A sea clutter data generation method based on improved SIRP

Abstract

The application relates to a sea clutter data generation method based on improved SIRP. First, an amplitude distribution characteristic function, a time correlation function and a space correlation function are extracted from actual sea clutter data. Second, speckle components satisfying the time correlation and space correlation characteristics are generated through convolution filtering. Third, texture components conforming to inverse Gaussian distribution are obtained through rejection sampling. Finally, the sea clutter data are generated by using the speckle components and the texture components. The purpose of the application is to generate sea clutter simulation data satisfying the amplitude distribution characteristics, the time correlation characteristics and the space correlation characteristics, so as to reduce the dependence on actual sea clutter data and improve the fitting degree of the simulation data and the true value.

Description

Technical Field

[0001] This invention relates to the field of radar digital signal processing technology, specifically a method for generating sea clutter data based on an improved SIRP. Background Technology

[0002] The development and application of sea surface target detection radars require in-depth research on sea clutter data. Methods for acquiring sea clutter data include both measured data acquisition and simulation generation. Acquiring measured sea clutter data is costly, and the repeatability of field data acquisition is poor, resulting in a limited number of measured sea clutter samples. However, evaluating radar performance and verifying the effectiveness of anti-clutter algorithms in indoor tests require a large amount of sea clutter data. Given these issues, methods for reproducing sea clutter data through simulation are essential. Realistic sea clutter data simulation technology can not only reproduce the true characteristics of sea clutter but also provide a realistic testing environment for radar performance evaluation, making the evaluation results more reliable.

[0003] Currently, sea clutter modeling methods mainly include two aspects: First, the study of the physical mechanism layer, that is, under certain radar operating parameters and external environmental parameters (such as waveband, polarization, sea state, and grazing angle), the electromagnetic scattering theory is used to calculate the backscattering coefficient of the sea surface, model its undulation, and obtain the radar echo of the simulated sea surface, such as patent CN202410223240.1 "A Simulation Method and Device for Spatial Correlated Sea Clutter Based on Sea Surface Structure" and patent CN202311604731.2 "A Dynamic Simulation Method for Sea Clutter Matching the Spatiotemporal Nature of the Real Marine Environment"; Second, establishing statistical models based on measured sea clutter data. The physical mechanism-based modeling approach involves conducting sea clutter measurement experiments on actual sea surfaces or wave pools. The measurement data is used to analyze the characteristics of sea clutter and their interrelationships with various factors such as radar operating parameters and marine environmental parameters. A statistical model of sea clutter is established, with or without considering the underlying mechanisms. Examples include patents CN202410619861.1 "Method and Device for Sea Clutter Amplitude Modulation and Simulation Based on Multi-Point Signal Model," CN202211256995.9 "A Method and Device for Simulating Clutter Data Based on Radar Parameter Information," and CN202210831350.7 "A High-Resolution Sea Clutter Modeling and Simulation Method." Because physical mechanism-based modeling often fails to accurately reflect the actual situation due to the complex composition of the sea surface and numerous influencing factors, statistical modeling methods that can describe the actual sea clutter characteristics to the greatest extent are considered a more effective approach.

[0004] Sea clutter statistical models are primarily described through amplitude distribution and correlation characteristics. Existing technologies have evolved through various distribution models, including normal, Rayleigh, log-normal, Weibull, composite Gaussian, and mixed distribution models. Sea clutter correlation is categorized into temporal and spatial correlation. Currently, sea clutter modeling methods mainly include memoryless nonlinear (ZMNL) transforms and spherically invariant random processes (SIRP) methods, such as patents CN202210831350.7 ("A High-Resolution Sea Clutter Modeling and Simulation Method"), CN202110668819.5 ("A Sea Clutter Parameter Estimation Method, System, Device, and Storage Medium"), and CN201810177009.8 ("A Radar Baseband Clutter Generation Device and Method"). These methods can generate sea clutter models that conform to the corresponding amplitude distribution and exhibit temporal correlation; however, few sea clutter models currently possess both spatial and temporal correlation. Furthermore, existing methods are unable to generate sea clutter simulation data in both the I and Q channels (radar echoes have two channels, one with in-phase component "I" and the other with quadrature component "Q") that satisfy both temporal and spatial correlation, resulting in a low degree of matching between simulation data and measured data. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of the existing technologies by proposing a sea clutter data generation method based on an improved SIRP. The aim is to generate simulated sea clutter data that simultaneously satisfies the amplitude distribution characteristics, time correlation characteristics, and spatial correlation characteristics of actual sea clutter and has a complex signal form.

