Method and apparatus for determining target scatter center structure type, and electronic device

Abstract

The embodiment of the application provides a kind of target scattering center structure type determination method, device and electronic equipment, method includes: the radar echo data of the target to be measured is sub-band segmented, obtains multiple sub-band data;Each sub-band data obtained is respectively carried out Fourier transform, obtains the one-dimensional sub-range image of each sub-band data;Each one-dimensional sub-range image obtained is spliced, obtains the one-dimensional range image of the radar echo data of the target to be measured;The one-dimensional range image of the radar echo data of the target to be measured is input into pre-trained structure type estimation model, obtains the type probability of each distance unit in one-dimensional range image;Wherein, structure type estimation model is obtained by training the neural network model constructed in advance;Based on type probability, determine the structure type of each scattering center in the target to be measured.The application can accurately determine the structure type of target scattering center.

Description

Methods, apparatus and electronic equipment for determining the structure type of target scattering center Technical Field

[0001] This invention relates to the field of target scattering technology, and in particular to a method, apparatus, and electronic device for determining the structure type of a target scattering center. Background Technology

[0002] The Geometrical Theory of Diffraction (GTD) is a mathematical model describing the mechanism of high-frequency electromagnetic scattering, providing the most accurate characterization of the scattering characteristics of a target in the high-frequency region. This model describes the frequency dependence of the scattering center amplitude as a half-integer power function of the frequency, with different values ​​of the power corresponding to different scattering center structures; therefore, this half-integer power is also known as the type parameter. Thus, by estimating the type parameter of the target's GTD model, the structural type of the target's scattering center can be determined.

[0003] However, as the dimension of the GTD model increases, although its accuracy in characterizing the electromagnetic scattering properties of the target improves, the computational complexity and the difficulty of parameter estimation also increase. Related techniques struggle to accurately estimate the type parameters of the GTD model, making it impossible to accurately determine the structural type of the target scattering center.

[0004] Therefore, there is an urgent need for a method, device, and electronic equipment for determining the structure type of the target scattering center to solve the above problems. Summary of the Invention

[0005] To address the problem that existing methods cannot accurately determine the structural type of a target scattering center, this invention provides a method, apparatus, and electronic device for determining the structural type of a target scattering center, which can accurately determine the structural type of the target scattering center.

[0006] In a first aspect, embodiments of the present invention provide a method for determining the structure type of a target scattering center, including:

[0007] The radar echo data of the target under test is divided into subbands to obtain multiple subband data.

[0008] Perform a Fourier transform on each sub-band data to obtain a one-dimensional sub-range image of each sub-band data;

[0009] Each one-dimensional sub-range image is stitched together to obtain a one-dimensional range image of the radar echo data of the target under test.

[0010] The one-dimensional range profile of the radar echo data of the target under test is input into a pre-trained structure type estimation model to obtain the type probability of each range unit in the one-dimensional range profile; wherein, the structure type estimation model is obtained by training a pre-constructed neural network model with the one-dimensional range profile of the known target radar echo data as input and the type probability of each range unit in the known target one-dimensional range profile as output.

[0011] Based on the type probability, the structure type of each scattering center in the target to be tested is determined.

[0012] In one possible design, the structural types include sharp-corner structures, curved-edge structures, straight-edge structures, single-curved surface structures, and planar structures.

[0013] In one possible design, the one-dimensional range profile of the known target radar echo data is obtained as follows:

[0014] Based on the known target's structure type, broadband radar echo data of the known target is generated using the GTD model;

[0015] The generated broadband radar echo data is divided into subbands to obtain multiple subband data.

[0016] Perform a Fourier transform on each sub-band data to obtain a one-dimensional sub-range image of each sub-band data;

[0017] Each one-dimensional sub-range image is stitched together to obtain a one-dimensional range image of the known target radar echo data.

[0018] In one possible design, the calculation formula for the GTD model is:

[0019]

[0020] In the formula, y n For radar echo data, k n Let k be the wave number. c The central wave number, f n f is the radar observation frequency. c Here, c is the center frequency, c is the speed of light, M is the number of scattering centers, and A is the center frequency. m r is the scattering amplitude coefficient of the m-th scattering center. m Let α be the radial position of the m-th scattering center. m For type parameters, e n To measure noise, N z This represents the total number of frequency points.

