Sea clutter filtering method, equipment, storage medium and product

Abstract

The present invention discloses a method, device, storage medium and product for filtering sea clutter. The method comprises calculating the specific value of each characteristic quantity of a first sample data set according to a first sampling point, and calculating the specific value of each characteristic quantity of the first sample data set according to a second sampling point; generating a logarithmic data histogram of the ratio of the sea clutter conditional probability of the characteristic quantity to the precipitation echo conditional probability according to the specific values ​​of the characteristic quantities of the first sampling point and the second sampling point; calculating the specific value of each characteristic quantity according to the actual sampling point; determining the actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each characteristic quantity of the actual sampling point according to the specific value of each characteristic quantity of the actual sampling point and the logarithmic data histogram; filtering sea clutter according to the actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of all characteristic quantities of the actual sampling point. The present invention effectively filters out sea clutter in radar echoes.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and in particular relates to a method, device, storage medium and product for filtering sea clutter based on volume scanning data of a dual-polarization phased array weather radar. Background Art

[0002] At present, the methods for processing sea clutter include but are not limited to the following methods:

[0003] (1) Signal processing method

[0004] This clutter detection method is performed in the radar's signal processor, i.e., before the actual radar (image) product is generated. For Doppler radars, filters that use Doppler velocity and spectrum width are very effective. Targets with zero or very low velocity are likely to be stationary ground targets (i.e., ground clutter), while precipitation typically has non-zero velocity; however, Doppler clutter filters are not very effective in detecting sea clutter echoes originating from the ocean surface, which have non-zero velocity due to the motion of the waves.

[0005] (2) Pattern classification method

[0006] Modification of the radar signal processor is an advanced task usually applicable only to non-operational research radars, whereas most sea clutter detection methods are developed and applied in operational environments where many users depend on uninterrupted data transmission. These methods are based on radar data products after the signal processor, using radar quantities, CAPPI products (i.e., Contour Plane Position Indicator), combinations of CAPPI or similar products, to extract various features from these data that allow distinguishing between clutter and precipitation.

[0007] Pattern classification methods include clutter map methods, feature quantity spatial decorrelation methods, and data fusion methods. One of the simplest methods for dealing with clutter is to generate a clutter map. By observing clutter in the absence of precipitation, integrating the observations over a period of time, and generating an average clutter map, the average clutter map is then used to effectively remove ground clutter. However, when precipitation echoes are mixed with clutter, there is a risk of eliminating precipitation echoes, and the clutter map method is only effective in detecting ground clutter. The sea clutter caused by abnormal propagation varies greatly in time and space, and it is impossible to generate an average map of these clutters. Therefore, the clutter map method is not suitable for the processing of sea clutter. The feature quantity spatial decorrelation method uses the difference in the spatial decorrelation time of clutter and precipitation signals. The decorrelation of clutter is faster than the decorrelation of precipitation, which is very effective for eliminating land clutter, but less effective for ocean clutter. There is also the use of 3D texture as a metric feature for clutter / precipitation discrimination. The reason is that abnormal propagation (AP) clutter only affects lower radar elevation angles, so the geometric characteristics of the vertical range, gradient, and variability of radar echoes are useful.

[0008] Data fusion methods include satellite data fusion methods and propagation modeling prediction methods. The satellite data fusion method uses third-party data, such as multispectral satellite observations in the visible spectrum, infrared spectrum, and thermal infrared spectrum, to map precipitation areas; several methods that use geostationary satellite observations have shown promising results. The propagation modeling prediction method uses information from radiosondes or numerical weather prediction (NWP) models to predict when abnormal propagation will occur; the refractive index of the atmosphere can be calculated based on parameters such as temperature, air pressure, and humidity, thereby determining the electromagnetic propagation path. Observations from sondes (weather balloons that measure vertical atmospheric profiles of variables such as temperature, pressure, and humidity) or predictions of meteorological parameters output by NWP models are usually used.

[0009] (3) New radar technology

[0010] Recent advances in radar technology, especially dual-polarization radar, provide new methods for clutter detection. Operating the radar in multiple polarization modes yields important information about the radar target that can be used for clutter identification. The correlation coefficient of horizontal and vertical power and the differential phase shift are good characteristic quantities.

[0011] (4) Statistical methods

[0012] Clutter detection techniques employ various methods to infer class membership of echoes in radar data. Most methods use supervised classification techniques ranging from simple thresholds on input features to classification tree methods, neural networks, fuzzy logic classifiers, and naive Bayes classifiers.

[0013] Sea wave echoes often appear in radar echoes in coastal areas. This type of non-meteorological echo is mainly caused by super-refraction, which affects the quality of radar data to a certain extent and interferes with precipitation echoes. Sea wave echoes mainly appear in the sea surface area, composed of many needle-shaped echo bodies, fan-shaped outward radiating, with a relatively low echo height. They generally appear in the echo at the lowest radar detection elevation angle, and rarely appear in the second detection elevation angle. The radial velocity of sea wave echoes is relatively small, mainly concentrated in -10 to 5 m / s; compared with precipitation echoes, sea wave echoes have greater differences in echo morphology, vertical structure, radial velocity, etc., and have obvious double-bias clutter characteristics.