[0006] The technical solution adopted by this invention to achieve the above-mentioned technical objectives is as follows: a method for generating sea clutter data based on improved SIRP, comprising the following steps:

[0007] 1) Based on measured sea clutter data, the amplitude distribution characteristic function, temporal correlation function, and spatial correlation function describing sea clutter are extracted from them respectively;

[0008] 2) Based on the temporal and spatial correlation functions obtained in step 1), a convolution kernel is constructed. Then, the complex Gaussian white noise is processed by a convolution filter to generate speckle components that simultaneously satisfy the temporal and spatial correlation characteristics. ;

[0009] 3) Set two sets of Gaussian white noise data and use a low-pass filter to adjust them to generate two sets of highly correlated colored Gaussian noise. Then combine them to form a noise data sample set that conforms to the Rayleigh distribution. Finally, use the rejection sampling method to process the data to obtain the texture component that follows an inverse Gaussian distribution. ;

[0010] 4) Use the speckle component obtained in step 2). and the texture components obtained in step 3) Sea clutter data is generated using the following formula:

[0011]

[0012] In the formula, z represents the clutter component; The speckle component represents the Rayleigh distribution; Represents texture components that follow a specific distribution.

[0013] As an optimization scheme of the above-mentioned sea clutter data generation method based on improved SIRP, the specific operation of extracting the amplitude distribution characteristic function describing sea clutter in step 1) is as follows: the amplitude distribution of sea clutter data is fitted with the existing CG-IG distribution, and the values ​​of shape parameter v and scale parameter b are calculated. Then, the shape parameter v and scale parameter b are substituted into the CG-IG distribution to obtain the amplitude distribution characteristic function describing sea clutter.

[0014] As another optimization scheme for the above-mentioned sea clutter data generation method based on improved SIRP, the probability density function of the CG-IG distribution... The expression is:

[0015]

[0016] In the formula, e is the base of the natural logarithm; exp is the natural exponential function; Indicates the amplitude of sea clutter; Represents the shape parameters of the basic scatterer; The scale parameter represents the basic scatterer.

[0017] As another optimization scheme for the above-mentioned sea clutter data generation method based on improved SIRP, the specific operation of extracting the time correlation function describing sea clutter in step 1) is as follows:

[0018] For a certain distance unit The expression for the time correlation function of echo intensity between different pulses is:

[0019]

[0020] In the formula, m is the number of interval pulses, M represents the number of pulses in the sample, and ii represents the ii-th pulse. Distance unit The intensity in the echo of the iith pulse;

[0021] Then, the time correlation functions obtained from all distance units are averaged to obtain the averaged time correlation function. , This represents the time correlation function of sea clutter.

[0022] As another optimization scheme for the above-mentioned sea clutter data generation method based on improved SIRP, the specific operation of extracting the spatial correlation function describing sea clutter in step 1) is as follows:

[0023] For a certain pulse The spatial correlation function expression for the echo intensity of cells at different distances is:

[0024]

[0025] In the formula, n is the number of interval distance units; N represents the number of distance units in the sample; jj represents the jj-th distance unit; Indicates a pulse The intensity at the jj-th distance unit;

[0026] Then, the spatial correlation functions obtained from all pulses are averaged to obtain the averaged spatial correlation function. , This represents the spatial correlation function of sea clutter.

[0027] As another optimization scheme for the above-mentioned sea clutter data generation method based on improved SIRP, step 2) generates speckle components that simultaneously satisfy both temporal and spatial correlation characteristics. The specific operation is as follows:

[0028] 2.1) A Gaussian distribution is used as the starting sample for generating correlated sample clutter.