[0021] In one possible design, the Fourier transform of each sub-band data is performed using the following formula:

[0022]

[0023] In the formula, E i Let y be the one-dimensional distance image of the i-th sub-band data. i (n) represents the data of the i-th sub-band, r j Let J be the j-th distance cell of the distance image, where J is the total number of distance cells.

[0024] In one possible design, the structure type estimation model is determined as follows:

[0025] Construct a neural network model; wherein the neural network includes a first layer, a second layer, a third layer and a fourth layer connected in sequence, the first layer and the second layer have one channel each, the third layer has the same number of channels as the number of types of the structure, and the fourth layer is a Softmax layer;

[0026] The neural network model is optimized.

[0027] The optimized neural network model is trained using the one-dimensional range profile of known target radar echo data and the type probability of each range cell in the one-dimensional range profile of known target, to obtain the structure type estimation model.

[0028] In one possible design, optimizing the neural network model includes:

[0029] The neural network model is optimized using the cross-entropy loss function as the cost function and the stochastic gradient descent method.

[0030] Secondly, embodiments of the present invention also provide a device for determining the structure type of a target scattering center, comprising:

[0031] The segmentation module is used to segment the radar echo data of the target under test into sub-bands to obtain multiple sub-band data.

[0032] The transformation module is used to perform Fourier transform on each sub-band data to obtain a one-dimensional sub-range image of each sub-band data;

[0033] The stitching module is used to stitch together each obtained one-dimensional sub-range image to obtain a one-dimensional range image of the radar echo data of the target under test;

[0034] The input module is used to input the one-dimensional range profile of the radar echo data of the target under test into a pre-trained structure type estimation model to obtain the type probability of each range unit in the one-dimensional range profile; wherein, the structure type estimation model is obtained by training a pre-constructed neural network model with the one-dimensional range profile of the known target radar echo data as input and the type probability of each range unit in the known target one-dimensional range profile as output.

[0035] The determination module is used to determine the structural type of each scattering center in the target under test based on the type probability.

[0036] Thirdly, embodiments of the present invention also provide an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements the method described in any embodiment of this specification.

[0037] Fourthly, embodiments of the present invention also provide a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to perform the methods described in any embodiment of this specification.

[0038] This invention provides a method, apparatus, and electronic device for determining the structure type of a target scattering center. The method first constructs a structure type estimation model, then trains this model using known target samples to obtain a trained sparse estimation model. Thus, for radar echo data of the target, after preprocessing to obtain a one-dimensional range profile, and inputting this one-dimensional range profile into the structure type estimation model, the type probability of each range cell in the one-dimensional range profile can be obtained. Finally, based on these type probabilities, the structure type of each scattering center in the target can be determined. Therefore, this invention can accurately determine the structure type of a target scattering center. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 is a flowchart of a method for determining the structure type of a target scattering center according to an embodiment of the present invention;

[0041] Figure 2 is a hardware architecture diagram of an electronic device provided in an embodiment of the present invention;

[0042] Figure 3 is a structural diagram of a device for determining the structure type of a target scattering center according to an embodiment of the present invention. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0044] The following describes the specific implementation of the above concept.

[0045] Please refer to Figure 1. An embodiment of the present invention provides a method for determining the structure type of a target scattering center, the method comprising:

[0046] Step 100: Subband segmentation is performed on the radar echo data of the target to be measured to obtain multiple subband data;

[0047] Step 102: Perform Fourier transform on each sub-band data to obtain a one-dimensional sub-range image of each sub-band data;

[0048] Step 104: Stitch together each one-dimensional sub-range image to obtain a one-dimensional range image of the radar echo data of the target under test;

[0049] Step 106: Input the one-dimensional range profile of the radar echo data of the target to be tested into the pre-trained structure type estimation model to obtain the type probability of each range unit in the one-dimensional range profile; wherein, the structure type estimation model is obtained by training a pre-constructed neural network model with the one-dimensional range profile of the known target radar echo data as input and the type probability of each range unit in the known target one-dimensional range profile as output.

[0050] Step 108: Determine the structural type of each scattering center in the target under test based on the type probability.

[0051] This invention first constructs a structure type estimation model, then trains this model using known target samples to obtain a trained sparse estimation model. Thus, for radar echo data of the target, after preprocessing to obtain a one-dimensional range profile, and inputting this one-dimensional range profile into the structure type estimation model, the type probability of each range cell in the one-dimensional range profile can be obtained. Finally, based on these type probabilities, the structure type of each scattering center in the target can be determined. Therefore, this invention can accurately determine the structure type of a target's scattering center.