[0014] At present, the clutter map method cannot solve the problem of large variations in the temporal and spatial distribution range of sea clutter; the feature quantity spatial decorrelation algorithm is very effective for land clutter, but less effective for ocean clutter; among the data fusion methods, the satellite data fusion method not only needs to rely on satellite data in addition to the radar-based data itself, but also the temporal resolution of satellite data is usually at the sub-hour level, which is far lower than the sub-minute time accuracy of dual-polarization radar, making it difficult to assimilate; the same problem also exists in the sounding instrument or model data required for the propagation modeling prediction method. Summary of the invention

[0015] The object of the present invention is to provide a sea clutter filtering method, device, storage medium and product to solve the problem that traditional methods cannot filter sea clutter or the filtering effect is poor.

[0016] The present invention solves the above technical problems through the following technical solutions: A method for filtering sea clutter, comprising:

[0017] Constructing a first sample data set; wherein the first sample data set includes each valid sampling point and its label in each radar volume scan data, the label includes sea clutter and precipitation echo, defining the valid sampling point labeled as sea clutter as the first sampling point, and the valid sampling point labeled as precipitation echo as the second sampling point;

[0018] Calculating a specific value of each characteristic quantity of the first sampling point according to the first sampling point of the first sample data set, and calculating a specific value of each characteristic quantity of the second sampling point according to the second sampling point of the first sample data set; wherein the characteristic quantity is a parameter capable of distinguishing sea clutter from precipitation echo;

[0019] Generate a logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the corresponding feature quantity according to the specific value of each feature quantity at the first sampling point and the specific value of each feature quantity at the second sampling point;

[0020] Acquire test sample volume scan data, and calculate specific values ​​of each feature quantity of the actual sampling point according to the actual sampling point in the test sample volume scan data;

[0021] Determining actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity at the actual sampling point according to the specific value of each feature quantity at the actual sampling point and the logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity;

[0022] Sea clutter is filtered out according to actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of all characteristic quantities of the actual sampling points.

[0023] Further, the characteristic quantities include the standard deviation of the differential phase shift, the mean value of the correlation coefficient and the probability of precipitation echo of the echo block;

[0024] The specific value of the standard deviation of the differential phase shift is calculated as follows:

[0025] ;

[0026] Among them, σ(Φ dp ) represents a specific value of the standard deviation of the differential phase shift of the first sampling point or a specific value of the standard deviation of the differential phase shift of the second sampling point, n represents the window size of the first sampling point or the second sampling point along the azimuth and radial distance bin, represents the differential phase shift of the jth valid sampling point in the window of the first sampling point or the second sampling point, The mean of the differential phase shifts of all valid sampling points within the window representing the first sampling point or the second sampling point;

[0027] The calculation formula for the specific value of the correlation coefficient mean is:

[0028] ;

[0029] in, represents the specific value of the mean value of the correlation coefficient of the first sampling point or the specific value of the mean value of the correlation coefficient of the second sampling point; CC j represents the correlation coefficient of the jth valid sampling point in the window of the first sampling point or the second sampling point;

[0030] The calculation formula for the specific value of the probability of precipitation echo in the echo block is:

[0031] R pre =A P / A va ;

[0032] Among them, R pre Indicates the specific value of the probability of precipitation echo in the echo block, A P A represents the area of ​​precipitation echo in the echo block in the sea area. va Indicates the echo block area in the sea surface area.

[0033] Furthermore, when calculating the specific value of the standard deviation of the differential phase shift, the differential phase shift of the effective sampling point is also unfolded, and the specific process includes:

[0034] Copying the differential phase shifts of n valid sampling points within the window of the first sampling point or the second sampling point to obtain a first differential phase shift array;

[0035] Adding the negative value in the first differential phase shift array to the radar folding period to obtain a second differential phase shift array;

[0036] Calculating a first specific value of the standard deviation of the differential phase shift of the first sampling point or the second sampling point according to the first differential phase shift array; calculating a second specific value of the standard deviation of the differential phase shift of the first sampling point or the second sampling point according to the second differential phase shift array;

[0037] The smaller one of the first specific value and the second specific value is selected as the specific value of the standard deviation of the differential phase shift of the current first sampling point or the current second sampling point.

[0038] Further, a logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the corresponding feature quantity is generated according to the specific value of each feature quantity of the first sampling point and the specific value of each feature quantity of the second sampling point, including:

[0039] Determine a value range of the corresponding feature quantity according to the specific value of each feature quantity of each first sampling point or the second sampling point, and divide the value range into N intervals;

[0040] According to the specific value of each characteristic quantity of each first sampling point, the conditional probability of sea clutter of the corresponding characteristic quantity in the i-th interval of the value range is calculated; according to the specific value of each characteristic quantity of each second sampling point, the conditional probability of precipitation echo of the corresponding characteristic quantity in the i-th interval of the value range is calculated;

[0041] According to the sea clutter conditional probability and precipitation echo conditional probability of each characteristic quantity in the i-th interval, the logarithmic data of the ratio of the sea clutter conditional probability and the precipitation echo conditional probability of the corresponding characteristic quantity in the i-th interval is calculated;

[0042] A logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the corresponding feature quantity is generated according to the logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity in N intervals.

