Burst identification method based on radar range-doppler plane

Abstract

The application relates to a burst point recognition method based on a radar range-Doppler plane, which can be used for judging whether a target area is successfully hit by a shell and also can evaluate the training result of artillery shooting. The method determines the distribution area of the burst point in the plane by respectively performing a range dimension CFAR, range dimension and Doppler dimension binary accumulation and other multiple treatments on the radar range-Doppler plane of the obtained target, then extracts six kinds of features of the burst point distribution area, such as target Doppler dimension spread, center symmetry, total energy of the target, total energy of the target average image, average energy of the target average image and the number of unit cells occupied by the target, and finally inputs the feature vector into a support vector machine for burst point recognition. The application has excellent real-time performance, can recognize the burst point while the target area is hit by the shell, and can report detailed information such as the position of the burst point at the same time; the application does not need to measure the site after the war or training, and has no safety hidden danger.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of target recognition, and relates to a burst point recognition method based on a radar range-Doppler plane. BACKGROUND

[0002] With the continuous development of science and technology and the continuous improvement of the combat level worldwide, weapon equipment is developing in the direction of informatization, intelligentization and automation, etc., among which, burst point recognition is the basis for shooting calibration and damage assessment, and the burst point position and the position deviation between the burst point and the target are also one of the important indicators for evaluating the combat effectiveness of the target on the ground. Reasonably mastering and utilizing the information of the burst point can help to correct the projection process and improve the projection level, and can further improve the combat capability of the army. Based on this factor, the automatic recognition of the burst point is an important link in the construction of the intelligent and digital target range of the army.

[0003] At present, the detection of the burst point position in the target range still mainly adopts the method of manual measurement. Specifically, after shooting, the measurement personnel uses global positioning system or laser level and other tools to observe and measure on site. However, this method has many problems, mainly including poor real-time performance, safety hazards, low measurement accuracy and the like. In recent years, with the rapid development of digital image recognition technology, it has also been applied to the field of burst point recognition by domestic and foreign scholars, and many effective algorithms have been proposed, including but not limited to frame difference method, optical flow method, background pruning method and block matching method. However, the algorithms proposed at present are designed for specific places, and there is no algorithm suitable for all backgrounds. Similarly, with the continuous development of infrared technology and the deepening of its application in the field of military defense, some scholars have proposed that due to the temperature difference and infrared radiation characteristics between the shell and the background, the infrared wave radiated by the burst point can pass through the infrared camera, while the background cannot, based on which the burst point and the background can be distinguished. However, the existing various detection and positioning methods are also proposed for specific scenarios, and are highly targeted with certain limitations.

[0004] When the burst point is hit by the shell, an explosion circle with the shell hitting position as the center and surrounded by fast moving fragments will appear on the ground. These fragments are densely distributed in the initial explosion stage, have fast and slow moving speeds, are distributed in a large range, and the moving direction is 360° full coverage. The distance and speed information of the fragment distribution can be reflected in the radar range-Doppler plane. Therefore, by analyzing the burst point area on the radar range-Doppler plane, extracting features, and through a specific classifier, the burst point recognition based on the radar range-Doppler plane can be realized. SUMMARY

[0005] The technical problem solved by the present application is:

[0006] In order to overcome the defects of the existing burst point identification method, which is limited by the scene, the present application provides a burst point identification method based on radar range-Doppler plane, which uses threshold method to determine the burst point area in the radar range-Doppler plane, extracts multi-class features of the burst point, and realizes the identification of the burst point through a specific classifier.

