] like Figure 2 As shown, Figure 2 The schematic diagram of the scattering aperture imaging model is shown in Figure 2. Due to the existence of the obstacle wall, the imaging scene is divided into the viewing area and the non-viewing area. The two-dimensional scanning millimeter wave radar R is located in the viewing area, and the target T nlos Located in the non-line-of-sight area, the radar transmit beam width is The electromagnetic wave strikes the rough scattering surface, forming a beam footprint of a certain size. This beam footprint is called the scattering aperture Ω0. The scattering aperture contains a number of scattering points, two of which are P1 and P2. In the scattering aperture, any scattering point can scatter the radar's signal into the entire non-line-of-sight area and can also reflect the echo from the non-line-of-sight area back to the radar's receiving antenna. Therefore, the three-reflection path of the scattering aperture imaging model is:

[0081]

[0082] According to the above formula, the radar is defined as The propagation delay between The time delay between them is t2, so the phase function of the radar signal is expressed as:

[0083]

[0084] Among them, t r Represents distance to fast time, phase function and They are introduced by the two-way delay of the radar signal propagating in the line-of-sight area and the non-line-of-sight area respectively. needs to be compensated, and the non-viewing target information only exists in the phase function Therefore, this method uses a distance compensation algorithm to remove the phase function

[0085] The original spectrum resolution Δf and range resolution Δr of the echo signal are:

[0086]

[0087] Among them, f sr is the range sampling rate, N is the number of range sampling points, c is the speed of light, B r is the signal bandwidth.

[0088] In order to improve the range resolution, it is necessary to perform time domain zero padding on the distance of the echo signal. The spectrum resolution Δf′ and range resolution Δr′ after zero padding are:

[0089]

[0090] Where N1 is the length of the zero-filled signal. Performing a distance Fourier transform on the echo signal yields:

[0091]

[0092] Among them, K r is the range modulation frequency, n is the range unit index;

[0093] The first sin c(·) function represents the position where the non-viewing target is focused;

[0094] The second sin c(·) function represents the position of the scattering aperture;

[0095] The index position that meets the following conditions can be considered as the range index corresponding to the scattering aperture position:

[0096]

[0097] Where n1 and n2 represent the starting and ending indices of the continuous range slices close to the scattering surface, and m is the index position corresponding to the spectrum peak. Therefore, the distance between the radar and the scattering aperture is m·Δr′. The round-trip delay between

[0098]

[0099] Based on the above derivation, a range compensation filter can be constructed to compensate for the echo signal in the viewing area. The formula of the compensation filter is as follows:

[0100] h(t r )=exp(―2jπf c t1―2jπK r tt1+jπK r t1 2 ), where f c Transmit signal carrier frequency.

[0101] Therefore, compared with formula (1), the propagation path of the scattering aperture imaging model can be rewritten as:

[0102]

[0103] Comparing Equation (1) and Equation (8), the echo signal received by the actual antenna array is converted into the echo signal received by the scattering aperture virtual array through accurate distance compensation.

[0104] Considering the two-dimensional antenna array, the three-dimensional echo signal after distance compensation can be expressed as:

[0105]

[0106] Where w(·) represents the range window function, K is the wave number, and R represents the scattering aperture position (x l ,y m ,0) and any scattering point (x t ,y t ,z t ) distance. σ is the target backscattering sparseness, and ∑0 represents the set of target scattering points.

[0107] In millimeter-wave near-field imaging scenarios, target scattering points are sparse relative to the imaging space, and the scattering aperture itself exhibits sparse sampling characteristics. Furthermore, to improve reconstruction efficiency, the echo signal is sparsely sampled in the slow time dimension. Therefore, these sparse characteristics all meet the requirements of sparse imaging.

[0108] Figure 3 The relationship between the radar sampling interval and the scattering aperture sampling interval is given. Assuming that the angle between the radar trajectory and the normal of the scattering surface is θ, and the radar sampling interval in the slow time dimension is Δx, then the sampling interval of the scattering aperture array is Then divide the imaging grid and perform range Fourier transform on Equation (9). Select a corresponding range slice to obtain the discrete echo form represented by the grid points:

[0109]

[0110] in, is the wave number behind the pulse pressure, and R(l,m,p,q) is the distance between the scattering aperture array sampling position index (l,m) and the scattering point position index (p,q).

[0111] Equation (10) is reconstructed into a matrix, and the echo signal matrix and the target scattering coefficient matrix are arranged into a one-dimensional vector form, which can be rewritten as:

[0112] s=Ψe (11)

[0113] Where s is the LM × 1 echo signal vector, Ψ is the LM × PQ measurement matrix, L and M are the number of sampling points of the scattering aperture in azimuth and elevation, respectively, and P and Q are the number of pixels of the target aperture in azimuth and elevation, respectively. e is the scattering coefficient vector.