[0043] Further, the sea clutter conditional probability of the feature quantity in the i-th interval of the value range is equal to the ratio of the number of specific values ​​of the feature quantity of the first sampling point belonging to the i-th interval of the value range to the number of specific values ​​of the feature quantity of all first sampling points;

[0044] The conditional probability of precipitation echo of the characteristic quantity in the i-th interval of the value range is equal to the ratio of the number of specific values ​​of the characteristic quantity of the second sampling points belonging to the i-th interval of the value range to the number of specific values ​​of the characteristic quantity of all second sampling points.

[0045] Furthermore, the calculation formula for the logarithmic data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo for each feature quantity in the i-th interval is:

[0046] ;

[0047] Among them, LD ki The specific value of the kth feature is X k The logarithmic data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo in the ith interval, K represents the number of characteristic quantities, P((X k ∈R ki ) |wave) indicates that the specific value of the kth feature is X k The conditional probability of sea clutter in the ith interval is P((X k ∈R ki ) |pre) indicates that the specific value of the kth feature is X k The conditional probability of precipitation echo in the ith interval when R ki Represents the i-th interval of the value range.

[0048] Further, the sea clutter is filtered out according to the actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of all characteristic quantities of the actual sampling points, including:

[0049] Calculate the probability of sea clutter relative to precipitation echo at each actual sampling point. The calculation formula is:

[0050] ;

[0051] Where T represents the probability of sea clutter relative to precipitation echo at the actual sampling point, ω k represents the weight of the kth feature, K represents the number of features, The specific value of the kth feature is X k The actual logarithm of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo at the time, P((X k ∈R ki ) |wave) indicates that the specific value of the kth feature is X k The conditional probability of sea clutter in the ith interval is P((X k ∈R ki ) |pre) indicates that the specific value of the kth feature is X k The conditional probability of precipitation echo in the ith interval when R ki Represents the i-th interval of the value range;

[0052] When the probability of sea clutter at an actual sampling point relative to precipitation echo is ≤ the probability threshold, the actual sampling point is a precipitation echo; when the probability of sea clutter at an actual sampling point relative to precipitation echo is > the probability threshold, the actual sampling point is sea clutter and is filtered out.

[0053] Furthermore, the process of obtaining the probability threshold and the weight of each feature quantity includes:

[0054] Constructing a second sample data set; wherein the second sample data set includes each valid sampling point and its label in each radar volume scan data, and the label includes sea clutter and precipitation echo;

[0055] Calculate specific values ​​of all feature quantities of each valid sampling point in the second sample data set;

[0056] Determine the logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity at each valid sampling point in the second sample data set according to the specific values ​​of all feature quantities at each valid sampling point in the second sample data set and the logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity;

[0057] Initialize the probability threshold and the weight of each feature;

[0058] Calculating the probability of sea clutter relative to precipitation echo at each valid sampling point in the second sample data set according to the logarithm data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo of each feature quantity at each valid sampling point in the second sample data set and the initial value of the weight of each feature quantity;

[0059] Determining the category of each valid sampling point in the second sample data set according to the probability of sea clutter relative to precipitation echo at each valid sampling point in the second sample data set and an initial value of the probability threshold;

[0060] Calculate the classification result metric according to the category and corresponding label of each valid sampling point in the second sample data set;

[0061] The probability threshold and the weight of each feature quantity are changed, and the probability calculation of the sea clutter relative to the precipitation echo for each valid sampling point in the second sample data set, the category determination of each valid sampling point in the second sample data set, and the calculation of the classification result metric index are repeated until the termination condition is met;

[0062] The probability threshold corresponding to the optimal classification result measurement indicator and the weight of each feature quantity are selected as the optimal result.

[0063] Based on the same concept, the present invention provides an electronic device, including a memory, a processor, and a computer program / instruction stored in the memory, wherein the processor executes the computer program / instruction to implement the sea clutter filtering method as described above.

[0064] Based on the same concept, the present invention provides a computer-readable storage medium having a computer program / instruction stored thereon. When the computer program / instruction is executed by a processor, the sea clutter filtering method as described above is implemented.

[0065] Based on the same concept, the present invention provides a computer program product, including a computer program / instruction, which implements the sea clutter filtering method as described above when executed by a processor.

[0066] Beneficial Effects

[0067] Compared with the prior art, the advantages of the present invention are:

[0068] The present invention makes full use of the spatial high-resolution information and dual-polarization characteristics of the radar, combines the characteristic quantity information for distinguishing sea clutter from precipitation echoes, performs statistics on the manually selected sample data set to obtain a characteristic quantity histogram, and then performs sea clutter recognition on the weather radar volume scan data to be processed based on the characteristic quantity histogram, thereby realizing the classification of precipitation echoes and sea clutter in radar echoes, and further realizing effective filtering of sea clutter, solving the problem that traditional methods cannot filter out sea clutter or the filtering effect is poor. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only one embodiment of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0070] Figure 1 is a flow chart of a method for filtering sea clutter in an embodiment of the present invention;

[0071] Figure 2 is a PPI (i.e., plane position indication) diagram of echo reflectivity at an elevation angle of 0.75° when case 1 is not filtered in an embodiment of the present invention;

[0072] Figure 3 This is a PPI graph of echo reflectivity with an elevation angle of 0.75° after filtering using the method of the present invention in Case 1 of the embodiment of the present invention;