[0007] In order to solve the above technical problem, the technical scheme adopted by the present application is:

[0008] A burst point identification method based on radar range-Doppler plane, characterized by the following steps:

[0009] Step 1: Obtain the radar range-Doppler plane of the target;

[0010] Step 2: Preprocess the plane image, including range dimension CFAR and range dimension binary accumulation;

[0011] Step 3: Determine the range dimension boundary of the burst point distribution area according to the binary accumulation vector;

[0012] Step 4: Determine the Doppler dimension boundary of the burst point data according to the range dimension boundary;

[0013] Step 5: Determine the burst point area according to the range dimension boundary and the Doppler dimension boundary;

[0014] Step 6: Extract six types of features from the burst point distribution area; the six types of features include target Doppler dimension spread, center symmetry, total energy of the target, total energy of the target average image, average energy of the target average image, and the number of cells occupied by the target;

[0015] Step 7: Input the features into the support vector machine for burst point identification.

[0016] Further technical scheme of the present application: the radar range-Doppler plane of the target in step 1 is a matrix M, with a size of d*r, d represents the length of the Doppler dimension, different Doppler values correspond to different radial velocities, and r represents the length of the range dimension, representing the distance between the target and the radar.

[0017] Further technical scheme of the present application: the binary accumulation in step 2 is specifically:

[0018] Assign 1 to the values in matrix M that exceed the range dimension CFAR threshold, and assign 0 to the values that do not exceed the threshold, sum the values of all Doppler channels along the range dimension, and a vector with a length of r is obtained, called binary accumulation vector, denoted as S=(s1,s2,s3,…s r ).

[0019] A further technical solution of the present invention: Step 3 is as follows:

[0020] First, the threshold value for the distance to the detonation point is determined to be one-fifth of the number of Doppler accumulation points:

[0021] Next, starting from the initial value s1 of the vector, the judgment is made. Positions that exceed the threshold value are considered to have potential explosion points, and their corresponding distance dimension r is remembered.

[0022] Since a range-Doppler image cannot be assumed to have only one exploded point, the range dimension boundaries to be determined are not necessarily limited to one set. Each set of range dimension boundaries includes the minimum value R of the range dimension. min With the maximum value R max Two data points;

[0023] Before performing vector value judgment, the distance dimension boundary is assumed to be 0 groups, meaning there are 0 explosion points. num =0; the position of the vector value judgment check num Starting from the first value in the vector, i.e., check num =1; Sets the flag isR for determining whether to check the boundary of the new distance dimension. min A value of 1 indicates that a new detonation point boundary needs to be determined, while a value of 0 indicates that a new detonation point boundary does not need to be determined.

[0024] check num Increment by 1 sequentially; it is necessary to determine the check of the binary accumulation vector. num Does the value exceed the threshold value th? r When the threshold value is exceeded, determine isR. min Is it 1?

[0025] a)isR min 1:

[0026] The number of explosion point boundaries needs to be increased by 1, i.e., ZD. num ++;

[0027] Minimum value of distance dimension R min With the maximum value R max Assign the current distance value, R min =check num R max =check num ;

[0028] Next, it is necessary to determine the maximum value R of the boundary distance of the explosion point. max Therefore, when the next vector value exceeds the threshold, a new explosion point boundary should not be created, and the marker of the new distance dimension boundary, isR, should be set. min Set to 0;

[0029] b) isR min is 0:

[0030] check num isR max is less than or equal to 2, it is considered that the blast point range is still within the same range, and the blast point boundary distance dimension maximum value R max is updated to the latest check num , if the difference is more than 2, it is considered that the judgment distance has exceeded the range of the blast point, and it is necessary to judge whether there is a next blast point, so the new distance dimension boundary flag isR min is set to 1.

[0031] The further technical scheme of the present application is that step 4 is specifically:

[0032] The blast point data in each distance dimension boundary is sequentially accumulated along the Doppler dimension in binary, and a column vector with a length of Doppler dimension is sequentially obtained, the Doppler threshold value is set to 2, the positions of the first points exceeding the threshold value at the upper and lower ends (D up ,D down ) are determined, the difference values between the upper and lower positions and the starting position and the end position of the vector are calculated, that is, (|D up -D start |, |D end -D down |), and the minimum difference value is the channel Del num , that is

[0033] Del num = min(|D up -D start |, |D end -D down |)

[0034] The data is subtracted by the corresponding channel number Del num , and the obtained Doppler minimum value D min and maximum value D max are the Doppler boundaries of the group of blast point data:

[0035] (D min ,D max )=(D start +Del num ,D end -Del num ).

