Sparse HRRP identification method based on compressed sensing imaging

Abstract

The invention proposes a sparse HRRP recognition method based on compressed sensing imaging, and relates to the technical field of high-resolution range profile target recognition, and the method comprises the steps: carrying out the reconstruction of a radar echo signal at a low sampling rate through a compressed sensing algorithm, and reconstructing the sparse HRRP data with improved resolution; performing data preprocessing on the acquired sparse HRRP data, wherein the data preprocessing comprises amplitude normalization and gravity center alignment processing; extracting sequence features and network features of the preprocessed sparse HRRP data, fusing the sequence features and the network features and performing normalization processing to form feature vectors of the HRRP of each target; and the feature vectors are sent to a random forest classifier for training and testing, and an identification result is obtained. Based on the method provided by the invention, the resolution, anti-noise performance and subsequent recognition performance of HRRP imaging are improved, and the storage space of an HRRP database is saved.

Description

Technical Field

[0001] The present application relates to the technical field of high-resolution range image target recognition, and in particular to a sparse HRRP recognition method based on compressed sensing imaging. Background Art

[0002] Target recognition is of great significance in the military field and can provide support for the reconnaissance and surveillance of sea and air targets. High-resolution range profile (HRRP) is the distribution information of target scattering points along the radar radial obtained from broadband radar echoes. Compared with the information obtained from narrowband radar echoes, it can be used for more accurate target identification; compared with the two-dimensional images obtained from broadband radars, it is easier to obtain and process.

[0003] In the traditional imaging system, obtaining higher imaging resolution requires higher transmission signal bandwidth and higher receiver sampling rate, which puts pressure on the data transmission and processing of radar systems. Compressed sensing is a new signal sampling theory, which points out that under the premise that the signal has a priori sparsity, the signal can be accurately reconstructed with fewer samples specified by the Nyquist sampling theorem. The scattering response of radar targets in the high-frequency region has a sparse characteristic and can be represented by the vector sum of the scattering responses of a small number of scattering centers. Therefore, the high-resolution imaging process of radar can be transformed into a sparse reconstruction problem.

[0004] In recent years, many scholars have proposed HRRP target recognition methods, which mainly involve first extracting features and then inputting the features into the model for classifier training and testing. The extracted features mainly include equivalent length, number of strong scattering points, etc. However, due to the limitations of the Nyquist sampling theorem, under certain sampling conditions, the imaging resolution of the existing HRRP imaging method is insufficient, which has a certain impact on the subsequent recognition performance. At the same time, the high-dimensional HRRP database will also take up a lot of storage space. Summary of the invention

[0005] The present application aims to solve one of the technical problems in the related art at least to some extent.

[0006] To this end, the first objective of the present application is to propose a sparse HRRP identification method based on compressed sensing imaging.

[0007] The second objective of the present application is to propose a sparse HRRP identification device based on compressed sensing imaging.

[0008] The third objective of the present application is to provide an electronic device.

[0009] A fourth objective of the present application is to provide a computer-readable storage medium.

[0010] A fifth object of the present application is to provide a computer program product.

[0011] To achieve the above-mentioned purpose, the first embodiment of the present application proposes a sparse HRRP identification method based on compressed sensing imaging, including:

[0012] The radar echo signal at low sampling rate is reconstructed using compressed sensing algorithm to reconstruct sparse high-resolution range image HRRP data with improved resolution;

[0013] Performing data preprocessing on the reconstructed sparse HRRP data, wherein the data preprocessing includes amplitude normalization and center of gravity alignment processing;

[0014] Extracting sequence features and network features of the preprocessed sparse HRRP data, fusing the sequence features with the network features and performing normalization processing to form a feature vector of the HRRP of each target;

[0015] The feature vector is sent to a random forest classifier for training and testing to obtain a recognition result.

