Non-uniform sea clutter partitioning method and device based on M-order moment feature set

Abstract

The invention belongs to the technical field of radar signal processing, and particularly relates to a non-uniform sea clutter partitioning method and device based on an M-order moment feature set. The method comprises the following steps: S1, carrying out distance and azimuth grid division on sea clutter echoes to form K grids, and determining a distance center R (k) and an azimuth angle center Az (k) of the kth grid; s2, calculating an M-order moment feature of the kth grid; s3, constructing a sea clutter partition feature vector set based on the distance center R (k), the azimuth angle center Az (k) and M-order moment features of the kth grid; s4, the sea clutter partition feature vector set is clustered, an L * K partition matrix U is obtained, and L is an expected partition number; and S5, dividing the grids corresponding to the columns where the same row elements with the same maximum value as each column are located in the partition matrix U into the same partition. According to the invention, the accuracy of parameter estimation of a subsequent detection algorithm is improved.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of radar signal processing, and particularly relates to a non-uniform sea clutter partitioning method and device based on an M-order moment feature set. BACKGROUND

[0002] When an airborne radar detects the sea, the received sea clutter characteristics change with the wind field on the sea surface, the energy distribution of the sea wave, the geometric structure of the airborne radar observing the sea surface, and the like, and present the characteristics of spatial non-uniformity and time variation. At this time, target detection in the sea clutter has the following problems:

[0003] (1) In a conventional adaptive target detection method, it is assumed that the entire sea surface obeys a single sea clutter model, but the complexity of the sea clutter makes it difficult to describe the sea clutter in the radar search range by using a single model, which leads to the problem that the clutter model assumed by the detection algorithm deviates from the real clutter model, thereby causing large differences in detection performance in different regions.

[0004] (2) The fluctuation degree of each region of the sea clutter is different, and the data in the reference unit at the strong-weak boundary of the sea clutter is reduced in parameter estimation accuracy due to non-independent and identically distributed, which leads to more false alarms or missed alarms of the conventional target detection algorithm using fixed processing parameters. SUMMARY

[0005] To solve the above problems, the application provides a non-uniform sea clutter partitioning method and device based on an M-order moment feature set, which performs grid division on the sea clutter in the space domain, constructs a local uniform environment, extracts the M-order moment statistics of each grid to represent the statistical distribution characteristics of the sea clutter, and performs fine partitioning on the clutter region based on the M-order moment statistics.

[0006] The first aspect of the application provides a non-uniform sea clutter partitioning method based on an M-order moment feature set, mainly comprising:

[0007] Step S1, performing grid division on sea clutter echoes in the range and azimuth directions to form K grids, and determining the range center R(k) and the azimuth center Az(k) of the kth grid;

[0008] Step S2, calculating the M-order moment feature of the kth grid ;

[0009] Step S3, constructing a sea clutter partitioning feature vector set based on the range center R(k), the azimuth center Az(k), and the M-order moment feature of the kth grid ;

[0010] Step S4, clustering the sea clutter partitioning feature vector set to obtain a partitioning matrix U of L*K, wherein L is the expected number of partitions.

[0011] Step S5, dividing the grids corresponding to the columns in which the elements in the same row as the maximum value of each column in the partition matrix U are located into the same partition.

[0012] Preferably, step S2 further comprises:

[0013] Step S21, calculating the m-th moment feature of the k-th grid:

[0014] ;

[0015] wherein N is the number of echo samples, is the n-th echo sample;

[0016] Step S22, constructing the M-th moment feature of the sea clutter echo of the k-th grid:

[0017] .

[0018] Preferably, in step S2, the dimension M of the M-th moment feature is 3-5.

[0019] Preferably, step S3 further comprises:

[0020] Step S31, constructing the sea clutter feature set S based on the distance center R(k), the azimuth angle center Az(k) and the M-th moment feature of the k-th grid .