[0029] set up and It is an M×N order Gaussian white noise matrix. and Composition of complex Gaussian white noise matrix:

[0030]

[0031] In the formula, j represents the complex unit. Indicates by and Forming a complex Gaussian white noise matrix;

[0032] 2.2) Based on the spatial correlation function The convolution kernel for the corresponding two-dimensional convolution is generated based on the time correlation function:

[0033]

[0034] Where m and n represent the number of data points extracted from the time correlation function and the spatial correlation function, respectively. The spatial correlation function of sea clutter is represented. represents the time correlation function of sea clutter; K represents the convolution kernel, which is also a complex matrix;

[0035] 2.3) For complex Gaussian white noise matrix Two-dimensional convolution is performed to generate speckle components that simultaneously satisfy temporal and spatial correlation characteristics. The calculation formula is as follows:

[0036]

[0037] In the formula, Indicates the position of the output speckle component matrix. The value; It is the input complex Gaussian white noise matrix relative to Position offset The value; Is the convolution kernel at position The value of .

[0038] As another optimization scheme for the above-mentioned sea clutter data generation method based on improved SIRP, the texture components obtained in step 3) follow an inverse Gaussian distribution. The specific operation is as follows:

[0039] 3.1) Set up two sets of Gaussian white noise data and Two sets of highly correlated colored Gaussian noise data were generated by controlling the noise through low-pass filters. and ;

[0040] The mathematical expressions for the filtering process of the two sets of Gaussian white noise data are as follows:

[0041]

[0042] In the formula, h is the convolution kernel of the low-pass filter. ; Represents the first of two sets of Gaussian white noise data. There are 1, 2, 3 elements; k represents the k-th element in the convolution kernel h.

[0043] 3.2) From colored Gaussian noise data and Given data x that follows a Rayleigh distribution, the set of all possible values ​​of x constitutes the sample space S, i.e.:

[0044]

[0045] In the formula, b is the scale parameter in the amplitude distribution characteristic function of sea clutter. Denotes the first random number in a Rayleigh distribution. Each element, because Two sets of colored noise data and The first Composed of elements, the two are equivalent. Represents all positive integers;

[0046] 3.3) Sample in the sample space S, and generate texture components that characterize the inverse Gaussian distribution of the shape parameter v and scale parameter b of the sea clutter using the rejection sampling method. .

[0047] As another optimization scheme for the above-mentioned sea clutter data generation method based on improved SIRP, step 3.1) generates two sets of highly correlated colored Gaussian noise data. and , refers to and The correlation coefficient should be as close to 1 as possible.

[0048] As another optimization scheme for the above-mentioned sea clutter data generation method based on improved SIRP, in step 3.3), the rejection sampling method generates texture components that characterize the inverse Gaussian distribution of the sea clutter shape parameter v and scale parameter b. The pseudocode is:

[0049] ;

[0050] Where e represents one-dimensional texture component data, totaling M×N; This represents the texture component data of M×N; This command converts data e into an M×N matrix; W is a constant; u is a random number sampled from a uniform distribution [0,1]; q(x) is the distribution of the enclosing distribution that is consistent with the distribution characteristics of the sample space S; p(x) is the target distribution, i.e., the inverse Gaussian distribution of the shape parameter v and the scale parameter b of the sea clutter.

[0051] As another optimization scheme for the aforementioned sea clutter data generation method based on improved SIRP, to ensure that the enclosing distribution q(x) is sufficiently close to the target distribution p(x), thereby minimizing the number of rejected samples, and satisfying that for all x of the target distribution... , where W takes the value 2.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] 1) Actual radar echoes have two channel components, I and Q, which are usually represented by complex numbers. In order to generate signals that conform to the radar mechanism, this invention uses complex signal form to describe speckle components and convolution filters in the underlying signal processing, and finally generates sea clutter simulation data with relevant features that conform to the actual I and Q channel signals, thereby greatly improving the fit between the generated simulation data and the real values.