[0036] The further technical scheme of the present application is that: step 5 is specifically: four coordinates of each burst point in the distance-Doppler plane are obtained from step 3 and step 4, and a region of a burst point in the plane is determined according to the four coordinate information.

[0037] The further technical scheme of the present application is that: the target Doppler spread: the number of cells occupied by the determined region in the Doppler velocity;

[0038] The center symmetry: assuming that the distribution of the determined region is a matrix Sig, containing m*n elements, the matrix is sequentially flipped horizontally and vertically, and a flipped matrix Sig flip When the Doppler distribution of the sample completely presents the center symmetry, the difference between the flipped matrix and the original matrix is 0, and the center symmetry feature calculation formula is as follows:

[0039]

[0040] The total energy of the target: the cell in which the value higher than the distance dimension CFAR threshold η in the matrix Sig is the cell where the target is located, the cell where the target is located is extracted, the energy value of each cell is calculated and summed, and the total energy value of the target can be calculated:

[0041] T E =∑|sig>η| 2

[0042] The total energy of the target average image: the values of each row vector in the Doppler matrix Sig are sorted from large to small, and a matrix Sig sort is obtained, and the internal elements of the matrix are specifically represented as:

[0043]

[0044] The average value of the first j element values of each row vector in the matrix Sig sort , a column vector Sig mean containing m elements can be obtained, which is called the target average image:

[0045]

[0046] The calculation formula of the total energy of the target average image is specifically:

[0047]

[0048] The calculation process of the average energy of the target average image is:

[0049]

[0050] Wherein, m is the length of the target average image.

[0051] The target-occupied cell number calculation process is specifically that the cell where the value higher than the distance dimension CFAR threshold value η in the matrix Sig is located is considered as the cell where the target is located, and the total of the target-occupied cells is calculated, that is, the target-occupied cell number.

[0052] A further technical solution of the application is that the support vector machine in step 7 is a binary classification vector machine.

[0053] The application has the beneficial effects that:

[0054] The application provides a burst point recognition method based on a radar range-Doppler plane, which can be used for judging whether a target area is successfully hit by a shell in real time in any ground environment, and can also be used for evaluating a result of artillery shooting training in real time. The application solves the problems of the current burst point detection, such as poor real-time performance, safety hazards and limited applicable environment of the current algorithm. The application determines the distribution area of the burst point in the plane by performing a plurality of processing, such as range dimension CFAR, range dimension and Doppler dimension binary accumulation, on the radar range-Doppler plane of the acquired target, extracts six types of features, such as target Doppler dimension spread, center symmetry, total energy of the target, total energy of the target average image, average energy of the target average image and target-occupied cell number, from the burst point distribution area, and finally inputs the feature vector into a support vector machine for burst point recognition.

[0055] Compared with the prior art, the application has the beneficial effects as follows:

[0056] 1) excellent real-time performance, which can recognize the burst point while the target area is hit by the shell, and can report the position and other detailed information of the burst point at the same time;

[0057] 2) no need to measure the site after the war or training, and no safety hazards;

[0058] 3) no special requirements for applicable ground environment. BRIEF DESCRIPTION OF DRAWINGS

[0059] The accompanying drawings are only used for the purpose of illustrating specific embodiments and are not considered as limiting the application, and in the whole drawings, the same reference signs represent the same parts.

[0060] Figure 1 Flowchart of the burst point recognition method based on the radar range-Doppler plane;

[0061] Figure 2 Schematic diagram of range dimension binary accumulation;

[0062] Figure 3 Flowchart of determining the range dimension boundary of the burst point;

[0063] Figure 4 Coordinate determination of the bomb point area. DETAILED DESCRIPTION

[0064] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application. In addition, the technical features involved in the various embodiments of the present application described below can be combined with each other as long as they do not conflict with each other.