[0021] ;

[0022] wherein ;

[0023] Step S32, performing weighted processing on the sea clutter feature set S to obtain a sea clutter partition feature vector set .

[0024] ;

[0025] wherein W is a weighting vector, and the dimension of W is consistent with that of the feature vector .

[0026] Preferably, step S5 further comprises:

[0027] Step S51, determining the maximum value of the k-th column in the partition matrix U;

[0028] Step S52, finding one or more elements identical to the maximum value in the row l in which the maximum value is located;

[0029] Step S53, determining the column sequence number in which the element is located;

[0030] Step S54: Divide the grid corresponding to one or more column numbers into the same partition.

[0031] The second aspect of this application provides a non-uniform sea clutter partitioning device based on an M-order moment feature set, mainly comprising:

[0032] The grid parameter determination module is used to divide the sea clutter echo into range and azimuth grids, forming K grids, and determine the range center R(k) and azimuth center Az(k) of the kth grid.

[0033] The M-order moment feature calculation module is used to calculate the M-order moment feature of the k-th grid. ;

[0034] The partition feature vector set construction module is used to construct features based on the distance center R(k), azimuth center Az(k), and M-order moment features of the k-th grid. Constructing a sea clutter partition feature vector set ;

[0035] The clustering module is used to partition the feature vector set of sea clutter. Perform clustering to obtain an L*K partition matrix U, where L is the desired number of partitions;

[0036] The partitioning module is used to divide the grid corresponding to the row elements in the partitioning matrix U that have the same maximum value in each column into the same partition.

[0037] Preferably, the M-order moment feature calculation module includes:

[0038] The moment feature calculation unit is used to calculate the m-th moment feature of the k-th grid.

[0039] ;

[0040] Where N is the number of echo samples, This is the nth echo sample;

[0041] The M-order moment feature calculation unit is used to construct the M-order moment feature of the sea clutter echo in the k-th grid:

[0042] .

[0043] Preferably, the dimension M of the M-order moment feature is 3 to 5.

[0044] Preferably, the partition feature vector set construction module includes:

[0045] Sea clutter feature set construction unit, used for the distance center R(k), azimuth center Az(k), and M-order moment features based on the k-th grid. Constructing the sea clutter feature set S:

[0046] ;

[0047] in, ;

[0048] The weighted processing unit is used to perform weighted processing on the sea clutter feature set S to obtain the sea clutter partition feature vector set. :

[0049] ;

[0050] Where W is the weighted vector and the eigenvector. The dimensions are consistent.

[0051] Preferably, the partitioning module includes:

[0052] The maximum value determination unit is used to determine the maximum value in the k-th column of the partition matrix U;

[0053] The same value lookup unit is used to find one or more elements that are the same as the maximum value in row l where the maximum value in column k is located;

[0054] The column number determination unit is used to determine the column number of the element.

[0055] A grid defining unit is used to divide the grid corresponding to one or more column numbers into the same partition.

[0056] This application realizes the regionalized characterization of sea clutter, effectively improves the mismatch between the detection algorithm model and the radar observation global sea clutter model, and enhances the accuracy of subsequent detection algorithm parameter estimation. Attached Figure Description

[0057] Figure 1 This is a flowchart of a preferred embodiment of the non-uniform sea clutter partitioning method based on the M-order moment feature set of this application.

[0058] Figure 2 This is a schematic diagram of a distance-angle two-dimensional grid. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this application will be described in more detail below with reference to the accompanying drawings. In the drawings, the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The described embodiments are only some, not all, of the embodiments of this application. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application. The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0060] The first aspect of this application provides a non-uniform sea clutter partitioning method based on the M-order moment feature set, such as... Figure 1 As shown, it mainly includes:

[0061] Step S1: Divide the sea clutter echo into range and azimuth grids to form K grids, and determine the range center R(k) and azimuth center Az(k) of the kth grid.