[0052] In some implementations, the structural types include sharp-corner structures, curved-edge structures, straight-edge structures, single-curved surface structures, and planar structures.

[0053] In some implementations, the one-dimensional range profile of the known target radar echo data is obtained as follows:

[0054] Step A1: Based on the known target's structure type, generate broadband radar echo data of the known target using the GTD model;

[0055] Step A2: The generated broadband radar echo data is divided into subbands to obtain multiple subband data.

[0056] Step A3: Perform Fourier transform on each sub-band data to obtain a one-dimensional sub-range image of each sub-band data.

[0057] Step A4: Stitch together each one-dimensional sub-range image to obtain a one-dimensional range image of the known target radar echo data.

[0058] Regarding step A1, in some implementations, the calculation formula for the GTD model is as follows:

[0059]

[0060] In the formula, y n For radar echo data, k n Let k be the wave number. c The central wave number, f n f is the radar observation frequency. c Here, c is the center frequency, c is the speed of light, M is the number of scattering centers, and A is the center frequency. m r is the scattering amplitude coefficient of the m-th scattering center. m Let α be the radial position of the m-th scattering center. m For type parameters, e n To measure noise, N z This represents the total number of frequency points.

[0061] In some implementations, the type parameter α m The values ​​for α include -1, -0.5, 0, 0.5, and 1, with different type parameters corresponding to different structure types. For example, when α... m =-1 represents a sharp-angled structure, α m =-0.5 represents a curved edge structure, α m =0 represents a straight edge structure, α m =0.5 represents a single-curved surface structure, α m=1 represents a planar structure. Therefore, for the target to be measured, as long as the type parameter of a certain scattering center is determined, the structural type of that scattering center can be determined.

[0062] Furthermore, the radar echo data generated based on the GTD model is broadband data. In order to accurately obtain the one-dimensional range profile of the echo data, it is necessary to divide the echo data into sub-bands and then generate a one-dimensional range profile for each sub-band.

[0063] Regarding step A2, in some implementations, subband segmentation is achieved in the following manner:

[0064] Let y n The total bandwidth is B z With bandwidth B z A segment of frequency domain data column vector y1 is extracted from the radar echo data y(n), with bandwidth B and center frequency f. c1 f c1 = f0 + B / 2, where f0 is the starting frequency corresponding to y(n).

[0065] Define the step center frequency as Δf c , with f c2 =f c1 +Δf c Let B be the center frequency and B be the bandwidth. The second segment of the frequency domain data column vector y2 is extracted. In some implementations, Δf is taken. c =Δf, where Δf is the radar step frequency, which achieves better smoothing. Of course, Δf c Other values ​​are also possible, as long as they do not exceed the preset frequency. This application is not limited to these values.

[0066] f ci =f c1 +(i-1)Δf (i=3,4,…) is the center frequency, and B is the bandwidth. Extract the i-th segment of frequency domain data column vector y. i .

[0067] And so on, until f cL =f c1 +(L-1)Δf is the center frequency, B is the bandwidth, and the Lth segment of frequency domain data column vector y is extracted. L ;in, And L is an integer.

[0068] For step A3, in some implementations, the Fourier transform of each sub-band data obtained is performed using the following formula:

[0069]

[0070] In the formula, Ei Let y be the one-dimensional distance image of the i-th sub-band data. i (n) represents the data of the i-th sub-band, r j Let J be the j-th distance cell of the distance image, where J is the total number of distance cells.

[0071] Regarding step A4, in some implementations, each one-dimensional distance image column vector E is... i By stitching together the data, a one-dimensional range image of the known target radar echo data is obtained: Input = […, E i [ , ...], this matrix is ​​the input to the structure type estimation model.

[0072] In some implementations, the type probability of each distance cell in the known one-dimensional range image of the target is used as the network label Label(j,k) of a pre-built neural network model. This network label is a J×6 sparse matrix. When the distance cell does not have a scattering center, the first column of the row is 1, and the remaining columns are 0. When the distance cell has a scattering center, the column corresponding to the type of that scattering center is 1, and the remaining columns are 0. The expression for this network label is as follows:

[0073] Label(j, k) = 1;

[0074] in,

[0075] In the formula, r d ≠r m This indicates that there is no scattering center in this range cell, r d =r m This indicates that a scattering center exists within this range cell. Furthermore, due to α in the GTD model... m The value of Label can only be one of five possibilities: -1, -0.5, 0, 0.5, and 1. Therefore, the value of each element of Label is an integer between 1 and 6.

