A method for evaluating quality of radar high resolution range profile data

Abstract

The application discloses a radar high-resolution range profile data quality evaluation method, and relates to the technical field of radar, comprising the following steps: obtaining a group of continuous N times of high-resolution range profile (HRRP) data; selecting strong scattering points in each HRRP data and generating a first set H, the first set H comprising distance unit positions of each strong scattering point and normalized amplitudes; calculating the similarity of all adjacent two times of HRRP data in the N times of HRRP data according to the first set H corresponding to each HRRP data, obtaining a similarity set; and calculating the mean value and variance of the similarity of all adjacent two times of HRRP data according to the similarity set, and evaluating the robustness of the HRRP data. The evaluation result is more objective, and the HRRP image does not need to be observed by using a display device, so that the cost is reduced.

Description

Technical Field

[0001] This invention belongs to the field of radar technology, specifically relating to a method for evaluating the quality of high-resolution radar range image data. Background Technology

[0002] In the field of radar target recognition, recognition performance is mainly affected by two factors: data quality and algorithm model. Currently, most research focuses on improving the algorithm model. However, in some cases, poor recognition performance is not caused by the algorithm model, but rather by poor data quality. Data quality, as one of the two major factors affecting recognition performance and also the front end of the algorithm model, is extremely important. Data robustness is a crucial aspect reflecting data quality, therefore, analyzing data robustness is of great significance.

[0003] Existing technologies lack effective methods for evaluating the quality of HRRP (High Resolution Range Profile) data, primarily relying on HRRP signal-to-noise ratio (SNR) or observation of HRRP images via display devices. However, since HRRP SNR is mainly influenced by both range and the target's radar cross-section, it cannot reflect the differences in HRRP data quality under different target attitudes. Furthermore, HRRP SNR cannot describe data robustness and is only suitable for simple radar target identification tasks. Observing HRRP images via display devices is costly, labor-intensive, and its evaluation metrics are highly subjective. Summary of the Invention

[0004] To address the aforementioned problems in the existing technology, this invention provides a method for evaluating the quality of high-resolution radar range image data. The technical problem to be solved by this invention is achieved through the following technical solution:

[0005] This invention provides a method for evaluating the quality of high-resolution radar range image data, comprising:

[0006] Acquire a set of N consecutive high-resolution range profiles (HRRP) data;

[0007] Strong scattering points are selected from each HRRP data set and a first set H is generated. The first set H includes the range cell position and normalized amplitude of each strong scattering point.

[0008] Based on the first set H corresponding to each HRRP data, calculate the similarity of all adjacent HRRP data in N HRRP data, and obtain the similarity set;

[0009] Based on the aforementioned similarity set, the mean and variance of the similarity between all two adjacent HRRP data sets are calculated, and the robustness of the HRRP data is evaluated.

[0010] In one embodiment of the present invention, before the step of selecting strong scattering points in each HRRP data set and generating a first set H, the method further includes: preprocessing each HRRP data set; wherein...

[0011] The steps for preprocessing HRRP data for each iteration include:

[0012] Obtain the sampling frequency and radar bandwidth of each HRRP data, and downsample the HRRP data when the sampling frequency is greater than the radar bandwidth;

[0013] Obtain the distance cell location corresponding to the maximum amplitude of the N / 2th HRRP data;

[0014] The HRRP data for each iteration is windowed and cropped based on the distance cell position corresponding to the maximum amplitude.

[0015] In one embodiment of the present invention, the step of selecting strong scattering points in each HRRP data set and generating a first set H includes:

[0016] Calculate the maximum amplitude of each HRRP data point and multiply it by a preset coefficient to obtain the amplitude threshold value used to select strong scattering points;

[0017] Determine the distance cell in each HRRP data whose amplitude is greater than the amplitude threshold, and select the scattering point in the distance cell as the strong scattering point;

[0018] Record the range cell position and amplitude of each strong scattering point in each HRRP data set, and normalize the amplitude of the strong scattering points in each HRRP data set to generate the first set H:

[0019]

[0020] In the formula, h1, h2, ..., h r These represent the range cell positions of the 1st, 2nd, ..., rth strong scattering points in a single HRRP dataset. These represent the normalized amplitudes of the 1st, 2nd, ..., rth strong scattering points, respectively.