[0062] Step S2: Calculate the M-th moment characteristic of the k-th grid. ;

[0063] Step S3: Based on the distance center R(k), azimuth center Az(k), and M-order moment characteristics of the k-th grid. Constructing a sea clutter partition feature vector set ;

[0064] Step S4: Analyze the feature vector set of sea clutter partitions. Perform clustering to obtain an L*K partition matrix U, where L is the desired number of partitions;

[0065] Step S5: Divide the grid corresponding to the row element in the partition matrix U that has the same maximum value as each column into the same partition.

[0066] In statistics, the M-order moment is an important characteristic representing the statistical distribution of data. For radar echoes, the M-order moment characteristic is the main basis for detector design. This application segments the entire radar observation area by range and azimuth, uses the M-order moment statistic to represent the statistical distribution characteristics of sea clutter in a single segmented area, and employs a clustering method to connect and merge the segmented areas, achieving a regionalized representation of sea clutter. This effectively improves the mismatch between the detection algorithm model and the sea clutter model of the entire radar observation area, enhances the accuracy of parameter estimation in the detection algorithm, and the algorithm is simple and convenient, possessing significant engineering application value.

[0067] First, in step S1, the entire radar observation area is divided into two-dimensional grids for distance and angle to construct a locally uniform environment.

[0068] Sea clutter echo characteristics typically vary slowly with distance and azimuth in the airspace. Therefore, partitioning can create a locally homogeneous environment. Assuming the distance gridding interval is... =2km, azimuth gridding interval is =3°, then the full-range echo can be divided into Each grid, such as Figure 2 As shown. (The following is a list of numbers / items / etc.) The distance to the center of each grid is The center of the azimuth is .

[0069] Then, in step S2, the M-order moment features of sea clutter are extracted for each grid.

[0070] In some alternative implementations, step S2 further includes:

[0071] Step S21: Calculate the m-th moment characteristic of the k-th grid:

[0072] ;

[0073] Where N is the number of echo samples, This is the nth echo sample;

[0074] Step S22: Construct the M-order moment characteristics of the sea clutter echo in the k-th grid:

[0075] .

[0076] In some alternative implementations, in step S2, the dimension M of the M-order moment feature takes the value of 3 to 5.

[0077] In this embodiment, for Performing m-fold square root processing brings the different m-order features of sea clutter to a similar order of magnitude, facilitating subsequent processing. M represents the dimension of this feature set, which, according to general empirical distribution models of sea clutter, is typically a two-parameter model. For example, the maximum value for the M-order moment feature set is the third moment, which can characterize the statistical distribution of sea clutter. It should be noted that the value of M should not be too large, as this is important because the data used is from measured data... When estimating, the larger M is, the greater the impact of disturbances on the estimation accuracy.

[0078] Then, in step S3, a smart partition feature vector set for sea clutter is constructed.

[0079] In some alternative implementations, step S3 further includes:

[0080] Step S31: Based on the distance center R(k), azimuth center Az(k), and M-order moment characteristics of the k-th grid. Constructing the sea clutter feature set S:

[0081] ;

[0082] in, , is the set of sea clutter feature vectors for the k-th partition;

[0083] Step S32: Perform weighted processing on the sea clutter feature set S to obtain the sea clutter partition feature vector set. :

[0084] ;

[0085] Where W is the weighted vector and the eigenvector. The dimensions are consistent.

[0086] In this embodiment, in order to facilitate the adjustment of the weight of each grid feature, the sea clutter feature set S is weighted in step S32.

[0087] Next, in step S4, the FCM algorithm is used for clustering to divide the sea clutter into L regions. The FCM algorithm is a commonly used algorithm for cluster analysis, which can be used based on the feature matrix. Generate a segmentation matrix ,in The desired number of partitions. Let be the partition matrix, and be The matrix.

[0088] .

[0089] Finally, in step S5, based on the segmentation matrix Intelligent zoning of sea clutter.