[0016] Optionally, the reconstructing radar echo signals at a low sampling rate using a compressed sensing algorithm to reconstruct sparse HRRP data with improved resolution includes:

[0017] According to the input downsampling rate α, a Gaussian random matrix of size M×N is constructed as the measurement matrix Ψ, where And construct an IDFT matrix Φ of size N×N = {φ m,n =ω -mn ,n=0,...,N-1,m=0,...N-1} as the sparse basis matrix Φ;

[0018] Projecting the input radar target echo signal s with a size of N×1 onto the measurement matrix Ψ to obtain a measurement value y=Ψs of the radar echo signal at a low sampling rate with a size of M×1;

[0019] The sensing matrix A=ΨΦ is constructed by using the measurement matrix Ψ and the sparse basis matrix Φ, whose size is M×N, and Θ is set to represent the HRRP of the target, and the process of HRRP imaging is represented as a sparse reconstruction problem y=AΘ;

[0020] The SAMP algorithm is used to solve the sparse reconstruction problem and obtain a sparse approximate solution of HRRP.

[0021] Optionally, solving the sparse reconstruction problem to obtain a sparse approximate solution of HRRP includes:

[0022] Initialize the residual to r0 = y and initialize the support set to Initialize the atomic matrix as The number of iterations is t = 0;

[0023] Perform the following iterations in sequence: Find the index λ of the column most correlated with the residual t =argmax| <r t-1 ,a j >|, and update the support set Λ t =Λ t-1 ∪λ t and the corresponding column matrix Solve the sparse representation coefficient estimation under the support set Update residual Update the number of iterations t = t + 1;

[0024] The iteration is stopped until the residual decreases, and the sparse approximate solution of HRRP is obtained. Its size is N×1.

[0025] Optionally, the performing data preprocessing on the reconstructed sparse HRRP data includes:

[0026] For the sparse approximate solution Amplitude normalization is performed, where the amplitude normalization part uses l2 norm normalization, and the calculation method of the normalized range image is:

[0027] Calculate the amplitude-normalized Θ norm The center of gravity g is expressed as:

[0028]

[0029] Calculate the start and end positions of the translation based on the center of gravity, translate the center of gravity to the center of the pane, and norm After cropping and splicing, the sparse HRRP data X after centroid alignment is obtained.

[0030] Optionally, the process of extracting sequence features of the preprocessed sparse HRRP data includes:

[0031] Extract sparse HRRP data X = [x1, x2, ..., x N ], the sequence features include the number of strong scattering points, equivalent target length, 2-4 order center distance, entropy, wherein the number of strong scattering points is sparse HRRP data X = [x1, x2, ..., x N ] is greater than the mean; the equivalent target length is the sparse HRRP data X = [x1, x2, ..., x N ] is the length from the first non-zero data to the last non-zero data; the calculation formula for the l-order center distance is: Wherein, m1 is the first-order center distance; the entropy calculation formula is:

[0032] Optionally, the process of extracting network features from the preprocessed sparse HRRP data includes:

[0033] For sparse HRRP data X = [x1, x2, ..., x N ] to construct the graph and transform the sparse HRRP data X = [x1, x2, ..., x N ] are considered as nodes of the graph. When two nodes x i and x j The node x between it t Satisfy(x i -x t ) / (ti)>(x i -x j ) / (ji), an edge is established between the two nodes, and its edge weight is the tangent of the two nodes, until all nodes that meet the above conditions are connected, and a complex graph network is obtained;

[0034] The graph density, global efficiency, independence number, graph index complexity and TSP problem cost of the complex graph network are extracted, wherein the graph density is calculated by dividing the actual number of edges in the complex graph network by the maximum possible number of edges; the global efficiency is calculated by the average value of the reverse shortest path lengths between all nodes; the independence number refers to the size of the largest independent set in the complex graph network; the TSP problem cost refers to the length of the shortest Hamiltonian circuit of the complex graph network; the graph index complexity GIC is calculated as follows:

[0035]

[0036] Among them, λ max represents the maximum eigenvalue of the adjacency matrix of the complex graph network, and V represents the number of nodes of the complex graph network.