[0073] Figure 4 is a correlation coefficient PPI graph of case 1 in the embodiment of the present invention when the elevation angle is 0.75° without filtering;

[0074] Figure 5 This is a correlation coefficient PPI graph of case 1 in the embodiment of the present invention with an elevation angle of 0.75° after filtering using the method of the present invention;

[0075] Figure 6 is a differential phase shift PPI diagram of the case 1 in the embodiment of the present invention when the elevation angle is 0.75° without filtering;

[0076] Figure 7 It is a differential phase shift PPI diagram with an elevation angle of 0.75° after filtering using the method of the present invention in Case 1 in the embodiment of the present invention;

[0077] Figure 8 is a PPI graph of echo reflectivity at an elevation angle of 0.75° when case 2 is not filtered in an embodiment of the present invention;

[0078] Fig. 9 This is a PPI graph of echo reflectivity with an elevation angle of 0.75° after filtering using the method of the present invention in Case 2 of the embodiment of the present invention;

[0079] Fig.10 is a differential phase shift PPI diagram of case 2 in an embodiment of the present invention when the elevation angle is 0.75° without filtering;

[0080] Fig.11 This is a differential phase shift PPI diagram with an elevation angle of 0.75° after filtering using the method of the present invention in Case 2 in the embodiment of the present invention;

[0081] Fig.12 is a PPI graph of echo reflectivity at an elevation angle of 0.75° when case 3 is not filtered in an embodiment of the present invention;

[0082] Fig.13 This is a PPI graph of echo reflectivity with an elevation angle of 0.75° after filtering using the method of the present invention in Case 3 of the embodiment of the present invention;

[0083] Fig.14 is a differential phase shift PPI diagram of case 3 in an embodiment of the present invention when the elevation angle is 0.75° without filtering;

[0084] Fig.15 This is a differential phase shift PPI diagram with an elevation angle of 0.75° after filtering using the method of the present invention in Case 3 of the embodiment of the present invention. DETAILED DESCRIPTION

[0085] The following is a clear and complete description of the technical solutions in the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0086] The technical solution of the present application is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0087] Example 1

[0088] like Figure 1 As shown, a sea clutter filtering method provided by an embodiment of the present invention comprises the following steps:

[0089] Step S1: construct a first sample data set.

[0090] Volume scan data containing sea clutter are screened out from the radar volume scan echo database, each volume scan data contains multiple sampling points, the volume scan data are preprocessed, invalid sampling points are removed, and each valid sampling point is obtained. The category of each valid sampling point is manually identified based on experience, and each valid sampling point is labeled. In this embodiment, the labels include sea clutter and precipitation echo, and the azimuth, elevation angle, and distance library range of the valid sampling points labeled as sea clutter and precipitation echo in each volume scan data are recorded to form a first sample data set.

[0091] In order to distinguish the valid sampling points labeled as sea clutter and the valid sampling points labeled as precipitation echo, the valid sampling points labeled as sea clutter are defined as the first sampling points, and the valid sampling points labeled as precipitation echo are defined as the second sampling points.

[0092] Step S2: Calculate the specific value of each feature quantity of each valid sampling point in the first sample data set.

[0093] By analyzing various features of valid sampling points labeled as sea clutter, it is found that the standard deviation of the differential phase shift, the mean value of the correlation coefficient, and the probability of precipitation echo in the echo block can better distinguish sea clutter from precipitation echo. Therefore, this embodiment selects the standard deviation of the differential phase shift, the mean value of the correlation coefficient, and the probability of precipitation echo in the echo block as feature quantities.

[0094] The specific value of the standard deviation of the differential phase shift is calculated as:

[0095] (1)

[0096] Among them, σ(Φ dp ) represents the specific value of the standard deviation of the differential phase shift of the first sampling point (i.e., the specific value of the standard deviation of the differential phase shift under sea clutter conditions) or the specific value of the standard deviation of the differential phase shift of the second sampling point (i.e., the specific value of the standard deviation of the differential phase shift under precipitation echo conditions), n represents the window size of the first sampling point or the second sampling point along the azimuth and radial range library, represents the differential phase shift of the jth valid sampling point in the window of the first sampling point or the second sampling point, Represents the mean of the differential phase shifts of all valid sampling points within the window of the first sampling point or the second sampling point.

[0097] When calculating the feature quantity, a sliding window method is used. The current first sampling point or the current second sampling point is windowed along the azimuth and radial distance libraries, and the specific values ​​of all feature quantities of the current first sampling point or the current second sampling point are calculated within the window range; the specific values ​​of all feature quantities of the next first sampling point or the next second sampling point are calculated in the same way. In this embodiment, the window size n=3×11, that is, the number of azimuths is 3, and the number of distance libraries is 11.

[0098] Due to the differential phase shift of the effective sampling point Φ dp There is a folding characteristic. In order to improve the calculation accuracy of the standard deviation of the differential phase shift, it is also necessary to perform an unfolding process on the differential phase shift of the effective sampling points in the window. In a specific embodiment of the present invention, the unfolding process on the differential phase shift of the effective sampling points in the window includes:

[0099] Step S2.1: The differential phase shift Φ of n valid sampling points in the window of the first sampling point or the second sampling point is dp The first differential phase shift array includes the differential phase shifts Φ of n valid sampling points. dp .