[0090] In some alternative implementations, step S5 further includes:

[0091] Step S51: Determine the maximum value in the k-th column of the partition matrix U;

[0092] Step S52: In row l where the maximum value in column k is located, find one or more elements that are the same as the maximum value;

[0093] Step S53: Determine the column number of the element;

[0094] Step S54: Divide the grid corresponding to one or more column numbers into the same partition.

[0095] In this embodiment, in the segmentation matrix output by the FCM algorithm, the element equal to the maximum value of each column belongs to the same partition. Therefore, in the sea clutter partitioning method, the first step is to find the segmentation matrix. Maximum value for each column Then determine that the l-th partition is the partition matrix. The value in the lth row and Equal grid size. Simulation results based on radar measurement data show that this method can achieve intelligent partitioning of non-uniform sea clutter while ensuring the continuity of the sea clutter airspace, and the statistical distribution characteristics of sea clutter in each region are basically consistent.

[0096] The second aspect of this application provides a non-uniform sea clutter partitioning device based on an M-order moment feature set, corresponding to the above method, mainly comprising:

[0097] The grid parameter determination module is used to divide the sea clutter echo into range and azimuth grids, forming K grids, and determine the range center R(k) and azimuth center Az(k) of the kth grid.

[0098] The M-order moment feature calculation module is used to calculate the M-order moment feature of the k-th grid. ;

[0099] The partition feature vector set construction module is used to construct features based on the distance center R(k), azimuth center Az(k), and M-order moment features of the k-th grid. Constructing a sea clutter partition feature vector set ;

[0100] The clustering module is used to partition the feature vector set of sea clutter. Perform clustering to obtain an L*K partition matrix U, where L is the desired number of partitions;

[0101] The partitioning module is used to divide the grid corresponding to the row elements in the partitioning matrix U that have the same maximum value in each column into the same partition.

[0102] In some optional implementations, the M-order moment feature calculation module includes:

[0103] The moment feature calculation unit is used to calculate the m-th moment feature of the k-th grid.

[0104] ;

[0105] Where N is the number of echo samples, This is the nth echo sample;

[0106] The M-order moment feature calculation unit is used to construct the M-order moment feature of the sea clutter echo in the k-th grid:

[0107] .

[0108] In some alternative implementations, the dimension M of the M-order moment feature takes the value of 3 to 5.

[0109] In some optional implementations, the partition feature vector set construction module includes:

[0110] Sea clutter feature set construction unit, used for the distance center R(k), azimuth center Az(k), and M-order moment features based on the k-th grid. Constructing the sea clutter feature set S:

[0111] ;

[0112] in, ;

[0113] The weighted processing unit is used to perform weighted processing on the sea clutter feature set S to obtain the sea clutter partition feature vector set. :

[0114] ;

[0115] Where W is the weighted vector and the eigenvector. The dimensions are consistent.

[0116] In some alternative implementations, the partitioning module includes:

[0117] The maximum value determination unit is used to determine the maximum value in the k-th column of the partition matrix U;

[0118] The same value lookup unit is used to find one or more elements that are the same as the maximum value in row l where the maximum value in column k is located;

[0119] The column number determination unit is used to determine the column number of the element.

[0120] A grid defining unit is used to divide the grid corresponding to one or more column numbers into the same partition.

[0121] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A non-uniform sea clutter partitioning method based on a set of M-moment features, characterized in that, The method comprises the following steps: Step S1, grid division of the sea clutter echo in range and azimuth is performed to form K grids, and the range center R(k) and the azimuth center Az(k) of the kth grid are determined; Step S2, calculating the Mth moment characteristic of the kth grid ; Step S3, based on the distance center R(k) of the kth grid, the azimuth angle center Az(k), and the M-order moment feature Constructing the sea clutter partition feature vector set ; Step S4, partitioning the sea clutter feature vector set performing clustering to obtain a partition matrix U of L*K, where L is the expected number of partitions; Step S5, the grid corresponding to the column in which the same row elements as the maximum value of each column in the partition matrix U are located is divided into the same partition.