[0021] In one embodiment of the present invention, the step of calculating the similarity of all adjacent HRRP data in N HRRP data according to the first set H corresponding to each HRRP data, and obtaining the similarity set, includes:

[0022] Obtain the first set H corresponding to the k-th HRRP data. kThe first set H corresponding to the (k+1)th HRRP data k+1 Based on the distance cell positions of each strong scattering point in the k-th HRRP data and the (k+1)-th HRRP data, the transfer distance between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data is calculated, k = 1, 2, ..., N-1;

[0023] Based on the normalized amplitudes of each strong scattering point in the k-th HRRP data and the (k+1)-th HRRP data, the transfer amplitude between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data is solved by linear programming.

[0024] Based on the transfer distance and the transfer amplitude, calculate the transfer cost between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data;

[0025] Calculate the bulldozer distance EMD between the kth HRRP data and the (k+1)th HRRP data based on the transfer cost;

[0026] Based on the bulldozer distance EMD, calculate the similarity S between the k-th HRRP data and the (k+1)-th HRRP data. k ;

[0027] After traversing the N HRRP data, we obtain N-1 similarity sets S = [s1, s2, ..., snp.] for adjacent HRRP data. N-1 ], where s1, s2……s N-1 These represent the similarity between the 1st, 2nd, ..., N-1th HRRP data and the next HRRP data, respectively.

[0028] In one embodiment of the present invention, the first set H corresponding to the k-th HRRP data is obtained. k The first set H corresponding to the (k+1)th HRRP data k+1 The steps for calculating the transfer distance between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data, based on the distance cell positions of each strong scattering point in the k-th and (k+1)-th HRRP data, include:

[0029] Obtain the first set H corresponding to the k-th HRRP data. k The first set H corresponding to the (k+1)th HRRP data k+1 ;

[0030] Based on the distance cell positions of each strong scattering point in the k-th HRRP data and the (k+1)-th HRRP data, calculate the Euclidean distance between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data, and use this distance as the transfer distance between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data.

[0031] In one embodiment of the present invention, the step of solving the transfer amplitude between each strong scattering point in the k-th HRRP data and each strong scattering point in the k+1-th HRRP data through linear programming based on the normalized amplitude of each strong scattering point in the k-th HRRP data and the k+1-th HRRP data includes:

[0032] Construct the following linear programming problem:

[0033]

[0034] Constraints:

[0035] f ij ≥0 1≤i≤m,1≤j≤n;

[0036]

[0037]

[0038] In the formula, m and n represent the number of strong scattering points in the k-th HRRP data and the (k+1)-th HRRP data, respectively, and d ij f represents the transfer distance between the i-th strong scattering point in the k-th HRRP data and the j-th strong scattering point in the (k+1)-th HRRP data. ij This represents the transfer magnitude between the i-th strong scattering point in the k-th HRRP data and the j-th strong scattering point in the (k+1)-th HRRP data. This represents the normalized amplitude of the i-th strong scattering point in the k-th HRRP data. This represents the normalized amplitude of the j-th strong scattering point in the (k+1)-th HRRP data;

[0039] By solving the linear programming problem, the transfer amplitude between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data is obtained.

[0040] In one embodiment of the present invention, the step of calculating the transfer cost between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data based on the transfer distance and the transfer amplitude includes:

[0041] For the transfer amplitude f ijWith transfer distance d ij Perform a dot product operation to obtain the transfer cost between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data.

[0042] In one embodiment of the present invention, the step of calculating the bulldozer distance EMD between the k-th HRRP data and the (k+1)-th HRRP data based on the transfer cost includes:

[0043] Calculate the sum of transfer costs between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data to obtain the bulldozer distance EMD between the k-th HRRP data and the (k+1)-th HRRP data.