[0100] Step S2.2: Add the negative value in the first differential phase shift array to the radar folding period to obtain a second differential phase shift array.

[0101] In this embodiment, the radar folding period is 360°, the positive differential phase shifts in the first differential phase shift array remain unchanged, and only the negative differential phase shifts are added with one radar folding period.

[0102] Step S2.3: Calculate a first specific value of the standard deviation of the differential phase shift of the first sampling point or the second sampling point according to the first differential phase shift array; calculate a second specific value of the standard deviation of the differential phase shift of the first sampling point or the second sampling point according to the second differential phase shift array.

[0103] Using formula (1), specific values ​​of the standard deviation of the differential phase shift are calculated for the first differential phase shift array and the second differential phase shift array respectively.

[0104] Step S2.4: Select the smaller one of the first specific value and the second specific value as the specific value of the standard deviation of the differential phase shift of the current first sampling point or the current second sampling point.

[0105] The specific value of the mean correlation coefficient is calculated as follows:

[0106] (2)

[0107] in, represents the specific value of the mean correlation coefficient of the first sampling point (i.e., the specific value of the mean correlation coefficient under sea clutter conditions) or the specific value of the mean correlation coefficient of the second sampling point (i.e., the specific value of the mean correlation coefficient under precipitation echo conditions); CC j Represents the correlation coefficient of the jth valid sampling point in the window of the first sampling point or the second sampling point.

[0108] The calculation formula for the specific value of the probability of precipitation echo in the echo block is:

[0109] R pre =A P / A va (3)

[0110] Among them, R pre Indicates the specific value of the probability of precipitation echo in the echo block, A P A represents the area of ​​precipitation echo in the echo block in the sea area. va Indicates the echo block area in the sea surface area.

[0111] The present invention classifies the wave clutter into wave types based on the reflectivity Z and radial velocity V of the single deflection, and obtains the probability of precipitation echo of the echo block. The determination of precipitation echo depends on the wave echo intensity and radial velocity threshold in Table 1, that is, according to Table 1, it is determined whether each valid sampling point (at this time, it is not distinguished whether it is sea clutter or precipitation echo) belongs to the first type of wave echo and the second type of wave echo. If the valid sampling point does not belong to the first type of wave echo and the second type of wave echo, the valid sampling point is a precipitation echo. In Table 1, Z low Indicates the current elevation reflectivity, Z high Indicates the elevation reflectivity of the previous layer, R indicates the radial distance, and V indicates the radial velocity. The area of ​​the precipitation echo in the echo block in the sea area A P A is the number of valid sampling points for precipitation echoes in the window. The total number of valid sampling points in the window minus the number of valid sampling points judged as the first type of sea wave echoes and the second type of sea wave echoes according to Table 1 is obtained. P The echo block area A in the sea area va is the number of valid sampling points in the window. pre It reflects the degree to which the echo block is a precipitation echo, R pre The larger it is, the more likely the echo block is a precipitation echo and the less likely it is a wave echo.

[0112] Table 1 Wave echo intensity and velocity thresholds

[0113]

[0114] Step S3: Generate a logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the corresponding feature quantity according to the specific value of each feature quantity of the first sampling point and the specific value of each feature quantity of the second sampling point in the first sample data set.

[0115] In a specific embodiment of the present invention, a logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity is generated, including:

[0116] Step S3.1: Based on the specific value X of the feature quantity k of each first sampling point or second sampling point k Determine the value range R of the feature quantity k k , and the value range R k Divided into N intervals, the i-th interval can be expressed as R ki In this embodiment, N is 32.

[0117] Step S3.2: Based on the specific value X of the feature quantity k of each first sampling point k Calculate the feature value k in the i-th interval R ki The conditional probability of sea clutter; according to the specific value X of the characteristic quantity k of each second sampling point k Calculate the feature value k in the i-th interval R ki The conditional probability of precipitation echo.

[0118] Count the specific value X of the feature quantity k of the first sampling point in the i-th interval of the value range k Quantity, the specific value X of the feature quantity k of the first sampling point in the N intervals of the statistical value range k Quantity (i.e., the specific value X of the feature quantity k of all first sampling points k The ratio of the two is the specific value of the characteristic quantity k is X k When in the i-th interval R ki The conditional probability of sea clutter P((X k ∈R ki ) |wave). Similarly, the specific value X of the feature quantity k of the second sampling point in the i-th interval of the value range is counted k Quantity, the specific value X of the feature quantity k of the second sampling point within the N intervals of the statistical value range k Quantity (i.e., the specific value X of the feature quantity k of all second sampling points k The ratio of the two is the specific value of the characteristic quantity k is X k When in the i-th interval R ki The conditional probability of precipitation echo P((X k ∈R ki ) |pre).

[0119] The specific value of the feature quantity k is X k The specific value of the characteristic quantity k of the conditional probability characterization of sea clutter in the i-th interval of the value range is X k The probability of falling within the i-th interval of the value range under sea clutter conditions, the specific value of the characteristic quantity k is X k The specific value of the characteristic quantity k representing the conditional probability of precipitation echo in the i-th interval of the value range is X k The probability that falls within the i-th interval of the value range under precipitation echo conditions.