[0037] To achieve the above-mentioned purpose, the second embodiment of the present application proposes a sparse HRRP identification device based on compressed sensing imaging, comprising:

[0038] The compressed sensing module is used to reconstruct radar echo signals at low sampling rates using a compressed sensing algorithm to reconstruct sparse HRRP data with improved resolution;

[0039] A data preprocessing module, used for performing data preprocessing on the acquired sparse high-resolution range image HRRP data, wherein the data preprocessing includes amplitude normalization and center of gravity alignment processing;

[0040] A feature extraction module is used to extract the sequence features and network features of the preprocessed sparse HRRP data, fuse the sequence features with the network features and perform normalization processing to form a feature vector of the HRRP of each target;

[0041] The classifier module is used to send the feature vector to the random forest classifier for training and testing to obtain the recognition result.

[0042] To achieve the above-mentioned purpose, the third aspect of the present application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0043] The memory stores computer-executable instructions;

[0044] The processor executes the computer-executable instructions stored in the memory to implement the method as described in any one of the first aspects.

[0045] To achieve the above-mentioned purpose, the fourth aspect embodiment of the present application proposes a computer-readable storage medium, in which computer-readable storage medium is stored computer execution instructions, and when the computer execution instructions are executed by a processor, they are used to implement the method as described in any one of the first aspects.

[0046] To achieve the above-mentioned purpose, the fifth aspect of the present application proposes a computer program product, which implements any method in the first aspect when executed by a processor.

[0047] The present application provides a sparse HRRP recognition method, device, electronic device and storage medium based on compressed sensing imaging. On the one hand, the HRRP obtained using the compressed sensing algorithm can improve the imaging resolution under the condition of the same sampling rate. On the other hand, the HRRP obtained by sparse reconstruction removes a large amount of noise in non-target areas, that is, while obtaining a certain noise reduction performance, the HRRP obtained is sparse and more conducive to storage. In general, the solution of the present application improves the resolution and noise resistance of HRRP imaging and the subsequent recognition performance, and saves the storage space of HRRP.

[0048] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0050] Figure 1 A schematic diagram of a flow chart of a sparse HRRP identification method based on compressed sensing imaging provided in an embodiment of the present application;

[0051] Figure 2 The effect diagram of the imaging result of the method of the present application provided in the embodiment of the present application;

[0052] Figure 3 The effect diagram of the imaging result of the traditional method provided in the embodiment of the present application

[0053] Figure 4 A schematic diagram of the structure of a sparse HRRP recognition device based on compressed sensing imaging provided in an embodiment of the present application;

[0054] Figure 5 A schematic diagram of the processing flow of a sparse HRRP identification device based on compressed sensing imaging provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0056] Due to the limitations of the Nyquist sampling theorem, under certain sampling conditions, the imaging resolution of the existing HRRP imaging method is insufficient, which has a certain impact on the subsequent recognition performance. At the same time, the high-dimensional HRRP database will also take up a lot of storage space.

[0057] To address this problem, an embodiment of the present application provides a sparse HRRP identification method based on compressed sensing imaging. A compressed sensing algorithm is used to reconstruct radar echo signals at a low sampling rate to obtain sparse HRRP data with improved resolution. At the same time, a graph structure that is more suitable for sparse data is introduced for subsequent feature extraction and recognition. The network features can better reflect the structural characteristics of the strong scattering points of HRRP, and have a higher recognition rate than traditional feature extraction methods.

[0058] Figure 1 The flowchart of a sparse HRRP identification method based on compressed sensing imaging provided by an embodiment of the present application is as follows. Figure 1 As shown, the method comprises the following steps:

[0059] Step 101: reconstruct radar echo signals at a low sampling rate using a compressed sensing algorithm to reconstruct sparse high-resolution range profile (HRRP) data with improved resolution.

[0060] In the embodiment of the present application, this step mainly includes a compressed sensing sampling process and a sparse reconstruction process. The compressed sensing sampling process is used to obtain radar echo data at a low sampling rate, and the sparse reconstruction process mainly uses a sparsity adaptive greedy pursuit algorithm (SAMP) algorithm.

[0061] Specifically, step 101 also includes:

[0062] Step 201, according to the input downsampling rate α, construct a Gaussian random matrix of size M×N as the measurement matrix Ψ, where And construct an IDFT (inverse discrete Fourier transform) matrix Φ of size N×N = {φ m,n =ω -mn ,n=0,...,N-1,m=0,...N-1} as the sparse basis matrix Φ.