[0076] In some implementations, the structure type estimation model is determined as follows:

[0077] Step B1: Construct a neural network model; wherein the neural network includes a first layer, a second layer, a third layer and a fourth layer connected in sequence. The first and second layers have one channel each, the third layer has the same number of channels as the number of structure types, and the fourth layer is a Softmax layer.

[0078] Step B2: Optimize the neural network model;

[0079] Step B3: Train the optimized neural network model using the one-dimensional range profile of the known target radar echo data and the type probability of each range cell in the one-dimensional range profile of the known target to obtain the structure type estimation model.

[0080] In step B1, the first and second layers are convolutional layers with 1 channel, a kernel size of 3*3, a stride of 1*3, and padding of (2, 0). After two convolutional layers, the output matrix Output2 is J×G dimensional. The third layer is a convolutional layer with 6 channels, a kernel size of 1*G, a stride of 1*1, and padding of (0, 0). The output matrix Output3 is J×6 dimensional.

[0081] The fourth layer is the Softmax layer. Its function is to transform each element in Output3 into a probability of a type, with values ​​between 0 and 1. The expression for its output matrix Output4(j, a) is:

[0082]

[0083] Regarding step B2, in some implementations, the neural network model is optimized, including:

[0084] The cross-entropy loss function is used as the cost function, and the stochastic gradient descent method is used to optimize the neural network model.

[0085] In this step, the expression for the cross-entropy loss function is:

[0086]

[0087] Where k is the column in the network label Label that is equal to 1, i.e., Label(j,k) = 1.

[0088]

[0089] Regarding step B3, in some implementations, the neural network model is trained based on a loss function. The network optimization method uses stochastic gradient descent, and the corresponding parameter update formula is:

[0090]

[0091] In the formula, θ t Let θ be the model parameters after the t-th iteration. t+1 Let represent the model parameters after the (t+1)th iteration, and η be the learning rate parameter.

[0092] Set a threshold of δ for the cost function. When Loss < δ, stop training the neural network model.

[0093] Finally, for step 108, for each row of Output4, the probability of each type of scattering center is taken. For any row in the Output4 matrix, if the element corresponding to its first column is the largest in that row, then there is no scattering center at that position; if the element corresponding to its second column is the largest in that row, then the type parameter of the scattering center in that row is determined to be -1, and the scattering center is a sharp-angled structure; if the element corresponding to its third column is the largest, then the type parameter of the scattering center in that row is determined to be -0.5, and the scattering center is a curved-edge structure; if the element corresponding to its fourth column is the largest, then the type parameter of the scattering center in that row is determined to be 0, and the scattering center is a straight-edge structure; if the element corresponding to its fifth column is the largest, then the type parameter of the scattering center in that row is determined to be 0.5, and the scattering center is a single-curved surface structure; if the element corresponding to its sixth column is the largest, then the type parameter of the scattering center in that row is determined to be 1, and the scattering center is a planar structure.

[0094] As shown in Figures 2 and 3, this embodiment of the invention provides a device for determining the structure type of a target scattering center. The device can be implemented in software, hardware, or a combination of both. From a hardware perspective, Figure 2 shows a hardware architecture diagram of the electronic device housing the device for determining the structure type of a target scattering center provided in this embodiment. Besides the processor, memory, network interface, and non-volatile memory shown in Figure 2, the electronic device in this embodiment may also include other hardware, such as a forwarding chip responsible for processing packets. Taking software implementation as an example, as shown in Figure 3, as a logical device, it is formed by the CPU of the electronic device reading the corresponding computer program from the non-volatile memory into memory and running it. This embodiment provides a device for determining the structure type of a target scattering center, including:

[0095] The segmentation module 300 is used to perform sub-band segmentation on the radar echo data of the target under test to obtain multiple sub-band data.

[0096] The transformation module 302 is used to perform Fourier transform on each sub-band data to obtain a one-dimensional sub-range image of each sub-band data;

[0097] The stitching module 304 is used to stitch together each obtained one-dimensional sub-range image to obtain a one-dimensional range image of the radar echo data of the target under test;

[0098] The input module 306 is used to input the one-dimensional range image of the radar echo data of the target under test into a pre-trained structure type estimation model to obtain the type probability of each range unit in the one-dimensional range image; wherein, the structure type estimation model is obtained by training a pre-built neural network model with the one-dimensional range image of the known target radar echo data as input and the type probability of each range unit in the known target one-dimensional range image as output.