[0120] Step S3.3: Calculate the logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the corresponding feature quantity in the i-th interval according to the sea clutter conditional probability and the precipitation echo conditional probability of each feature quantity in the i-th interval.

[0121] When the characteristic quantity k is in the i-th interval of the value range, the conditional probability of sea clutter P((X k ∈R ki ) |wave) or the conditional probability P((X k ∈R ki ) |pre) is 0, it is necessary to perform Laplace smoothing on the valid sampling points, and then convert the sea clutter conditional probability P((X k ∈R ki ) |wave) and the precipitation echo conditional probability P((X k ∈R ki ) | pre) and take the logarithm to generate the specific value of feature quantity k: X k The logarithmic data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo in the ith interval at time , the specific formula is:

[0122] (4)

[0123] Among them, LD ki The specific value of the kth feature is X k The logarithmic data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo in the ith interval. Similarly, the specific value of the characteristic quantity k can be calculated as X k The logarithmic data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo in N intervals at time .

[0124] Step S3.4: Generate a logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the corresponding feature quantity according to the logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity in N intervals.

[0125] The logarithmic data LD of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo in N intervals based on the characteristic quantity k k A logarithmic data histogram of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo of the characteristic quantity k is generated. In the logarithmic data histogram, the horizontal axis is the value range of N intervals, and the vertical axis is the logarithmic data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo in the corresponding interval.

[0126] By querying the specific value X of the feature quantity k k In the range of R k The position in which X k ∈R ki When , the interval number i in the histogram of the logarithmic data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo is obtained, and the corresponding logarithmic data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo can be obtained. Substituting it into formula (11), the probability of sea clutter relative to precipitation echo at any sampling point can be obtained.

[0127] Step S4: Acquire the test sample volume scan data, and calculate the specific value of each feature quantity of the actual sampling point according to the actual sampling point in the test sample volume scan data.

[0128] According to the calculation method recorded in step S2 (such as formulas (1) to (3)), the standard deviation of the differential phase shift of the actual sampling points in the test sample volume scan data, the mean value of the correlation coefficient and the probability of precipitation echo in the echo block can be calculated.

[0129] Step S5: determining actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity at the actual sampling point according to the specific value of each feature quantity at the actual sampling point and the logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity.

[0130] According to the standard deviation of the differential phase shift at the actual sampling point, the logarithmic data histogram of the standard deviation of the differential phase shift is searched, and the actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the standard deviation of the differential phase shift at the actual sampling point can be obtained. Similarly, by searching the corresponding logarithmic data histogram, the actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the correlation coefficient mean at the actual sampling point and the actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the echo block precipitation echo probability at the actual sampling point can be obtained.

[0131] Step S6: filtering out sea clutter based on actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of all characteristic quantities at actual sampling points.

[0132] The present invention uses the weighted naive Bayesian classification principle to filter out sea clutter. According to the naive Bayesian classification formula, the conditional probability of sea clutter is:

[0133] (5)

[0134] Among them, K represents the number of features, X k Indicates the specific value of the kth feature quantity; Represents the multiplication operator; P(wave|X 1 ,X 2 ,…,X K ) represents the posterior probability of sea clutter, that is, for a valid sampling point in the volume scan data, when the specific value of the K feature quantity of the valid sampling point is X 1 ,X 2 ,…,X K When combined, the probability that the valid sampling point is sea clutter; P(wave) represents the prior probability of sea clutter, that is, the probability that any valid sampling point is sea clutter, which is obtained by statistically analyzing the sample data set; P(X k |wave) represents the feature value X of any valid sampling point among the valid sampling points labeled as sea clutter k The probability of P(X 1 ,X 2 ,…,X K ) indicates that the value of K feature quantities at any valid sampling point is X 1 ,X 2 ,…,X K The probability of a combined event occurring does not need to be calculated when classifying P(X 1 ,X 2 ,…,X K ), considered as coefficient term.

[0135] According to the naive Bayes classification formula, the probability of precipitation echo is:

[0136] (6)

[0137] Among them, P(pre|X 1 ,X 2 ,…,X K ) represents the posterior probability of precipitation echo, that is, for a valid sampling point in the volume scan data, when the specific value of the K characteristic quantities of the valid sampling point is X 1 ,X 2 ,…,X K When combined, the probability that the valid sampling point is a precipitation echo; P(pre) represents the prior probability of precipitation echo, that is, the probability that any valid sampling point is a precipitation echo, which is obtained by statistically analyzing the sample data set; P(X k|pre) represents the feature value X of any valid sampling point among the valid sampling points labeled as precipitation echoes k probability.

[0138] Without considering the prior probability, it is assumed that the prior probability of sea clutter at the valid sampling points in the volume scan data is equal to the prior probability of precipitation echo, that is, P(wave) = P(pre) = 0.5. By comparing the values ​​of formula (5) and formula (6), the naive Bayes classification results of precipitation echo and sea clutter can be obtained, where c represents the echo category (i.e., precipitation echo or sea clutter):

[0139] (7)

[0140] Where P(c) represents the probability that the echo category is c, that is, P(wave) or P(pre); Used to determine the category c with the highest probability.