[0063] This step is used to construct the measurement matrix Ψ and the sparse basis matrix Φ, and, in order to satisfy the constrained isometry criterion, the measurement matrix adopts a Gaussian random matrix.

[0064] For the measurement matrix Ψ, M is calculated based on the downsampling rate α and the total number of samples N, that is, M is the result of the downsampling rate α multiplied by the total number of samples N, rounded down. This is done to meet the requirements for the number of samples in the compressed sensing theory, that is, the number of samples M should be less than the total number of samples N of the signal but still ensure that the signal can be accurately reconstructed.

[0065] For the sparse basis matrix Φ, ω is the Nth root of the complex unit, that is, ω = e -j2π / N , which is used to transform the signal from the time domain to the frequency domain so that it can be sparsely represented in the frequency domain.

[0066] The construction of these two matrices is to reconstruct the original signal with a small number of measurements under the compressed sensing framework. The measurement matrix Ψ is used to obtain low sampling rate measurements of the signal, while the sparse basis matrix Φ is used to represent the sparse characteristics of the signal in the frequency domain. Through these two matrices, the reconstruction problem of the original signal can be transformed into an optimization problem, which can then be solved using the compressed sensing algorithm.

[0067] Step 202 , projecting the input radar target echo signal s of size N×1 onto the measurement matrix Ψ, obtaining the measurement value y=Ψs of the radar echo signal at a low sampling rate, of size M×1.

[0068] This step is used to compress the sampled measurement values ​​y.

[0069] Step 203, using the measurement matrix Ψ and the sparse basis matrix Φ to construct a sensing matrix A = ΨΦ, whose size is M×N, let Θ represent the HRRP of the target, then the HRRP imaging process can be expressed as a sparse reconstruction problem y = AΘ.

[0070] This step is used to construct a sparse reconstruction problem according to the HRRP imaging process.

[0071] Step 204: Use the SAMP algorithm to solve the sparse reconstruction problem and obtain a sparse approximate solution of the HRRP.

[0072] In the embodiment of the present application, the SAMP algorithm is used to calculate the sparse reconstruction problem, specifically:

[0073] First, initialize the residual to r0=y, and initialize the support set to Initialize the atomic matrix as The number of iterations is t=0.

[0074] Then, the following iterations are performed in sequence: Find the index λ of the column most correlated with the residual t =argmax| <r t-1 ,a j >|, and update the support set Λ t =Λ t-1 ∪λ t and the corresponding column matrix Solve the sparse representation coefficient estimation under the support set Update residual Update the number of iterations t=t+1.

[0075] The iteration is stopped until the residual decreases, and finally the sparse approximate solution of HRRP is obtained. Its size is N×1.

[0076] Step 102: preprocess the reconstructed sparse HRRP data. The data preprocessing includes amplitude normalization and center of gravity alignment.

[0077] In the embodiment of the present application, for the sparse approximate solution obtained in step 101, Data preprocessing is performed to eliminate the amplitude sensitivity and translation sensitivity of HRRP data. This step includes amplitude normalization and center of gravity alignment.

[0078] As a possible implementation method, we first solve the sparse approximate solution Perform amplitude normalization.

[0079] Among them, amplitude normalization is to adjust the amplitude of the data to a standard range. In the embodiment of the present application, the amplitude normalization processing part uses l2 norm normalization, and the calculation method of the normalized range image is: The purpose of this formula is to Each element of is divided by the square root of the sum of the squares of all elements, so that the l2 norm of the normalized data (that is, the square root of the sum of the squares of all elements) is 1.

[0080] Next, calculate the amplitude-normalized Θ norm The centroid g is the sum of the products of all elements in the data and their indices divided by the sum of all elements. The calculation formula is:

[0081]

[0082] Next, we find the starting and ending positions after translation based on the center of gravity g, and calculate θ norm The sparse HRRP data X after centroid alignment is obtained by cropping.

[0083] The purpose of this process is to ensure that the amplitude and position sensitivity of the HRRP data is eliminated before feature extraction and classification, thereby improving the accuracy and robustness of recognition.