[0054] 2) This invention, while ensuring that the generated sea clutter meets the amplitude distribution characteristics, introduces temporal and spatial correlation features into the sea clutter data by constructing a convolution kernel for two-dimensional convolution filtering, thereby forming a sea clutter modeling method that can accurately reproduce the characteristics of actual sea clutter. Compared with existing sea clutter data generation methods, it achieves a higher degree of agreement with measured sea clutter.

[0055] 3) This invention uses rejection sampling to generate the texture components of sea clutter, offering the advantage of flexible control over the distribution of these texture components. Furthermore, by using rejection sampling to generate texture components following an inverse Gaussian distribution from highly correlated Rayleigh distribution data, the resulting texture components are also highly correlated, which does not affect the correlation characteristics of the sea clutter data. Therefore, the amplitude distribution characteristics of the sea clutter data are controlled by the texture components, while the correlation characteristics are controlled by the speckle components; there is no coupling between the amplitude distribution characteristics and the correlation characteristics of the sea clutter data. Thus, compared to other methods, this invention can more accurately generate sea clutter data with predetermined amplitude distribution and correlation characteristics.

[0056] 4) This invention is also applicable to sea clutter distribution models with composite K-distribution, generalized Pareto distribution, and other composite Gaussian distributions, and does not require nonlinear calculations, thus having wide applicability. Attached Figure Description

[0057] Figure 1 This is a flowchart illustrating the implementation process of the present invention;

[0058] Figure 2 For example verification, a comparison diagram is shown between the amplitude distribution of sea clutter data obtained according to the method of the present invention and the amplitude distribution of measured sea clutter sample data;

[0059] Figure 3 For example verification, a comparison graph is shown between the time correlation function of sea clutter data obtained according to the method of the present invention and the time correlation function of measured sea clutter sample data;

[0060] Figure 4 For example verification, a comparison graph is shown between the spatial correlation function of sea clutter data obtained according to the method of the present invention and the spatial correlation function of measured sea clutter sample data. Detailed Implementation

[0061] The technical solution of the present invention will be further described in detail below with reference to specific embodiments. Parts not explained in the following embodiments of the present invention are all considered to be prior art known or should be known by those skilled in the art.

[0062] Example 1

[0063] A method for generating sea clutter data based on an improved SIRP, such as Figure 1 As shown, it includes the following steps:

[0064] 1) Based on measured sea clutter data, the amplitude distribution characteristic function, temporal correlation function, and spatial correlation function describing sea clutter are extracted from them respectively;

[0065] The specific operation for extracting the amplitude distribution feature function is as follows: the amplitude distribution of the sea clutter data is fitted with the existing CG-IG distribution, and the values ​​of the shape parameter v and the scale parameter b are calculated. Then, the shape parameter v and the scale parameter b are substituted into the CG-IG distribution to obtain the amplitude distribution feature function describing the sea clutter.

[0066] The specific steps for extracting the time correlation function describing sea clutter are as follows:

[0067] For a certain distance unit The expression for the time correlation function of echo intensity between different pulses is:

[0068]

[0069] In the formula, m is the number of interval pulses, M represents the number of pulses in the sample, and ii represents the ii-th pulse. Distance unit The intensity in the echo of the iith pulse;

[0070] Then, the time correlation functions obtained from all distance units are averaged to obtain the averaged time correlation function. , This represents the time correlation function of sea clutter;

[0071] The specific steps for extracting the spatial correlation function describing sea clutter are as follows:

[0072] For a certain pulse The spatial correlation function expression for the echo intensity of cells at different distances is:

[0073]

[0074] In the formula, n is the number of interval distance units; N represents the number of distance units in the sample; jj represents the jj-th distance unit; Indicates a pulse The intensity at the jj-th distance unit;

[0075] Then, the spatial correlation functions obtained from all pulses are averaged to obtain the averaged spatial correlation function. , This represents the spatial correlation function of sea clutter;

[0076] Based on the temporal and spatial correlation functions obtained in step 1), a convolution kernel is constructed. Then, a convolution filter is used to process the complex Gaussian white noise, generating speckle components that simultaneously satisfy the temporal and spatial correlation characteristics. The specific operation is as follows:

[0077] 2.1) A Gaussian distribution is used as the starting sample for generating correlated sample clutter.