[0044] In one embodiment of the present invention, the similarity S between the k-th HRRP data and the (k+1)-th HRRP data is calculated according to the following formula. k :

[0045]

[0046] In the formula, EMD(H) k H k+1 MaxDistance(H) represents the bulldozer distance between the k-th HRRP data and the (k+1)-th HRRP data. k H k+1 ) represents the farthest distance between distance cells in the k-th HRRP data and the (k+1)-th HRRP data, where MaxDistance(H k H k+1 The greater of the following values ​​is: the distance between the first distance cell in the k-th HRRP data and the last distance cell in the (k+1)-th HRRP data, or the distance between the last distance cell in the k-th HRRP data and the first distance cell in the (k+1)-th HRRP data.

[0047] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0048] This invention provides a method for evaluating the quality of high-resolution range image data of radar, which can be used to evaluate the robustness of HRRP data. Compared with the HRRP signal-to-noise ratio used in the prior art, the evaluation results of this invention are more objective, and there is no need to use a display device to observe the HRRP image, which helps to reduce costs.

[0049] Furthermore, calculating similarity by selecting strong scattering points from the HRRP data improves computational performance compared to directly calculating 256 points of HRRP data.

[0050] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0051] Figure 1 This is a flowchart of a method for evaluating the quality of high-resolution range image data of radar provided in an embodiment of the present invention;

[0052] Figure 2 This is a schematic diagram of a set of continuous HRRP data provided in an embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram of the preprocessed HRRP data provided in an embodiment of the present invention;

[0054] Figure 4a This is a radar one-dimensional range profile described using preprocessed HRRP data, as provided in an embodiment of the present invention.

[0055] Figure 4b This is a radar one-dimensional range profile described using selected strong scattering points, provided by an embodiment of the present invention;

[0056] Figure 5 This is a schematic diagram illustrating the calculation of similarity sets provided in an embodiment of the present invention;

[0057] Figure 6 This is a two-dimensional schematic diagram of the transfer distance between each strong scattering point in the first HRRP data and each strong scattering point in the second HRRP data provided in the embodiment of the present invention;

[0058] Figure 7 This is a two-dimensional schematic diagram of the transfer distance between each strong scattering point in the first HRRP data and each strong scattering point in the second HRRP data provided in the embodiment of the present invention;

[0059] Figure 8 This is a flowchart illustrating the robustness comparison between signal HRRP data and noise HRRP data provided in an embodiment of the present invention.

[0060] Figure 9 This is a schematic diagram illustrating the similarity between the comparison signal HRRP data and the noise HRRP data provided in an embodiment of the present invention. Detailed Implementation

[0061] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0062] Figure 1 This is a flowchart of a method for evaluating the quality of high-resolution radar range image data provided in an embodiment of the present invention. For example... Figure 1 As shown, this embodiment of the invention provides a method for evaluating the quality of high-resolution radar range image data, including:

[0063] S1. Obtain a set of N consecutive high-resolution range profiles (HRRP) data;

[0064] S2. Select strong scattering points from each HRRP data and generate a first set H. The first set H includes the range cell position of each strong scattering point and the normalized amplitude.

[0065] S3. Based on the first set H corresponding to each HRRP data, calculate the similarity of all adjacent HRRP data in N HRRP data, and obtain the similarity set.

[0066] S4. Based on the similarity set, calculate the mean and variance of the similarity between all two adjacent HRRP data sets, and evaluate the robustness of the HRRP data.

[0067] Figure 2 This is a schematic diagram of a set of continuous HRRP data provided in an embodiment of the present invention. The sampling frequency is 1200MHz, the bandwidth is 500MHz, and the number of sampling points is 3000. It should be noted that "continuous" here means that the acquired N HRRP data points approximately describe the same state of the target. The value of N should be within a reasonable range, such as N=16. If the value of N is too small, the number of elements in the similarity set calculated in subsequent steps will be too small. This embodiment needs to evaluate the robustness of the HRRP data by calculating the mean and variance of the similarity set. When the number of elements in the similarity set is too small, the mean and variance are easily affected by individual elements, thus affecting the robustness evaluation result of the HRRP data. On the other hand, if N is too large, it indicates that the HRRP data duration is long. As time increases, the change in the target's motion state will also increase, which may cause the N HRRP data points to not describe the same state of the target.