[0065] The present application provides a bomb point recognition method based on radar range-Doppler plane, first acquires the radar range-Doppler plane of the target, then determines the distribution area of the bomb point in the plane through the range dimension CFAR, the binary accumulation of the range dimension and the Doppler dimension respectively, extracts six kinds of features such as the Doppler dimension spread of the target, the center symmetry, the total energy of the target, the total energy of the target average image, the average energy of the target average image, and the number of unit cells occupied by the target from the bomb point distribution area, and inputs the features into the support vector machine to complete the recognition of the bomb point.

[0066] Figure 1 For the implementation step flowchart of the present application, the implementation steps of the present application are as follows:

[0067] (1) Acquire the radar range-Doppler plane of the target;

[0068] (2) Preprocess the plane image, including range dimension CFAR and range dimension binary accumulation;

[0069] (3) Determine the range dimension boundary of the bomb point distribution area;

[0070] (4) Determine the Doppler dimension boundary of the bomb point distribution area;

[0071] (5) Determine the bomb point distribution area in the plane;

[0072] (6) Extract six kinds of features from the bomb point distribution area;

[0073] (7) Input the features into the support vector machine for bomb point recognition.

[0074] The above steps are as follows:

[0075] (1) Acquire the radar range-Doppler plane containing the bomb point.

[0076] The radar range-Doppler plane containing the bomb point is a matrix M, whose size is d*r, d represents the length of the Doppler dimension, different Doppler values correspond to different radial velocities, and r represents the length of the range dimension, representing the distance between the target and the radar.

[0077] (2) Preprocessing of the plan view image.

[0078] a) Range dimension CFAR on the radar range-Doppler plan view.

[0079] b) Binary accumulation along the range dimension after CFAR.

[0080] The values in matrix M that exceed the range dimension CFAR threshold are assigned a value of 1 and those that do not exceed the threshold are assigned a value of 0. The values along the range dimension for all Doppler bins are summed as shown in Figure 2 , which results in a set of vectors of length r, called binary accumulation vectors, denoted as S = (s1, s2, s3,... s r ).

[0081] (3) Determining the range dimension boundary of the burst region based on the binary accumulation vectors.

[0082] The flow chart of determining the range dimension boundary of the burst region based on the binary accumulation vectors is shown in Figure 3 . First, the burst range threshold is determined to be one fifth of the number of Doppler accumulation points;

[0083]

[0084] Second, the initial value s1 of the vector is used to determine the position of the suspected burst. If the value exceeds the threshold, the corresponding range dimension r is recorded.

[0085] Because it cannot be assumed that there is only one burst in a range-Doppler image, the determined range dimension boundary may not be only one set. Each set of range dimension boundary includes two data, the minimum range dimension R min and the maximum range dimension R max .

[0086] As shown in Figure 3 , before the value of the vector is determined, the range dimension boundary is first assumed to be 0, i.e., there are 0 bursts, ZD num = 0. The position of the vector value check num is determined from the first value of the vector, i.e., check num = 1. The flag isR min is set to 1, which indicates that a new burst boundary needs to be determined, or 0, which indicates that a new burst boundary does not need to be determined.

[0087] check num is incremented by 1, and the check num th r value of the binary accumulation vector is determined. When the value exceeds the threshold, isR minisR is 1:

[0088] a) isR min is 1:

[0089] The number of the shot boundary needs to be added 1 group, that is, ZD num ++;

[0090] The minimum value of the distance dimension R min and the maximum value R max is assigned to the current distance value, R min = check num , R max = check num ;

[0091] After that, it is necessary to determine the maximum value of the shot boundary distance dimension R max Therefore, when the next vector value exceeds the threshold value, a new shot boundary should not be created, at which time the flag isR min of the new distance dimension boundary is set to 0.