[0099] The determination module 308 is used to determine the structural type of each scattering center in the target under test based on the type probability.

[0100] In this embodiment of the invention, the segmentation module 300 can be used to execute step 100 in the above method embodiment, the transformation module 302 can be used to execute step 102 in the above method embodiment, the splicing module 304 can be used to execute step 104 in the above method embodiment, the input module 306 can be used to execute step 106 in the above method embodiment, and the determination module 308 can be used to execute step 108 in the above method embodiment.

[0101] In some implementations, the structural types include sharp-corner structures, curved-edge structures, straight-edge structures, single-curved surface structures, and planar structures.

[0102] In some implementations, the one-dimensional range profile of the known target radar echo data is obtained as follows:

[0103] Based on the known target's structure type, broadband radar echo data of the known target is generated using the GTD model;

[0104] The generated broadband radar echo data is divided into subbands to obtain multiple subband data.

[0105] Perform a Fourier transform on each sub-band data to obtain a one-dimensional sub-range image of each sub-band data;

[0106] Each one-dimensional sub-range image is stitched together to obtain a one-dimensional range image of the known target radar echo data.

[0107] In some implementations, the calculation formula for the GTD model is as follows:

[0108]

[0109] In the formula, y n For radar echo data, k n Let k be the wave number. c The central wave number, f n f is the radar observation frequency. cHere, c is the center frequency, c is the speed of light, M is the number of scattering centers, and A is the center frequency. m r is the scattering amplitude coefficient of the m-th scattering center. m Let α be the radial position of the m-th scattering center. m For type parameters, e n To measure noise, N z This represents the total number of frequency points.

[0110] In some implementations, the Fourier transform of each sub-band data is performed using the following formula:

[0111]

[0112] In the formula, E i Let y be the one-dimensional distance image of the i-th sub-band data. i (n) represents the data of the i-th sub-band, r j Let J be the j-th distance cell of the distance image, where J is the total number of distance cells.

[0113] In some implementations, the structure type estimation model is determined as follows:

[0114] Construct a neural network model; wherein the neural network includes a first layer, a second layer, a third layer and a fourth layer connected in sequence. The first and second layers have one channel each, the third layer has the same number of channels as the number of structure types, and the fourth layer is a Softmax layer.

[0115] Optimize the neural network model;

[0116] The optimized neural network model is trained using the one-dimensional range profile of known target radar echo data and the type probability of each range cell in the one-dimensional range profile of known target, to obtain the structure type estimation model.

[0117] In some implementations, the neural network model is optimized, including:

[0118] The cross-entropy loss function is used as the cost function, and the stochastic gradient descent method is used to optimize the neural network model.

[0119] It is understood that the structures illustrated in the embodiments of the present invention do not constitute a specific limitation on a device for determining a target scattering center structure type. In other embodiments of the present invention, a device for determining a target scattering center structure type may include more or fewer components than illustrated, or combine some components, or split some components, or arrange different components. The illustrated components may be implemented in hardware, software, or a combination of software and hardware.

[0120] The information interaction and execution process between the modules in the above-mentioned device are based on the same concept as the method embodiment of the present invention, and the specific details can be found in the description of the method embodiment of the present invention, and will not be repeated here.

[0121] This invention also provides an electronic device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, it implements a method for determining the structure type of a target scattering center according to any embodiment of this invention.

[0122] This invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform a method for determining the structure type of a target scattering center according to any embodiment of this invention.

[0123] Specifically, a system or apparatus equipped with a storage medium may be provided, on which software program code implementing the functions of any of the embodiments described above is stored, and the computer (or CPU or MPU) of the system or apparatus may read and execute the program code stored in the storage medium.

[0124] In this case, the program code read from the storage medium can itself implement the function of any of the above embodiments, and therefore the program code and the storage medium storing the program code constitute part of the present invention.

[0125] Examples of storage media used to provide program code include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD+RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, program code can be downloaded from a server computer via a communication network.

[0126] Furthermore, it should be clear that not only can the program code read by the computer be executed, but also the operating system or other components operating on the computer can be instructed based on the program code to perform some or all of the actual operations, thereby realizing the function of any of the embodiments described above.