[0141] The probability of each feature quantity k in the naive Bayes classification results of precipitation echo and sea clutter (as shown in expression (7)) is assigned a weight ω k , forming a weighted naive Bayes classification expression:

[0142] (8)

[0143] Superscript ω k represents the weight of feature quantity k, c only contains precipitation echoes or sea clutter, and Laplace smoothing is performed in advance, so that the probability P(X k |wave) is greater than 0, and it is assumed that P(wave) = P(pre), the argmax comparison in expression (8) can be changed to division, and converted to:

[0144] (9)

[0145] Taking the base 10 logarithm of both sides of formula (9) and simplifying it, we get:

[0146] (10)

[0147] Let the left side of formula (10) be:

[0148] (11)

[0149] Where T represents the probability of sea clutter relative to precipitation echo at the sampling point.

[0150] Set the weight of each feature value ω k , according to the specific value X of the characteristic quantity k of the actual sampling point kQuery the logarithmic data histogram of step S3 to obtain the specific value X of the feature quantity k of the actual sampling point k The interval where the specific value of the characteristic quantity of the actual sampling point is located can be obtained, and the interval number of the value range can be obtained, and then the actual logarithmic data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo of the specific value of the characteristic quantity of the actual sampling point can be obtained. Substituting it into formula (11), the probability of sea clutter relative to precipitation echo at the actual sampling point can be calculated.

[0151] When the probability T of the sea clutter relative to the precipitation echo at the actual sampling point is less than the probability threshold T 0 When the probability T of the sea clutter at the actual sampling point relative to the precipitation echo is greater than the probability threshold T 0 When , the actual sampling point is sea clutter, and the actual sampling point is filtered out.

[0152] In order to improve the calculation accuracy and classification accuracy of the probability of sea clutter relative to precipitation echo at the actual sampling point, it is necessary to obtain the appropriate weight ω of the feature quantity. k In a specific embodiment of the present invention, the probability threshold and the weight ω of each feature quantity k The acquisition process includes:

[0153] Step S6.1: construct a second sample data set; wherein the second sample data set includes each valid sampling point in each radar volume scan data and its label, and the label includes sea clutter and precipitation echo.

[0154] The construction process of the second sample data set is the same as that of the first sample data set, which will not be described in detail here. In another specific embodiment of the present invention, a sample data set is constructed in step S1, and then the sample data set is divided according to a proportion, wherein one sample data set is used to generate a logarithmic data histogram, and the other sample data set is used to obtain the optimal feature weight ω k and probability thresholds.

[0155] Step S6.2: Calculate the specific values ​​of all feature quantities of each valid sampling point in the second sample data set, and the specific calculation method is the same as the calculation method in step S2.

[0156] Step S6.3: Determine the logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity of each valid sampling point in the second sample data set based on the specific values ​​of all feature quantities of each valid sampling point in the second sample data set and the logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity, and the specific determination method is the same as the determination method in step S5.

[0157] Step S6.4: Initialize the probability threshold and the weight of each feature quantity.

[0158] Step S6.5: Calculate the probability of sea clutter relative to precipitation echo at each valid sampling point in the second sample data set according to the logarithmic data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo for each characteristic quantity of each valid sampling point in the second sample data set and the initial value of the weight of each characteristic quantity, as shown in formula (11).

[0159] Step S6.6: Determine the category of each valid sampling point in the second sample data set according to the probability of sea clutter relative to precipitation echo at each valid sampling point in the second sample data set and the initial value of the probability threshold.

[0160] Step S6.7: Calculate the classification result metric based on the category and corresponding label of each valid sampling point in the second sample data set.

[0161] In this embodiment, the classification result measurement indicators are detection probability POD and false alarm rate FAR.

[0162] Step S6.8: Change the probability threshold and the weight of each feature quantity, and repeat steps S6.5 to 6.7 until the termination condition is met.

[0163] The termination condition can be the maximum number of iterations, or a reasonable range and a reasonable step size. When the termination condition is the maximum number of iterations, the iteration stops when the number of iterations reaches the maximum number of iterations. When the termination condition is a reasonable range and a reasonable step size, the value is taken within a reasonable range according to the reasonable step size until all values ​​are taken.

[0164] Exemplarily, the weight range is 0 to 4, and the step size of the weight is 0.5; the probability threshold range is -4 to 4, and the step size of the probability threshold is 1. In this way, multiple combinations of 3 weights and probability thresholds can be obtained, and the classification result measurement index under each combination is calculated, so as to select the probability threshold corresponding to the optimal classification result measurement index and the weight of each feature quantity.

[0165] Step S6: Select the probability threshold corresponding to the optimal classification result measurement indicator and the weight of each feature quantity as the optimal result.

[0166] The probability threshold corresponding to the maximum detection probability POD and the minimum false alarm rate FAR and the weight of each feature quantity are taken as the optimal result. In this embodiment, the optimal weights of the standard deviation of the differential phase shift, the mean value of the correlation coefficient and the probability of precipitation echo of the echo block are 2, 0.5 and 1.5 respectively, and the optimal probability threshold is 0.

[0167] In order to verify the effect of the present invention, the volume scan data of two radars (ZK803 / ZK301) during typical precipitation from August 2023 to March 2024 were selected to generate the second sample data set. The effect of the method of the present invention was analyzed by comparing with the unfiltered and sea clutter filtered data. Some test periods are shown in Table 2.