[0068] Optionally, before the step of selecting strong scattering points from each HRRP data set and generating the first set H, the method further includes: preprocessing each HRRP data set; wherein,

[0069] The steps for preprocessing HRRP data for each iteration include:

[0070] Obtain the sampling frequency and radar bandwidth of each HRRP data, and downsample the HRRP data when the sampling frequency is greater than the radar bandwidth;

[0071] Obtain the distance cell location corresponding to the maximum amplitude of the N / 2th HRRP data;

[0072] Windowing is performed on each HRRP data point based on the distance cell position corresponding to the maximum amplitude.

[0073] Before selecting strong scattering points from each HRRP dataset, the HRRP data can be preprocessed, including downsampling and support region truncation. It should be understood that due to the sampling frequency, HRRP data often has multiple sampling points per range cell. To ensure that each range cell has only one sampling point, downsampling can be performed. Additionally, because of the large window size, HRRP data contains many pure noise sampling points; therefore, support region truncation is also necessary.

[0074] Specifically, firstly, the sampling frequency and radar bandwidth of the HRRP data are obtained and compared. When the sampling frequency is greater than the radar bandwidth, downsampling is required for each HRRP data; otherwise, downsampling is not required. Next, the range cell position corresponding to the largest amplitude value in the N / 2th HRRP data is obtained. Based on its position, windowing is performed on the N HRRP data respectively. The number of interception points can be set according to the specific physical size of the target, so that each intercepted HRRP data can completely cover the target range.

[0075] Figure 3 This is a schematic diagram of preprocessed HRRP data provided in an embodiment of the present invention. Figure 3 As shown, through the above preprocessing operations, the preprocessed HRRP data becomes 256 sampling points, and can simultaneously contain the support region of 16 HRRPs.

[0076] Optionally, step S2, which involves selecting strong scattering points from each HRRP dataset and generating the first set H, includes:

[0077] S201. Calculate the maximum amplitude of each HRRP data and multiply it by a preset coefficient to obtain the amplitude threshold value used to select strong scattering points;

[0078] S202. Determine the distance cell whose amplitude is greater than the amplitude threshold in each HRRP data, and select the scattering point in the distance cell as the strong scattering point;

[0079] S203. Record the distance cell position and amplitude of each strong scattering point in each HRRP data set, and normalize the amplitude of the strong scattering points in each HRRP data set to generate the first set H:

[0080]

[0081] In the formula, h1, h2, ..., h r These represent the range cell positions of the 1st, 2nd, ..., rth strong scattering points in a single HRRP dataset. These represent the normalized amplitudes of the 1st, 2nd, ..., rth strong scattering points, respectively.

[0082] For example, with a preset coefficient of 0.15, after normalizing the amplitude, the sum of the normalized amplitudes of strong scattering points in each HRRP data is 1.

[0083] Figure 4a This is a radar one-dimensional range profile described using preprocessed HRRP data, provided in an embodiment of the present invention. Figure 4b This is a radar one-dimensional range profile described using selected strong scattering points, provided in an embodiment of the present invention. Please refer to... Figures 4a-4b Taking the first HRRP data as an example, Figure 4a The radar one-dimensional range profile was described using HRRP data from 256 preprocessed sampling points. Figure 4b The radar one-dimensional range profile is described based on the 36 strong scattering points obtained. It can be seen that the two have a high degree of similarity, indicating that the scattering characteristics of the target can also be described by extracting strong scattering points.