[0092] b) isR min is 0:

[0093] check num and R max is calculated, if the difference is less than or equal to 2, it is considered to be within the same shot range, and the maximum value of the shot boundary distance dimension R max is updated to the latest check num , if the difference is more than 2, it is considered that the distance has exceeded the range of the shot, and it is necessary to determine whether there is a next shot, so the flag isR min of the new distance dimension boundary is set to 1.

[0094] (4) Determine the Doppler dimension boundary of the shot data according to the distance dimension boundary of the shot.

[0095] The shot data in each distance dimension boundary is sequentially accumulated along the Doppler dimension in binary, and a column vector with a length of the Doppler dimension can be obtained in sequence. The Doppler threshold value is set to 2, the positions of the first points exceeding the threshold value at the upper and lower ends (D up , D down ) are determined, the differences between the upper and lower positions and the starting position and the end position of the vector are calculated, that is, (|D up -D start |, |D end -D down |), and the minimum difference is taken as the channel Del num , that is

[0096] Del num = min(|Dup -D start |,|D end -D down |)

[0097] Subtract the corresponding channel number Del from the data above and below num , and the obtained Doppler minimum value D min and maximum value D max are the Doppler boundaries of the group of burst data:

[0098] (D min ,D max )=(D start +Del num ,D end -Del num )

[0099] (5) Determine the burst region according to the distance dimension boundary and the Doppler dimension boundary.

[0100] From the operation steps of steps (3) and (4), four coordinates (R min ,R max ,D min ,D max ) of each burst in the range-Doppler plane can be obtained, as shown in FIG. 6. According to the four coordinate information, the region of a burst in the plane can be determined. Figure 4

[0101] (6) Extract features of the burst distribution region.

[0102] The data within the boundary of each group of burst data can be extracted for feature extraction. The extracted features include: target Doppler dimension spread, center symmetry, total energy of the target, total energy of the target average image, average energy of the target average image, and the number of cells occupied by the target. The extracted features of each group of bursts can form a feature vector containing six feature elements.

[0103] a) Feature 1: Target Doppler dimension spread

[0104] The target Doppler dimension spread is the number of cells occupied by the determined region in the Doppler dimension. Because the clutter and other targets including pedestrians, vehicles, and unmanned aerial vehicles have a narrow spread in the Doppler dimension, and the burst has a wide spread, the Doppler dimension spread of the target image can be extracted as a feature to separate the burst from other targets.

[0105] b) Feature 2: Center symmetry

[0106] ​The difference between the target Doppler signal rotated 180° and the original signal can reflect the central symmetry of the target data. Since the distribution of the burst point in the range-Doppler plane is always centrally symmetric, while other targets and clutter do not exhibit such characteristics, this feature can classify the burst point from clutter and other targets that do not exhibit central symmetry.

[0107] Assuming that the distribution of the determined area is a matrix Sig containing m x n elements, the matrix is sequentially flipped horizontally and vertically to obtain the flipped matrix Sig flip When the Doppler distribution of the sample completely exhibits central symmetry, the difference between the flipped matrix and the original matrix is 0, so the central symmetry feature calculation formula is as follows:

[0108]

[0109] The stronger the central symmetry of the Doppler distribution of the sample, the smaller the feature value.

[0110] c) Feature 3: Total energy of the target

[0111] By comparing the Doppler distribution of the burst point and the clutter, it can be found that the energy of the burst point is much stronger than that of the clutter, so the total energy of the target can be used as a feature to classify the burst point and the clutter.

[0112] The calculation process of this feature is as follows: consider the cells in the matrix Sig that are higher than the range dimension CFAR threshold η in step (2) as the cells where the target is located, extract the cells where the target is located, calculate the energy value of each cell and sum them up, and then the total energy value of the target can be calculated. The calculation formula is as follows:

[0113] T E =∑|sig>η| 2

[0114] The stronger the intensity of the target in the sample, the more cells it occupies, and the stronger the energy of the target, the larger the feature value.