2. The M-moment feature set based non-uniform sea clutter partitioning method of claim 1, wherein, Step S2 further comprises: Step S21, the mth moment feature of the kth grid is calculated; ； where N is the number of echo samples, is the nth echo sample; Step S22, the Mth moment feature of the sea clutter echo of the kth grid is constructed; 。 3. The M-moment feature set based non-uniform sea clutter partitioning method of claim 1, wherein, In step S2, the dimension M of the Mth moment feature is 3-5.

4. The M-moment feature set based non-uniform sea clutter partitioning method of claim 1, wherein, Step S3 further comprises: Step S31, based on the distance center R(k) of the kth grid, the azimuth angle center Az(k) and the M-order moment feature Constructing the sea clutter feature set S: ； wherein ; Step S32, the sea clutter feature set S is weighted to obtain a sea clutter partition feature vector set : ； wherein W is a weight vector, consistent with the dimension of the eigenvector .

5. The M-moment feature set based non-uniform sea clutter partitioning method of claim 1, wherein, Step S5 further comprises: Step S51, the maximum value of the kth column in the partition matrix U is determined; Step S52, one or more elements same as the maximum value are found in the row l in which the maximum value is located; Step S53, the column serial number in which the element is located is determined; Step S54, the grid corresponding to the one or more column serial numbers is divided in the same partition.

6. A non-uniform sea clutter partitioning device based on an M-order moment feature set, characterized in that, The method comprises the following steps: The grid parameter determination module is configured to perform grid division of the sea clutter echo in range and azimuth to form K grids, and determine the range center R(k) and the azimuth center Az(k) of the kth grid; The M-order moment feature calculation module is configured to calculate the M-order moment feature of the kth grid ; a partition feature vector set construction module, configured to construct a partition feature vector set based on the distance center R(k) of the kth grid, the azimuth angle center Az(k), and the M-order moment feature constructing a sea clutter partition feature vector set ; The clustering module is used to partition the feature vector set of sea clutter. Perform clustering to obtain an L*K partition matrix U, where L is the desired number of partitions; The partition module is configured to divide the grid corresponding to the column in which the same row elements as the maximum value of each column in the partition matrix U are located into the same partition.

7. The device for partitioning nonhomogeneous sea clutter based on M-moment feature set according to claim 6, wherein, The Mth moment feature calculation module comprises: The moment feature calculation unit is configured to calculate the mth moment feature of the kth grid; ； where N is the number of echo samples, is the nth echo sample; The Mth moment feature calculation unit is configured to construct the Mth moment feature of the sea clutter echo of the kth grid; 。 8. The device for partitioning nonhomogeneous sea clutter based on M-moment feature set according to claim 6, wherein, The dimension M of the Mth moment feature is 3-5.

9. The device for partitioning nonhomogeneous sea clutter based on M-moment feature set according to claim 6, wherein, The partition feature vector set construction module comprises: a sea clutter feature set construction unit configured to construct a sea clutter feature set S based on the distance center R(k), the azimuth angle center Az(k), and the M-order moment feature of the kth grid construct a sea clutter feature set S: ； wherein ; The weighted processing unit is used to perform weighted processing on the sea clutter feature set S to obtain the sea clutter partition feature vector set. : ； wherein W is a weight vector, consistent with the dimension of the eigenvector .

10. The means for partitioning nonhomogeneous sea clutter based on a set of M-moment features as recited in claim 9, wherein, The partition module comprises: The maximum value determination unit is configured to determine the maximum value of the kth column in the partition matrix U; The same value finding unit is configured to find one or more elements same as the maximum value in the row l in which the maximum value is located; The column serial number determination unit is configured to determine the column serial number in which the element is located; The grid determination unit is configured to divide the grid corresponding to the one or more column serial numbers into the same partition.

Images