[0078] set up and It is an M×N order Gaussian white noise matrix. and Composition of complex Gaussian white noise matrix:

[0079]

[0080] In the formula, j represents the complex unit. Indicates by and Forming a complex Gaussian white noise matrix;

[0081] 2.2) Based on the spatial correlation function The convolution kernel for the corresponding two-dimensional convolution is generated based on the time correlation function:

[0082]

[0083] Where m and n represent the number of data points extracted from the time correlation function and the spatial correlation function, respectively. The spatial correlation function of sea clutter is represented. represents the time correlation function of sea clutter; K represents the convolution kernel, which is also a complex matrix;

[0084] 2.3) For complex Gaussian white noise matrix Two-dimensional convolution is performed to generate speckle components that simultaneously satisfy temporal and spatial correlation characteristics. The calculation formula is as follows:

[0085]

[0086] In the formula, Indicates the position of the output speckle component matrix. The value; It is the input complex Gaussian white noise matrix relative to Position offset The value; Is the convolution kernel at position The value;

[0087] Two sets of Gaussian white noise data are set and manipulated using a low-pass filter to generate two sets of highly correlated colored Gaussian noise. These are then combined to form a noise data sample set that conforms to a Rayleigh distribution. Finally, a rejection sampling method is used to process the data to obtain texture components that follow an inverse Gaussian distribution. The specific operation is as follows:

[0088] 3.1) Set up two sets of Gaussian white noise data and Two sets of highly correlated colored Gaussian noise data were generated by controlling the noise through low-pass filters. and ;

[0089] The mathematical expressions for the filtering process of the two sets of Gaussian white noise data are as follows:

[0090]

[0091] In the formula, h is the convolution kernel of the low-pass filter. ; Represents the first of two sets of Gaussian white noise data. There are 1, 2, 3 elements; k represents the k-th element in the convolution kernel h.

[0092] In this step, two sets of highly correlated colored Gaussian noise data are generated. and , refers to and The correlation coefficient should be as close to 1 as possible;

[0093] 3.2) From colored Gaussian noise data and Given data x that follows a Rayleigh distribution, the set of all possible values ​​of x constitutes the sample space S, i.e.:

[0094]

[0095] In the formula, b is the scale parameter in the amplitude distribution characteristic function of sea clutter. Denotes the first random number in a Rayleigh distribution. Each element, because Two sets of colored noise data and The first Composed of elements, the two are equivalent. Represents all positive integers;

[0096] 3.3) Sample in the sample space S, and generate texture components that characterize the inverse Gaussian distribution of the shape parameter v and scale parameter b of the sea clutter using the rejection sampling method. ;

[0097] 4) Use the speckle component obtained in step 2). and the texture components obtained in step 3) Sea clutter data is generated using the following formula:

[0098]

[0099] In the formula, z represents the clutter component; The speckle component represents the Rayleigh distribution; Represents texture components that follow a specific distribution.

[0100] Example 2

[0101] This embodiment is a further improvement on Embodiment 1. Its main body is the same as Embodiment 1, but the improvement lies in the probability density function of the CG-IG distribution in step 1). The expression is:

[0102]

[0103] In the formula, e is the base of the natural logarithm; exp is the natural exponential function; Indicates the amplitude of sea clutter; Represents the shape parameters of the basic scatterer; The scale parameter represents the basic scatterer.

[0104] Example 3

[0105] This embodiment is a further improvement on Embodiment 1. While the main body is the same as Embodiment 1, the improvement lies in step 3.3), where the rejection sampling method generates texture components representing the inverse Gaussian distribution of the shape parameter v and scale parameter b of the sea clutter. The pseudocode is:

[0106] ;

[0107] Where e represents one-dimensional texture component data, totaling M×N; This represents the texture component data of M×N; This command converts data e into an M×N matrix; W is a constant; u is a random number sampled from a uniform distribution [0,1]; q(x) is the distribution of the enclosing distribution that is consistent with the distribution characteristics of the sample space S; p(x) is the target distribution, i.e., the inverse Gaussian distribution of the shape parameter v and the scale parameter b of the sea clutter.