[0115] d) Feature 4: Total energy of the target average image

[0116] When calculating the average image of the target, the rule adopted is to average part of the range dimension of the matrix Sig, and the final average image can be expanded in the velocity dimension. The specific process is as follows:

[0117] Sort the values of each row vector in the Doppler matrix Sig from large to small to obtain the matrix Sig sort , the internal elements of the matrix are specifically represented as:

[0118]

[0119] Matrix Sig is obtained sort The average value of the first j elements of each row vector in the matrix Sig is calculated, and a column vector Sig containing m elements is obtained mean , which is called the target average image

[0120]

[0121] The target average image also contains partial characteristics of the target, including the spread of the target on the Doppler axis and the energy of the target. Therefore, by calculating the total energy of the target average image, the shot point can be classified with the clutter. The formula for calculating the total energy of the target average image is as follows:

[0122]

[0123] The stronger the energy of the target, the greater the value of this feature.

[0124] e) Feature 5: Average energy of target average image

[0125] Because the length of the target average image is m, the average energy of the average image also contains the same target characteristics as the total energy, and the calculation process of the average energy is as follows:

[0126]

[0127] The stronger the energy of the target, the stronger the average energy, and the greater the value of this feature.

[0128] f) Feature 6: Number of cells occupied by the target

[0129] In the previous shot point area analysis, it is known that the energy of the shot point is much stronger than that of the clutter, not only in the size of the intensity, but also in the number of cells occupied. The number of cells occupied by the shot point is greater than that of the clutter, and the continuity is stronger than that of the clutter. Therefore, the number of cells occupied by the target can be used as a feature for the identification and classification of shot points and clutter.

[0130] The calculation process of this feature is as follows: the cells in the matrix Sig that are higher than the distance dimension CFAR threshold η in step (2) are considered to be the cells occupied by the target, and the sum of the cells occupied by the target is calculated, which is the number of cells occupied by the target.

[0131] (7) Input the feature vector into the support vector machine for shot point recognition.

[0132] The support vector machine used in this method is a binary classification vector machine. Each feature vector calculated in step (6) is input into the binary classification support vector machine, and the output result is 1, indicating that the group of area recognition results is a shot point, and the output result is -1, indicating that the recognition result is not a shot point.

[0133] The above merely provides the specific implementation of the present application, but the protection scope of the present application is not limited thereto, and any skilled person in the art can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present application, and these modifications or replacements should be covered within the protection scope of the present application.

Claims

1. A burst identification method based on radar range-Doppler plane, characterized in that The steps are as follows: Step 1: Obtain the radar range-Doppler plane of the target; Step 2: Preprocess the plane image, including range dimension CFAR and range dimension binary accumulation; Step 3: Determine the range dimension boundary of the burst point distribution area according to the binary accumulation vector; specifically as follows: First, the threshold of the distance of the burst point is determined as one fifth of the number of Doppler accumulation points: ; Second, from the initial value of the vector Start to determine, beyond the threshold value of the position is considered to exist suspected explosion point, its corresponding distance dimension r Remember; Because it cannot be assumed that there is only one burst in a range-Doppler image, the range dimension of the decision does not necessarily have only one set of boundaries, each set of range dimension boundaries including a minimum value of the range dimension and a maximum value two data; First, the distance dimension boundary is set as 0 group, i.e. there are 0 burst points, before the vector value judgment is performed, ; the position of the vector value judgment Starting from the first value of the vector, i.e. ; Flag to set whether to judge new distance dimension boundary is 1, indicating that a new burst boundary needs to be determined, and is 0, indicating that a new burst boundary does not need to be determined; 1 is added successively, it is needed to judge whether the value of the binary accumulation vector is over the threshold value When the threshold value is over, it is judged whether it is 1:​ a) is 1 : The number of burst boundaries needs to be incremented by 1 group, i.e. ; distance dimension minimum value with maximum value assigned the current distance value, , ; The distance dimension maximum value of the burst point boundary needs to be determined Therefore, when the next vector value exceeds the threshold value, a new burst point boundary should not be created, and the flag of the new distance dimension boundary is set to 0; b) is 0: With Difference calculation, if the difference is less than or equal to 2, it is considered to be within the same range of the burst point, and the maximum value of the burst point boundary distance Update to the latest , if the difference is more than 2, it is considered that the judgment distance has exceeded the range of the burst point, and it is necessary to judge whether there is a next burst point, so the flag of the new distance dimension boundary Is set to 1; Step 4: Determine the Doppler dimension boundary of the burst point data according to the range dimension boundary; Step 5: Determine the burst point area according to the range dimension boundary and the Doppler dimension boundary; Step 6: Extract six types of features from the burst point distribution area; The six types of features include target Doppler dimension spread, center symmetry, total energy of the target, total energy of the target average image, average energy of the target average image, and the number of unit cells occupied by the target; Step 7: Input the features into the support vector machine for burst point recognition.