[0084] Optionally, step S3, which involves calculating the similarity between all adjacent HRRP data points in N HRRP data points based on the first set H corresponding to each HRRP data point, to obtain the similarity set, includes:

[0085] S301. Obtain the first set H corresponding to the k-th HRRP data. k The first set H corresponding to the (k+1)th HRRP data k+1 Based on the distance cell positions of each strong scattering point in the k-th HRRP data and the (k+1)-th HRRP data, the transfer distance between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data is calculated, k = 1, 2, ..., N-1;

[0086] S302. Based on the normalized amplitudes of each strong scattering point in the k-th HRRP data and the (k+1)-th HRRP data, solve the transfer amplitude between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data using linear programming.

[0087] S303. Based on the transfer distance and transfer amplitude, calculate the transfer cost between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data.

[0088] S304. Calculate the bulldozer distance EMD between the kth HRRP data and the (k+1)th HRRP data based on the transfer cost.

[0089] S305. Based on the bulldozer distance EMD, calculate the similarity S between the k-th HRRP data and the (k+1)-th HRRP data. k ;

[0090] S306. After traversing the HRRP data N times, we obtain the similarity set S = [s1, s2, ..., s1] of N-1 consecutive HRRP data sets. N-1 ], where s1, s2……s M-1 These represent the similarity between the 1st, 2nd, ..., N-1th HRRP data and the next HRRP data, respectively.

[0091] Specifically, in step S301, the first set H corresponding to the k-th HRRP data is obtained. k The first set H corresponding to the (k+1)th HRRP data k+1 The steps for calculating the transfer distance between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data, based on the distance cell positions of each strong scattering point in the k-th and (k+1)-th HRRP data, include:

[0092] Obtain the first set H corresponding to the k-th HRRP data. k The first set H corresponding to the (k+1)th HRRP data k+1 ;

[0093] Based on the distance cell positions of each strong scattering point in the k-th HRRP data and the (k+1)-th HRRP data, calculate the Euclidean distance between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data, and use this distance as the transfer distance between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data.

[0094] Figure 5 This is a schematic diagram illustrating the calculation of similarity sets provided in an embodiment of the present invention. For example... Figure 5 As shown, taking 16 HRRP data sets as an example, the similarity set should include: the similarity between the 1st HRRP data and the 2nd HRRP data, the similarity between the 2nd HRRP data and the 3rd HRRP data, ..., the similarity between the 15th HRRP data and the 16th HRRP data.

[0095] In step S301, please continue to refer to... Figure 5 Taking m=36 strong scattering points in the first HRRP data and n=29 strong scattering points in the second HRRP data as examples, the Euclidean distance between the 36 strong scattering points in the first HRRP data and the 29 strong scattering points in the second HRRP data is calculated, and the transfer distance between each strong scattering point in the first HRRP data and each strong scattering point in the second HRRP data is obtained, as shown below:

[0096]

[0097] Figure 6 This is a two-dimensional schematic diagram showing the transfer distance between each strong scattering point in the first HRRP data and each strong scattering point in the second HRRP data provided in this embodiment of the invention. Figure 6 As shown, the marker (1, 1, 1) indicates that the transfer distance from the first strong scattering point in the first HRRP data to the first strong scattering point in the second HRRP data is 1. Similarly, the marker (1, 29, 134) indicates that the transfer distance from the first strong scattering point in the first HRRP data to the 29th strong scattering point in the second HRRP data is 134.

[0098] Step S302, which involves solving the transfer amplitude between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data using linear programming based on the normalized amplitudes of each strong scattering point in the k-th HRRP data and the (k+1)-th HRRP data, includes:

[0099] Construct the following linear programming problem:

[0100]

[0101] Constraints:

[0102] f ij ≥0 1≤i≤m,1≤j≤n;

[0103]

[0104]

[0105] In the formula, m and n represent the number of strong scattering points in the k-th HRRP data and the (k+1)-th HRRP data, respectively, and d ij f represents the transfer distance between the i-th strong scattering point in the k-th HRRP data and the j-th strong scattering point in the (k+1)-th HRRP data. ij This represents the transfer magnitude between the i-th strong scattering point in the k-th HRRP data and the j-th strong scattering point in the (k+1)-th HRRP data. This represents the normalized amplitude of the i-th strong scattering point in the k-th HRRP data. This represents the normalized amplitude of the j-th strong scattering point in the (k+1)-th HRRP data;

[0106] By solving a linear programming problem, the transfer amplitude between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data is obtained.