[0127] Furthermore, it is understood that the program code read from the storage medium is written to the memory set in the expansion board inserted into the computer or to the memory set in the expansion module connected to the computer. Then, based on the instructions of the program code, the CPU or other components installed on the expansion board or expansion module execute some and all of the actual operations, thereby realizing the function of any of the above embodiments.

[0128] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus.

[0129] Those skilled in the art will understand that all or part of the steps of the above method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps of the above method embodiments. The aforementioned storage medium includes various media that can store program code, such as ROM, RAM, magnetic disk, or optical disk.

[0130] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for determining the structural type of a target scattering center, characterized in that, include: The radar echo data of the target under test is divided into subbands to obtain multiple subband data. Perform Fourier transform on each sub-band data to obtain a one-dimensional sub-range image of each sub-band data; stitch together each one-dimensional sub-range image to obtain a one-dimensional range image of the radar echo data of the target under test; The one-dimensional range profile of the radar echo data of the target under test is input into a pre-trained structure type estimation model to obtain the type probability of each range cell in the one-dimensional range profile. The structure type estimation model is obtained by training a pre-constructed neural network model with the one-dimensional range profile of the known target radar echo data as input and the type probability of each range cell in the known target one-dimensional range profile as output. Based on the type probability, the structure type of each scattering center in the target under test is determined.

2. The method according to claim 1, characterized in that, The structural types include sharp-corner structures, curved-edge structures, straight-edge structures, single-curved surface structures, and planar structures.

3. The method according to claim 2, characterized in that, The one-dimensional range image of the known target radar echo data is obtained as follows: based on the known target's structural type, broadband radar echo data of the known target is generated using a GTD model; the generated broadband radar echo data is divided into sub-bands to obtain multiple sub-band data; Fourier transform is performed on each sub-band data to obtain a one-dimensional sub-range image of each sub-band data; and each one-dimensional sub-range image is stitched together to obtain the one-dimensional range image of the known target radar echo data.

4. The method according to claim 3, characterized in that, The calculation formula for the GTD model is as follows: In the formula, For radar echo data, For wave number, The central wave number, ， ， For radar observation frequency, For the center frequency, Where is the speed of light, and M is the number of scattering centers. Let m be the scattering amplitude coefficient of the m-th scattering center. Let m be the radial position of the m-th scattering center. For type parameters, To measure noise, This represents the total number of frequency points.

5. The method according to claim 4, characterized in that, The Fourier transform of each sub-band data obtained is performed using the following formula: In the formula, For the one-dimensional distance image of the i-th sub-band data, For the i-th sub-band data, Let J be the j-th distance cell of the distance image, where J is the total number of distance cells.

6. The method according to claim 1, characterized in that, The structure type estimation model is determined as follows: a neural network model is constructed; wherein the neural network includes a first layer, a second layer, a third layer, and a fourth layer connected in sequence, the first layer and the second layer each have one channel, the third layer has the same number of channels as the number of structure types, and the fourth layer is a Softmax layer; the neural network model is optimized; the optimized neural network model is trained using a one-dimensional range profile of known target radar echo data and the type probability of each range unit in the one-dimensional range profile of known target to obtain the structure type estimation model.

7. The method according to claim 6, characterized in that, The optimization of the neural network model includes: using the cross-entropy loss function as the cost function and using the stochastic gradient descent method to optimize the neural network model.

8. A device for determining the structure type of a target scattering center, characterized in that, include: The system comprises the following modules: a segmentation module for segmenting the radar echo data of the target under test into sub-bands, resulting in multiple sub-band data; a transformation module for performing Fourier transforms on each sub-band data to obtain a one-dimensional sub-range image; a stitching module for stitching together the obtained one-dimensional sub-range images to obtain a one-dimensional range image of the radar echo data of the target under test; an input module for inputting the one-dimensional range image of the radar echo data of the target under test into a pre-trained structure type estimation model to obtain the type probability of each range unit in the one-dimensional range image; wherein the structure type estimation model is obtained by training a pre-constructed neural network model using the one-dimensional range image of the known target radar echo data as input and the type probability of each range unit in the known target one-dimensional range image as output; and a determination module for determining the structure type of each scattering center in the target under test based on the type probabilities.

9. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method as described in any one of claims 1-7.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed in the computer, it causes the computer to perform the method of any one of claims 1-7.

Images