[0084] Step 103 , extracting the sequence features and network features of the preprocessed sparse HRRP data, fusing the sequence features and the network features, and performing normalization processing to form a feature vector of the HRRP of each target.

[0085] In the embodiment of the present application, the extracted sparse HRRP data X=[x1, x2, ..., x N ]’s sequence features include the number of strong scattering points, equivalent target length, 2nd-4th order center distance, and entropy.

[0086] The number of strong scattering points is the sparse HRRP data X = [x1, x2, ..., x N ] is greater than the mean; the equivalent target length is the sparse HRRP data X = [x1, x2, ..., x N ] is the length from the first non-zero data to the last non-zero data; the calculation formula for the l-order center distance is: Where m1 is the first-order center distance; the entropy calculation formula is: Used to measure the uncertainty of data.

[0087] Secondly, for extracting sparse HRRP data X = [x1, x2, ..., x N ] network features, we first need to analyze the sparse HRRP data X = [x1, x2, ..., x N ] to construct the graph.

[0088] Specifically, the sparse HRRP data X = [x1, x2, ..., x N ] are considered as nodes of the graph. When two nodes x i and x j The node x between it t Satisfy(x i -x t) / (ti)>(x i -x j ) / (ji), an edge is established between the two nodes, and its edge weight is the tangent of the two nodes, until all the nodes that meet the above conditions are connected, and a complex graph network is obtained.

[0089] Then, the features of the complex graph network, including graph density, global efficiency, independence number, graph index complexity and TSP problem cost, are extracted.

[0090] Among them, the calculation method of graph density is the actual number of edges in the complex graph network divided by the maximum possible number of edges; the calculation method of global efficiency is the average value of the reverse shortest path length between all nodes; the independence number refers to the size of the largest independent set in the complex graph network; the TSP problem cost refers to the length of the shortest Hamiltonian circuit in the complex graph network; the calculation formula of graph index complexity GIC is:

[0091]

[0092] Among them, λ max represents the maximum eigenvalue of the adjacency matrix of the complex graph network, and V represents the number of nodes in the complex graph network.

[0093] Finally, the sequence features are fused with the network features and normalized, that is, the values ​​are limited to between 0 and 1 to form a feature vector.

[0094] It can be understood that the extraction of these feature vectors is to better understand and analyze HRRP data, thereby improving the accuracy of target recognition. By inputting these feature vectors into the classifier, different targets can be effectively classified.

[0095] Step 104, sending the feature vector to a random forest classifier for training and testing to obtain a recognition result.

[0096] In this embodiment, for the feature vectors of HRRP of each target obtained in step 103, the random forest classifier will use these feature vectors for training to learn how to distinguish different targets. After the training is completed, the random forest classifier will use test data to verify its recognition ability. Finally, the random forest classifier will output the recognition result, that is, the category or identity of each target.

[0097] It can be understood that random forest is an integrated learning method, which improves the accuracy and robustness of classification by constructing multiple decision trees and aggregating their prediction results. The method of training the classifier is well known to people in the field, and this application does not describe and explain it in detail.

[0098] In addition, the embodiment of the present application also provides a comparison result of the target recognition rate between a sparse HRRP recognition method based on compressed sensing imaging and a traditional HRRP imaging method. Through this comparison, the performance advantage of the method proposed in the present application can be demonstrated. The imaging results are as follows: Figure 2 and Figure 3 The comparison results of HRRP target recognition rate are shown in Table 1.

[0099] Table 1

[0100] method Recognition rate Traditional imaging + sequence features 75.10％ The method proposed in this application 84.63％

[0101] In comparison, the recognition rate of the traditional imaging method combined with sequence features is 75.10%, while the method proposed in this application achieves a recognition rate of 84.63% by using compressed sensing imaging technology and advanced feature extraction methods. This result shows that the method of this application is significantly better than the traditional method in terms of target recognition accuracy.