[0107] It should be understood that the constraint f ij≥0 indicates that the transfer amplitude between the i-th strong scattering point in the k-th HRRP data and the j-th strong scattering point in the (k+1)-th HRRP data must be non-negative, constraining the condition. This means that the amplitude transferred from the i-th strong scattering point in the k-th HRRP data to each strong scattering point in the (k+1)-th HRRP data must be less than or equal to its own normalized amplitude. Constraints This means that the amplitude received by the j-th strong scattering point in the (k+1)-th HRRP data must be less than its own normalized amplitude.

[0108] Figure 7 This is a two-dimensional schematic diagram showing the transfer distance between each strong scattering point in the first HRRP data and each strong scattering point in the second HRRP data provided in this embodiment of the invention. Figure 7 As shown, the marker (1, 1, 0.0181648) indicates that the transfer amplitude between the first strong scattering point in the first HRRP data and the first strong scattering point in the second HRRP data is 0.0181648, and the marker (1, 29, 0) indicates that the transfer amplitude between the first strong scattering point in the first HRRP data and the first strong scattering point in the second HRRP data is 0.

[0109] Step S303, which calculates the transfer cost between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data based on the transfer distance and transfer amplitude, includes:

[0110] For the transfer amplitude f ij With transfer distance d ij Perform a dot product operation to obtain the transfer cost between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data.

[0111] For example, the transfer cost can be expressed as:

[0112]

[0113] Further, step S304, the step of calculating the bulldozer distance EMD between the k-th HRRP data and the (k+1)-th HRRP data based on the transfer cost, includes:

[0114] Calculate the sum of transfer costs between each strong scattering point in the k-th HRRP data and each strong scattering point in the (k+1)-th HRRP data to obtain the bulldozer distance EMD between the k-th HRRP data and the (k+1)-th HRRP data.

[0115] In this embodiment, the similarity S between the k-th HRRP data and the (k+1)-th HRRP data is calculated according to the following formula. k :

[0116]

[0117] In the formula, EMD(H) k H k+1 MaxDistance(H) represents the bulldozer distance between the k-th HRRP data and the (k+1)-th HRRP data. k H k+1 ) represents the farthest distance between distance cells in the k-th HRRP data and the (k+1)-th HRRP data, where MaxDistance(H k H k+1 The greater of the following values ​​is: the distance between the first distance cell in the k-th HRRP data and the last distance cell in the (k+1)-th HRRP data, or the distance between the last distance cell in the k-th HRRP data and the first distance cell in the (k+1)-th HRRP data.

[0118] Below, the robustness of signal HRRP data and noise HRRP data is compared using the radar high-resolution range image data quality evaluation method provided by this invention.

[0119] Figure 8 This is a flowchart illustrating the robustness comparison between signal HRRP data and noise HRRP data provided in an embodiment of the present invention. Figure 8 As shown, the robustness of the two is compared by calculating the mean and variance of the similarity sets of N signal HRRP data and N noise HRRP data respectively.

[0120] Figure 9 This is a schematic diagram illustrating the similarity between comparative signal HRRP data and noise HRRP data provided in an embodiment of the present invention. Please refer to... Figure 9 The horizontal axis represents the index of similarity in the similarity set, and the vertical axis represents the similarity. The mean values ​​of the similarity sets of N signal HRRP data and N noise HRRP data are 0.9754 and 0.9232, respectively. The similarity of the signal HRRP data is significantly higher than that of the noise HRRP data.

[0121] Furthermore, the variances of the similarity sets of N-times signal HRRP data and N-times noise HRRP data are 8.2076e-05 and 93.9919e-05, respectively. This means that the similarity distribution of noise HRRP data fluctuates greatly, which is consistent with the fact that its data quality is poor and its robustness is poor, while the similarity distribution of signal HRRP data fluctuates less, which is consistent with the fact that its data quality is good and its robustness is good.