[0108] To ensure that the enclosing distribution q(x) is sufficiently close to the target distribution p(x), thereby minimizing the number of rejected samples, and satisfying that for all x of the target distribution... , where W takes the value 2.

[0109] To verify the effectiveness of the present invention, the following examples were conducted:

[0110] The data used were echo data of LFM transmitted signals from the data file "20210106155330_01_staring" in the first issue of 2020 of the "Radar Observation Data Sharing Program (SDRDSP)". A 500×500 matrix was extracted as a sample (the matrix ranges from 2001 to 2500 in the time dimension and from 2001 to 2500 in the distance dimension, and the number of data extracted in both the time correlation function and the spatial correlation function is 40 (m=40, n=40).

[0111] According to the method of this invention, sea clutter data is generated based on the characteristics of the sample data, and the generated sea clutter data is compared with the measured data. The results are shown in the appendix. Figure 2-4 As shown:

[0112] Figure 2 To obtain a comparison chart between the amplitude distribution of sea clutter data and the amplitude distribution of measured sea clutter sample data according to the method of this invention, from... Figure 2 As can be seen, the amplitude distribution of the sea clutter data generated by the method of the present invention is almost consistent with that of the measured sea clutter data, and it also confirms that the CG-IG distribution model can fit the sea clutter data well. Therefore, the present invention can generate sea clutter data that meets the set amplitude distribution.

[0113] Figure 3 To obtain a comparison chart between the time correlation function of sea clutter data and the time correlation function of measured sea clutter sample data according to the method of this invention, from... Figure 3 As can be seen from this, the time correlation function of the sea clutter data generated by this invention is almost identical to that of the measured data in both the real and imaginary parts. Therefore, this invention can generate sea clutter data that satisfies the set time correlation function.

[0114] Figure 4To obtain a comparison chart between the spatial correlation function of sea clutter data and the spatial correlation function of measured sea clutter sample data according to the method of this invention, from... Figure 4 As can be seen from this, the spatial correlation function of the sea clutter data generated by this invention has a high similarity in the real and imaginary parts compared with the spatial correlation function of the measured data. Therefore, this invention can generate sea clutter data that satisfies the set spatial correlation function.

[0115] In summary, the method of the present invention can generate sea clutter simulation data that simultaneously satisfies the actual sea clutter amplitude distribution characteristics, time correlation characteristics, and spatial correlation characteristics.

Claims

1. A method for generating sea clutter data based on improved SIRP, characterized in that, Includes the following steps: 1) Based on measured sea clutter data, the amplitude distribution characteristic function, temporal correlation function, and spatial correlation function describing sea clutter are extracted from them respectively; 2) Based on the temporal and spatial correlation functions obtained in step 1), a convolution kernel is constructed. Then, the complex Gaussian white noise is processed by a convolution filter to generate speckle components that simultaneously satisfy the temporal and spatial correlation characteristics. ; 3) Set two sets of Gaussian white noise data and use a low-pass filter to adjust them to generate two sets of highly correlated colored Gaussian noise. Then combine them to form a noise data sample set that conforms to the Rayleigh distribution. Finally, use the rejection sampling method to process the data to obtain the texture component that follows an inverse Gaussian distribution. ; 4) Use the speckle component obtained in step 2). and the texture components obtained in step 3) Sea clutter data is generated using the following formula: ； In the formula, z represents the clutter component; The speckle component represents the Rayleigh distribution; Represents texture components that follow a specific distribution.

2. The sea clutter data generation method based on improved SIRP according to claim 1, characterized in that, The specific operation for extracting the amplitude distribution characteristic function describing sea clutter in step 1) is as follows: the amplitude distribution of sea clutter data is fitted with the existing CG-IG distribution, and the values ​​of shape parameter v and scale parameter b are calculated. Then, the shape parameter v and scale parameter b are substituted into the CG-IG distribution to obtain the amplitude distribution characteristic function describing sea clutter.