[0114] The specific expression of formula (11) is as follows:

[0115]

[0116] Where S(L,M) represents the value of the echo signal at the array position index (L,M);

[0117] exp(―jkR(L,M,P,Q)) represents the value of the phase function at the corresponding array position index (L,M) and target space position index (P,Q);

[0118] σ(P,Q) represents the value of the scattering coefficient vector at the target spatial position index (P,Q);

[0119] Formula (11) is consistent with the standard model of sparse reconstruction, so the gOMP algorithm can be used to reconstruct the 3D non-viewing targets.

[0120] In this example, a millimeter-wave radar two-dimensional scanning system was built to verify the algorithm's imaging performance for non-line-of-sight targets under rough scattering surface conditions, and to illustrate the effectiveness of the scattering aperture three-dimensional sparse imaging method. The system consists of a millimeter-wave radar, a set of linear motion modules for two-dimensional scanning, and a host computer that controls radar acquisition and the movement of the linear motion modules.

[0121] The millimeter-wave radar used in this embodiment is the AWR2243BOOST evaluation board equipped with the DCA1000 module produced by Texas Instruments. The radar parameters are shown in Table 1.

[0122] Table 1. Millimeter-wave radar parameters

[0123] Transmit carrier frequency 77GHz Frequency modulation 79MHz / μs Sampling frequency 8MHz Distance sampling points 256 Pulse repetition frequency 50Hz Receive gain 30dB Distance resolution 0.0593m

[0124] The rough scattering surface selected in this embodiment is artificial turf with a 3mm undulation height, and the non-visual target is a smooth metal letter "E". The range pulse compression is performed on the echo signal after frequency modulation, and the range pulse compression result of one height slice is taken, such as Figure 4 As shown, the signal at a distance of 0.4 m represents the direct echo from the scattering surface, while the signal at a distance of 0.9 m represents the triple reflection echo from the non-line-of-sight target. Figure 5 The results of the scattering aperture position distribution are given. It can be seen that the scattering aperture position is distributed in three dimensions, which proves that the scattering of radar signals on the artificial turf surface is mainly diffuse reflection.

[0125] In order to convert the non-line-of-sight echo into the line-of-sight echo, the original echo signal is distance compensated to obtain Figure 6 , in which the direct echo of the scattering surface has been moved to the starting point of the range, and the echo of the non-line-of-sight target is completely retained. Therefore, the scattering aperture imaging algorithm transforms the radar imaging problem of the non-line-of-sight target in the range direction into the line-of-sight synthetic aperture radar imaging problem.

[0126] In order to obtain a sparse imaging model, the spacing of the scattering aperture virtual array is accurately calculated, and the measurement matrix is ​​obtained by cross-calculating the array position grid and the imaging space grid to establish a scattering aperture sparse imaging model.

[0127] This embodiment uses the gOMP algorithm to iteratively calculate the scattering coefficient of the non-viewing area target and completes the reconstruction of the metal letter "E". Figure 7 The slice imaging results at a distance of 0.53m are given. It can be seen that the algorithm proposed in this method reconstructs the non-viewing target more accurately, which verifies the effectiveness of the scattering aperture sparse imaging algorithm. Figure 8 The imaging results obtained by the traditional range migration algorithm (RMA) based on the specular reflection model are given. It can be seen that the algorithm has poor performance in reconstructing targets under the conditions of scattering aperture imaging scenes.

[0128] This embodiment provides a millimeter-wave scattering aperture sparse three-dimensional imaging system, comprising a signal transmitting module, a receiving module, a signal processing unit, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-described method when executing the program. The signal transmitting module is configured to transmit a linear frequency modulated continuous wave signal ranging from 76 GHz to 81 GHz, providing an excitation source for target detection. The receiving module is configured to receive direct waves from the scattering surface and echo signals reflected three times by the scattering aperture; the signal processing unit is configured to perform high-speed sampling, low-noise amplification, mixing, and filtering on the echo signals, thereby down-converting the high-frequency analog signal to an intermediate-frequency digital signal.

[0129] The above embodiments are merely preferred embodiments for the purpose of fully illustrating the present invention, and the scope of protection of the present invention is not limited thereto. Equivalent substitutions or modifications made by those skilled in the art based on the present invention are within the scope of protection of the present invention. The scope of protection of the present invention shall be subject to the claims.

Claims

1. A millimeter-wave scattering aperture sparse 3D imaging method, characterized by: The following steps are involved: S1. Obtain a three-dimensional millimeter-wave radar echo signal, fill the three-dimensional millimeter-wave radar echo signal with zeros in the distance upward, and perform range pulse compression to obtain a range signal; S2. Obtain the equivalent scattering aperture position by calculating the spectrum peak position based on the distance signal after pulse compression; S3. Perform distance compensation in the time domain based on the estimated scattering aperture position to convert the non-line-of-sight signal into a line-of-sight signal; S4. Sparse sampling of the slow time dimension of the distance-compensated three-dimensional echo; S5. Calculate the virtual array spacing of the scattering aperture using the scan angles between the radar and the scattering surface in altitude and azimuth; S6. Select the corresponding range slice, rearrange the echo signals, divide the imaging grid, and construct a measurement matrix corresponding to the distance unit; S7. A sparse reconstruction model for scattering aperture imaging is established, and the generalized orthogonal matching pursuit (gOMP) algorithm is used to iterate the residuals and reconstruct non-viewing targets.