[0122] To verify that calculating similarity using strong scattering points from HRRP data improves computational performance compared to directly calculating 256 HRRP data points, this embodiment calculates N-1 similarities based on strong scattering points and 256 HRRP data points in the Matlab R2023b software environment. The average time taken is 0.0251s and 8.0656s, respectively. It can be seen that the computational performance is improved by 321 times while ensuring the effectiveness of the method.

[0123] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0124] This invention provides a method for evaluating the quality of high-resolution range image data of radar, which can be used to evaluate the robustness of HRRP data. Compared with the HRRP signal-to-noise ratio used in the prior art, the evaluation results of this invention are more objective, and there is no need to use a display device to observe the HRRP image, which helps to reduce costs.

[0125] Furthermore, calculating similarity by selecting strong scattering points from the HRRP data improves computational performance compared to directly calculating 256 points of HRRP data.

[0126] In the description of this invention, the term "first" is used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0127] The use of terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples" indicates that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. In addition, those skilled in the art can combine and integrate the different embodiments or examples described in this specification.

[0128] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A method for evaluating the quality of high-resolution radar range image data, characterized in that, include: Acquire a set of N consecutive high-resolution range profiles (HRRP) data; Select strong scattering points from each HRRP dataset and generate the first set. The first set This includes the distance cell position of each of the strong scattering points and its normalized amplitude; Based on the first set corresponding to each HRRP data Calculate the similarity between all two adjacent HRRP data sets in N HRRP data sets to obtain a similarity set; Based on the aforementioned similarity set, calculate the mean and variance of the similarity between all two adjacent HRRP data sets, and evaluate the robustness of the HRRP data. Based on the first set corresponding to each HRRP data The steps for calculating the similarity between all adjacent HRRP data points in N HRRP datasets to obtain a similarity set include: Get the k The first set corresponding to the HRRP data and the k The first set corresponding to +1 HRRP data According to the k HRRP data and the first k The distance cell positions of each strong scattering point in the +1 HRRP data are calculated to obtain the first... k Each strong scattering point in the second HRRP data and the first k Transfer distance between strong scattering points in +1 HRRP data According to the first k HRRP data and the first k The normalized amplitudes of each strong scattering point in the +1 HRRP data are obtained by solving the linear programming problem. k Each strong scattering point in the second HRRP data and the first k The transfer amplitude between strong scattering points in the +1 HRRP data; based on the transfer distance and the transfer amplitude, calculate the... k Each strong scattering point in the second HRRP data and the first k The transfer cost between strong scattering points in the +1 HRRP data; calculate the transfer cost based on the transfer cost. k HRRP data and the first k +1 bulldozer distance EMD between HRRP data points; calculate the first bulldozer distance EMD based on the bulldozer distance EMD. k HRRP data and the first k Similarity S of +1 HRRP data k After traversing the N HRRP data sets, we obtain N-1 similarity sets between two consecutive HRRP data sets. ],in, They represent the first Similarity between the next HRRP data and the next HRRP data; Get the k The first set corresponding to the HRRP data and the k The first set corresponding to +1 HRRP data According to the k HRRP data and the first k The distance cell positions of each strong scattering point in the +1 HRRP data are calculated to obtain the first... k Each strong scattering point in the second HRRP data and the first k The steps for determining the transfer distance between strong scattering points in the +1 HRRP data include: Get the k The first set corresponding to the HRRP data and the k The first set corresponding to +1 HRRP data According to the first k HRRP data and the first k The distance cell location of each strong scattering point in the +1 HRRP data is calculated. k Each strong scattering point in the second HRRP data and the first k The Euclidean distance between strong scattering points in the +1 HRRP data is used as the... k Each strong scattering point in the second HRRP data and the first k The transfer distance between strong scattering points in the +1 HRRP data.