[0102] In order to implement the above embodiment, the present application also proposes a sparse HRRP identification device based on compressed sensing imaging. Figure 4 FIG. 1 is a schematic diagram of a sparse HRRP identification device 10 based on compressed sensing imaging provided in an embodiment of the present application. Figure 4 As shown, the device comprises:

[0103] The compressed sensing module 100 is used to reconstruct the radar echo signal at a low sampling rate using a compressed sensing algorithm to reconstruct sparse HRRP data with improved resolution;

[0104] A data preprocessing module 200 is used to perform data preprocessing on the acquired sparse high-resolution range image HRRP data, wherein the data preprocessing includes amplitude normalization and center of gravity alignment processing;

[0105] The feature extraction module 300 is used to extract the sequence features and network features of the pre-processed sparse HRRP data, fuse the sequence features with the network features and perform normalization processing to form the feature vector of the HRRP of each target;

[0106] The classifier module 400 is used to send the feature vector to the random forest classifier for training and testing to obtain the recognition result.

[0107] in, Figure 5 A schematic diagram of the processing flow of a sparse HRRP identification device based on compressed sensing imaging provided in an embodiment of the present application.

[0108] In order to implement the above embodiments, the present application also proposes an electronic device, comprising: a processor, and a memory communicatively connected to the processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory to implement the method provided by the above embodiments.

[0109] In order to implement the above embodiments, the present application also proposes a computer-readable storage medium, in which computer-executable instructions are stored. When the computer-executable instructions are executed by a processor, they are used to implement the methods provided by the above embodiments.

[0110] In order to implement the above embodiments, the present application also proposes a computer program product, including a computer program, which implements the methods provided by the above embodiments when executed by a processor.

[0111] The collection, storage, use, processing, transmission, provision and disclosure of user personal information involved in this application are in compliance with relevant laws and regulations and do not violate public order and good morals.

[0112] It should be noted that personal information from users should be collected for legitimate and reasonable purposes and should not be shared or sold outside of these legitimate uses. In addition, such collection / sharing should be carried out after receiving the user's informed consent, including but not limited to notifying the user to read the user agreement / user notice and sign the agreement / authorization including authorization of relevant user information before the user uses the function. In addition, any necessary steps should be taken to protect and safeguard access to such personal information data and ensure that others who have access to personal information data comply with its privacy policy and procedures.

[0113] The present application is expected to provide an implementation scheme for users to selectively block the use or access of personal information data. That is, the present disclosure is expected to provide hardware and / or software to prevent or block access to such personal information data. Once the personal information data is no longer needed, the risk can be minimized by limiting data collection and deleting the data. In addition, when applicable, such personal information is de-identified to protect the privacy of the user.

[0114] In the description of the aforementioned embodiments, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0115] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0116] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0117] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute the instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways if necessary, and then stored in a computer memory.

[0118] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0119] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0120] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0121] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A sparse HRRP identification method based on compressed sensing imaging, characterized in that: The following steps are involved: The radar echo signal at low sampling rate is reconstructed using compressed sensing algorithm to reconstruct sparse high-resolution range image HRRP data with improved resolution; Performing data preprocessing on the reconstructed sparse HRRP data, wherein the data preprocessing includes amplitude normalization and center of gravity alignment processing; Extracting sequence features and network features of the preprocessed sparse HRRP data, fusing the sequence features with the network features and performing normalization processing to form a feature vector of the HRRP of each target; The feature vector is sent to a random forest classifier for training and testing to obtain a recognition result.

2. The method according to claim 1, characterized in that The method of reconstructing radar echo signals at a low sampling rate using a compressed sensing algorithm to reconstruct sparse HRRP data with improved resolution includes: According to the input downsampling rate α, a Gaussian random matrix of size M×N is constructed as the measurement matrix Ψ, where And construct an IDFT matrix Φ of size N×N = {φ m,n =ω -mn ,n=0,...,N-1,m=0,...N-1} as the sparse basis matrix Φ; Projecting the input radar target echo signal s with a size of N×1 onto the measurement matrix Ψ to obtain a measurement value y=Ψs of the radar echo signal at a low sampling rate with a size of M×1; The sensing matrix A=ΨΦ is constructed by using the measurement matrix Ψ and the sparse basis matrix Φ, whose size is M×N, and Θ is set to represent the HRRP of the target, and the process of HRRP imaging is represented as a sparse reconstruction problem y=AΘ; The SAMP algorithm is used to solve the sparse reconstruction problem and obtain a sparse approximate solution of HRRP.