2. The radar range-Doppler plane based burst identification method of claim 1, wherein: The radar range-Doppler plane of the targets described in step 1 is a matrix M with dimensions d*r , d denotes the length of the Doppler dimension, different Doppler values correspond to different radial velocities, r denotes the length of the range dimension, representing the distance of the target from the radar.

3. The radar range-Doppler plane based burst identification method of claim 2, wherein: The binary accumulation in step 2 is specifically as follows: The matrix M is obtained by assigning 1 to the values of the matrix that exceed the range dimension CFAR threshold and 0 to the values that do not exceed the threshold. The sum of the values along the range dimension for all Doppler bins yields a set of vectors of length r , called binary accumulation vectors, denoted as .

4. The radar range-Doppler plane based burst identification method of claim 3, wherein: Step 4 is specifically as follows: The binary accumulation is performed along the Doppler dimension for the shot data within each group of distance dimension boundaries in sequence to obtain a group of column vectors with the length of the Doppler dimension in sequence, the Doppler threshold value is set to 2, the positions of the first points exceeding the threshold value at the upper and lower ends are determined , the difference values between the upper and lower positions and the start and end positions of the vector are calculated respectively, that is , and the channel with the minimum difference value is taken as the subtraction channel , that is Subtracting the corresponding channel number from the data up and down The minimum and maximum values of Doppler The minimum and maximum values of Doppler are the Doppler boundaries of the shot data 。 5. The radar range-Doppler plane based burst identification method of claim 4, wherein: Step 5 is specifically: four coordinates of each burst point in the range-Doppler plane are obtained from steps 3 and 4, and the area of a burst point in the plane is determined according to the four coordinate information.

6. The radar range-Doppler plane based burst identification method of claim 5, wherein: The target Doppler dimension spread: the number of unit cells occupied by the determined area in the Doppler dimension; The center symmetry: assuming that the distribution of the determined area is a matrix , containing elements, the matrix is sequentially flipped horizontally and vertically to obtain a flipped matrix When the Doppler distribution of the sample completely exhibits center symmetry, the difference between the flipped matrix and the original matrix is 0, and the center symmetry feature calculation formula is as follows: The total energy of the target: matrix The cell whose value is higher than the distance dimension CFAR threshold value is the cell where the target is located The cell whose value is higher than the distance dimension CFAR threshold value is the cell where the target is located The cell whose value is higher than the distance dimension CFAR threshold value is the cell where the target is located The total energy of the target average image: the Doppler matrix The values of each row vector in the matrix are sorted from large to small to obtain a matrix The internal elements of the matrix are specifically represented as matrix before each row vector inside the average of the values of the elements, a set of column vectors containing the target average image: The calculation formula of the total energy of the target average image is specifically as follows: The calculation process of the average energy of the target average image is as follows: Wherein, m is the length of the target average image; The target occupied cell number calculation process is specifically: considering that the cell in which the value of the matrix higher than the distance dimension CFAR threshold is the cell in which the target is located, and calculating the total number of cells occupied by the target is the target occupied cell number. ​ 7. The radar range-Doppler plane based burst identification method of claim 6, wherein: The support vector machine in step 7 is a binary classification vector machine.

Images