2. The method for evaluating the quality of high-resolution radar range image data according to claim 1, characterized in that, Select strong scattering points from each HRRP dataset and generate the first set. Before the steps, it also includes: preprocessing the HRRP data for each iteration; wherein, The steps for preprocessing HRRP data for each iteration include: Obtain the sampling frequency and radar bandwidth of each HRRP data, and downsample the HRRP data when the sampling frequency is greater than the radar bandwidth; Obtain the distance cell location corresponding to the maximum amplitude of the N / 2th HRRP data; The HRRP data for each iteration is windowed and cropped based on the distance cell position corresponding to the maximum amplitude.

3. The method for evaluating the quality of high-resolution radar range image data according to claim 2, characterized in that, Select strong scattering points from each HRRP dataset and generate the first set. The steps include: Calculate the maximum amplitude of each HRRP data point and multiply it by a preset coefficient to obtain the amplitude threshold value used to select strong scattering points; Determine the distance cell in each HRRP data whose amplitude is greater than the amplitude threshold, and select the scattering point in the distance cell as the strong scattering point; Record the range cell position and amplitude of each strong scattering point in each HRRP data set, and normalize the amplitude of the strong scattering points in each HRRP data set to generate the first set. : ； In the formula, These represent the first HRRP data. The distance cell location of each strong scattering point They represent the first The normalized amplitude of each strong scattering point.

4. The method for evaluating the quality of high-resolution radar range image data according to claim 1, characterized in that, According to the k HRRP data and the first k The normalized amplitudes of each strong scattering point in the +1 HRRP data are obtained by solving the linear programming problem. k Each strong scattering point in the second HRRP data and the first k The steps for calculating the transfer amplitude between strong scattering points in the +1 HRRP data include: Construct the following linear programming problem: ； Constraints: ； ； ； In the formula, m , n They represent the first k HRRP data and the first k The number of strong scattering points in +1 HRRP data Indicates the first k The first HRRP data in i The strong scattering point and the first k +1 HRRP data in the first j The transfer distance between strong scattering points Indicates the first k The first HRRP data in i The strong scattering point and the first k +1 HRRP data in the first j The transfer amplitude between strong scattering points Indicates the first k The first HRRP data in i The normalized amplitude of each strong scattering point Indicates the first k +1 HRRP data in the first j The normalized amplitude of each strong scattering point; By solving the linear programming problem, we obtain the first... k Each strong scattering point in the second HRRP data and the first k The transfer amplitude between strong scattering points in the +1 HRRP data.

5. The method for evaluating the quality of high-resolution radar range image data according to claim 4, characterized in that, Based on the transfer distance and the transfer amplitude, calculate the first... k Each strong scattering point in the second HRRP data and the first k The steps for transferring costs between strong scattering points in +1 HRRP data include: For the transfer amplitude and transfer distance Perform the dot product operation to obtain the first... k Each strong scattering point in the second HRRP data and the first k +1 HRRP data transfer cost between strong scattering points.

6. The method for evaluating the quality of high-resolution radar range image data according to claim 5, characterized in that, Calculate the first based on the transfer cost. k HRRP data and the first k The steps for bulldozer distance EMD between +1 HRRP data points include: Calculate the first k Each strong scattering point in the second HRRP data and the first k The sum of the transfer costs between strong scattering points in the +1 HRRP data yields the th k HRRP data and the first k Bulldozer Distance EMD between +1 HRRP data points.

7. The method for evaluating the quality of high-resolution radar range image data according to claim 5, characterized in that, Calculate the number according to the following formula. k HRRP data and the first k Similarity S of +1 HRRP data k : ； In the formula, Indicates the first k HRRP data and the first k Bulldozer distance between +1 HRRP data points Indicates the first k HRRP data and the first k The farthest distance between cells in +1 HRRP data, where, For: the k The first distance cell in the HRRP data and the second k The distance between the last distance unit in the +1 HRRP data, or the distance between the first and last distance units. k The last distance cell in the HRRP data is related to the first k The larger value of the distance between the first distance cells in the +1 HRRP data.

Images