3. The sea clutter data generation method based on improved SIRP according to claim 2, characterized in that, The probability density function of the CG-IG distribution The expression is: ； In the formula, e is the base of the natural logarithm; exp is the natural exponential function; Indicates the amplitude of sea clutter; Represents the shape parameters of the basic scatterer; The scale parameter represents the basic scatterer.

4. The sea clutter data generation method based on improved SIRP according to claim 1, characterized in that, The specific steps for extracting the time correlation function describing sea clutter in step 1) are as follows: For a certain distance unit The expression for the time correlation function of echo intensity between different pulses is: ； In the formula, m is the number of interval pulses, M represents the number of pulses in the sample, and ii represents the ii-th pulse. Distance unit The intensity in the echo of the iith pulse; Then, the time correlation functions obtained from all distance units are averaged to obtain the averaged time correlation function. , This represents the time correlation function of sea clutter.

5. A method for generating sea clutter data based on improved SIRP according to claim 1, characterized in that, The specific steps for extracting the spatial correlation function describing sea clutter in step 1) are as follows: For a certain pulse The spatial correlation function expression for the echo intensity of cells at different distances is: ； In the formula, n is the number of interval distance units; N represents the number of distance units in the sample; jj represents the jj-th distance unit; Indicates a pulse The intensity at the jj-th distance unit; Then, the spatial correlation functions obtained from all pulses are averaged to obtain the averaged spatial correlation function. , This represents the spatial correlation function of sea clutter.

6. The sea clutter data generation method based on improved SIRP according to claim 1, characterized in that, In step 2), speckle components that simultaneously satisfy both temporal and spatial correlation characteristics are generated. The specific operation is as follows: 2.1) A Gaussian distribution is used as the starting sample for generating correlated sample clutter. set up and It is an M×N order Gaussian white noise matrix. and Composition of complex Gaussian white noise matrix: ； In the formula, j represents the complex unit. Indicates by and Forming a complex Gaussian white noise matrix; 2.2) Based on the spatial correlation function Time correlation function Generate the corresponding 2D convolution kernel: ； Where m and n represent the number of data points extracted from the time correlation function and the spatial correlation function, respectively. The spatial correlation function of sea clutter is represented. represents the time correlation function of sea clutter; K represents the convolution kernel, which is also a complex matrix; 2.3) For complex Gaussian white noise matrix Two-dimensional convolution is performed to generate speckle components that simultaneously satisfy temporal and spatial correlation characteristics. The calculation formula is as follows: ； In the formula, Indicates the position of the output speckle component matrix. The value; It is the input complex Gaussian white noise matrix relative to Position offset The value; Is the convolution kernel at position The value of .

7. A method for generating sea clutter data based on improved SIRP according to claim 1, characterized in that, In step 3), texture components that follow an inverse Gaussian distribution are obtained. The specific operation is as follows: 3.1) Set up two sets of Gaussian white noise data and Two sets of highly correlated colored Gaussian noise data were generated by controlling the noise through low-pass filters. and ; The mathematical expressions for the filtering process of the two sets of Gaussian white noise data are as follows: ； In the formula, h is the convolution kernel of the low-pass filter. ; Represents the first of two sets of Gaussian white noise data. There are 1, 2, 3 elements; k represents the k-th element in the convolution kernel h. 3.2) From colored Gaussian noise data and Given data x that follows a Rayleigh distribution, the set of all possible values ​​of x constitutes the sample space S, i.e.: ； In the formula, b is the scale parameter in the amplitude distribution characteristic function of sea clutter. Denotes the first random number in a Rayleigh distribution. Each element, because Two sets of colored noise data and The obtained number Composed of elements, the two are equivalent. Represents all positive integers; 3.3) Sample in the sample space S, and generate texture components that characterize the inverse Gaussian distribution of the shape parameter v and scale parameter b of the sea clutter using the rejection sampling method. .

8. A method for generating sea clutter data based on improved SIRP according to claim 7, characterized in that, In step 3.1), two sets of highly correlated colored Gaussian noise data are generated. and , refers to and The correlation coefficient should be as close to 1 as possible.

Images