2. The millimeter wave scattering aperture sparse 3D imaging method according to claim 1, wherein: The process of obtaining the range direction signal by using the range pulse pressure in step S1 is specifically performed according to the following steps: Step S11: Establish a three-reflection path model and define the round-trip delay from the radar through the scattering aperture to the non-line-of-sight target: the propagation delay between the radar and the scattering aperture is t1; the propagation delay between the scattering aperture and the non-line-of-sight target is t2; the phase function of the echo signal is decomposed into the line-of-sight phase φ1(t r ,t1) and the non-viewing area phase φ2(t r ,t2); Step S12: Perform zero-filling operation on the radar echo signal in the time domain in the range direction: the original number of sampling points in the range direction is N, and the signal length after zero filling is N1 (N1>N); the spectrum resolution is increased to Δf'=f sr / N1, the range resolution is optimized to Δr'=c / (2B r )×N / N1; N is the number of sampling points in the range direction; N1 is the length of the signal after zero padding.

3. The millimeter wave scattering aperture sparse 3D imaging method according to claim 1, wherein: The spectrum peak position in step S2 is specifically determined in the following manner: Step S21: Perform distance Fourier transform on the zero-padded signal to obtain the spectrum expression: Among them, K r is the range modulation frequency, n is the range unit index; the first sin c(·) function represents the position of the non-line-of-sight target focus; the second sin c(·) function represents the position of the scattering aperture; Step S22: Determine the index position m of the scattering aperture by spectrum peak detection: Search for the maximum peak in the distance slice interval [n1, n2] close to the scattering surface to obtain the index m; calculate the actual distance between the radar and the scattering aperture, and derive the round-trip delay; Step S23: Construct a range compensation filter to eliminate the phase φ1 (t r ,t1), retain the non-viewing area phase φ2(t r ,t2).

4. The millimeter wave scattering aperture sparse 3D imaging method according to claim 1, wherein: The time domain zero-padding operation in step S12 satisfies: The length N1 of the signal after zero padding is an integer power of 2, where N1 is the length of the signal after zero padding; The zero-padding position is the tail of the echo signal's time domain waveform.

5. The millimeter wave scattering aperture sparse 3D imaging method according to claim 3, wherein: The method for determining the distance slice interval [n1, n2] in step S22 is: According to the prior position information of the scattering surface, the distance index range is selected; The position of the reflection peak of the scattering surface is calibrated by actual measurement, and n1 and n2 are adjusted dynamically.

6. The millimeter wave scattering aperture sparse 3D imaging method according to claim 1, wherein: The measurement matrix in step S6 is determined in the following manner: The echo signal matrix and the target scattering coefficient matrix are arranged into a one-dimensional vector form s = Ψe; the specific expression is as follows: Where s is the echo signal vector of size LM×1, Ψ is the measurement matrix of size LM×PQ, and e is the scattering coefficient vector.

7. The millimeter wave scattering aperture sparse 3D imaging method according to claim 1, wherein: The specific process of iterating the residual using the generalized orthogonal matching pursuit gOMP algorithm in step S7 is as follows: Initialized residual vector r0=s, initialize index sequence number set Initialize the column collection Iteration number i = 1; Calculate the inner product of each column of the observation matrix Ψ and the residual vector, select the K values ​​with the largest inner product, and form the column numbers corresponding to these values ​​into a set J0; Update the corresponding index number set and column set, Λ i =Λ i-1 ∪J0,Ψ i =Ψ i-1 ∪a j (a j ∈Ψ); Calculate the least squares solution ε i =(Ψ i T Ψ i ) -1 Ψ i T s, and update the residual vector r i =s-Ψ i ε i ; Let i = i + 1. When the number of iterations is equal to the target sparsity estimate or the residual is less than the preset error value, stop the iteration and output ε = ε i ,r=r i , otherwise continue to iterate; The compressed sensing model is s=Ψe, where s is a two-dimensional echo signal rearranged into a one-dimensional vector form, Ψ is a measurement matrix, and e is a scattering coefficient vector.

8. Millimeter-wave scattering aperture sparse 3D imaging system, characterized by: It includes a signal transmitting module, a receiving module and a signal processing unit; the signal transmitting module is used to transmit a linear frequency modulated continuous wave signal of 76GHz to 81GHz to provide an excitation source for target detection; the receiving module is used to receive the direct wave of the scattering surface and the echo signal of the scattering aperture three times reflection; the signal processing unit is used to perform high-speed sampling, low-noise amplification, mixing and filtering on the echo signal, and down-convert the high-frequency analog signal to an intermediate-frequency digital signal.

9. A millimeter wave scattering aperture sparse three-dimensional imaging system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

Images