3. The method according to claim 2, characterized in that The step of solving the sparse reconstruction problem to obtain a sparse approximate solution of the HRRP includes: Initialize the residual to r0 = y and initialize the support set to Initialize the atomic matrix as The number of iterations is t = 0; Perform the following iterations in sequence: Find the index λ of the column most correlated with the residual t =argmax| <r t-1 ,a j >|, and update the support set Λ t =Λ t-1 ∪λ t and the corresponding column matrix Solve the sparse representation coefficient estimation under the support set Update residual Update the number of iterations t = t + 1; The iteration is stopped until the residual decreases, and the sparse approximate solution of HRRP is obtained. Its size is N×1.

4. The method according to claim 3, characterized in that The data preprocessing of the reconstructed sparse HRRP data includes: For the sparse approximate solution Amplitude normalization is performed, where the amplitude normalization part uses l2 norm normalization, and the calculation method of the normalized range image is: Calculate the amplitude-normalized Θ norm The center of gravity g is expressed as: Calculate the start and end positions of the translation based on the center of gravity, translate the center of gravity to the center of the pane, and norm After cropping and splicing, the sparse HRRP data X after centroid alignment is obtained.

5. The method according to claim 4, characterized in that The process of extracting sequence features of preprocessed sparse HRRP data includes: Extract sparse HRRP data X = [x1, x2, ..., x N ], the sequence features include the number of strong scattering points, equivalent target length, 2-4 order center distance, entropy, wherein the number of strong scattering points is sparse HRRP data X = [x1, x2, ..., x N ] is greater than the mean; the equivalent target length is the sparse HRRP data X = [x1, x2, ..., x N ] is the length from the first non-zero data to the last non-zero data; the calculation formula for the l-order center distance is: Wherein, m1 is the first-order center distance; the entropy calculation formula is:

6. The method according to claim 5, characterized in that The process of extracting network features from preprocessed sparse HRRP data includes: For sparse HRRP data X = [x1, x2, ..., x N ] to construct the graph and transform the sparse HRRP data X = [x1, x2, ..., x N ] are considered as nodes of the graph. When two nodes x i and x j The node x between it t Satisfy(x i -x t ) / (ti)>(x i -x j ) / (ji), an edge is established between the two nodes, and its edge weight is the tangent of the two nodes, until all nodes that meet the above conditions are connected, and a complex graph network is obtained; The graph density, global efficiency, independence number, graph index complexity and TSP problem cost of the complex graph network are extracted, wherein the graph density is calculated by dividing the actual number of edges in the complex graph network by the maximum possible number of edges; the global efficiency is calculated by the average value of the reverse shortest path lengths between all nodes; the independence number refers to the size of the largest independent set in the complex graph network; the TSP problem cost refers to the length of the shortest Hamiltonian circuit of the complex graph network; the graph index complexity GIC is calculated as follows: Among them, λ max represents the maximum eigenvalue of the adjacency matrix of the complex graph network, and V represents the number of nodes of the complex graph network.

7. A sparse HRRP identification device based on compressed sensing imaging, characterized in that: include: The compressed sensing module is used to reconstruct radar echo signals at low sampling rates using a compressed sensing algorithm to reconstruct sparse HRRP data with improved resolution; A data preprocessing module, used for performing data preprocessing on the acquired sparse high-resolution range image HRRP data, wherein the data preprocessing includes amplitude normalization and center of gravity alignment processing; A feature extraction module is used to extract the sequence features and network features of the preprocessed sparse HRRP data, fuse the sequence features with the network features and perform normalization processing to form a feature vector of the HRRP of each target; The classifier module is used to send the feature vector to the random forest classifier for training and testing to obtain the recognition result.

8. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the method according to any one of claims 1 to 6 when executed by a processor.

10. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 6 when being executed by a processor.

Images