[0168] Table 2 Data of some test periods

[0169]

[0170] The recognition effect of echo sea clutter in some time periods is as follows Figures 2 to 15 As shown, case 1 is that the radar ZK301 is in a clear sky environment, and the body scan time is 18:01:00 on February 5, 2024 (i.e., 20240205180100); case 2 is that the radar ZK301 is in a strong convection environment, and the body scan time is 06:00:00 on August 22, 2023 (i.e., 20230822060000); case 3 is that the radar ZK803 is in a strong convection environment, and the body scan time is 15:00:00 on March 5, 2024 (20240305150000); the horizontal axis X represents the distance in the east-west direction, and the vertical axis Y represents the distance in the north-south direction, and the unit is km; Figure 2 , Figure 3 , Figure 8 , Fig. 9 , Fig.12 and Fig.13 The color scale in represents the reflectivity intensity in dB; Figure 4 and Figure 5 The color scale in indicates the size of the correlation coefficient; Figure 6 , Figure 7 , Fig.10 , Fig.11 , Fig.14 and Fig.15 The color scale in represents the differential phase shift in degrees. For example, Figure 2 This is the PPI diagram of the echo reflectivity of radar ZK301 at an elevation angle of 0.75° when it was not filtered at 20240205180100 in a clear sky environment.

[0171] Figures 2 to 7 In case 1, the ZK301 radar is in a clear sky environment. Figure 2 , Figure 4 and Figure 6 It can be seen that the reflectivity of the southeast sea surface echo in the unfiltered data is uniform and flake-shaped, the correlation coefficient is significantly low, the differential phase shift is chaotic, and the sea clutter characteristics are clear, which is obvious sea clutter; Figure 3 , Figure 5 and Figure 7 It can be seen that after being processed by the filtering method of the present invention, Figure 3 , Figure 5 and Figure 7 The sea clutter in the image is completely filtered out, and the effect is good.

[0172] Figures 8 to 11 In case 2, the ZK301 radar echo shows convective weather, Figure 8 and Fig.10 It can be seen that the weather echo and sea clutter on the southeast sea surface are mixed in the unfiltered data. The reflectivity of the sea clutter is uniform and flake-like, and the differential phase shift is messy, which is obvious sea clutter. Fig. 9 and Fig.11 It can be seen that after being processed by the filtering method of the present invention, Fig. 9 and Fig.11 The sea clutter in the image is completely filtered out, and there is basically no misidentification of weather echoes, with a good effect.

[0173] Figures 12 to 15 In case 3, the ZK803 radar echo shows convective weather, Fig.12 and Fig.14 It can be seen that in the unfiltered data, the sea clutter and weather echoes on the eastern and southeastern sea surfaces are mixed, and the reflectivity of the sea clutter is uniform and flake-shaped, and the differential phase shift is chaotic, which is obvious sea clutter. Fig.13 and Fig.15 It can be seen that after being processed by the filtering method of the present invention, Fig.13 and Fig.15 The sea clutter in the image is completely filtered out, and there is basically no misidentification of weather echoes, with a good effect.

[0174] Example 2

[0175] An embodiment of the present invention further provides an electronic device, which includes: a memory, a processor, and a computer program / instruction stored in the memory, and the processor executes the computer program / instruction to implement the sea clutter filtering method in the embodiment of the present application.

[0176] Although not shown, the electronic device includes a processor that can perform various appropriate operations and processes according to the programs and / or data stored in the read-only memory (ROM) or the programs and / or data loaded from the storage part into the random access memory (RAM). The processor can be a multi-core processor or can include multiple processors. In some embodiments, the processor can include a general-purpose main processor and one or more special coprocessors, such as a central processing unit, a graphics processing unit (GPU), a neural network processor (NPU), a digital signal processor (DSP), etc. In RAM, various programs and data required for device operation are also stored. The processor, ROM, and RAM are connected to each other via a bus. The input / output (I / O) interface is also connected to the bus.

[0177] The processor and memory are used together to execute the program / instructions stored in the memory. When the program / instructions are executed by the computer, the methods, steps or functions described in the above embodiments can be implemented.

[0178] Although not shown, an embodiment of the present invention further provides a computer-readable storage medium on which a computer program / instruction is stored. When the computer program / instruction is executed by a processor, the sea clutter filtering method in the embodiment of the present application is implemented.

[0179] Storage media in embodiments of the present invention include permanent and non-permanent, removable and non-removable items that can be implemented by any method or technology to store information. Examples of storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0180] Readable storage media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include temporary computer-readable media (transitory media), such as modulated data signals and carrier waves.

[0181] Although not shown, an embodiment of the present invention further provides a computer program product, including: a computer program / instruction, which implements the sea clutter filtering method in the embodiment of the present application when the computer program / instruction is executed by a processor.

[0182] What is disclosed above is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or modifications within the technical scope disclosed in the present invention, which should be covered within the protection scope of the present invention.

Claims

1. A method for filtering sea clutter, characterized in that: The filtering method comprises: Constructing a first sample data set; wherein the first sample data set includes each valid sampling point and its label in each radar volume scan data, the label includes sea clutter and precipitation echo, defining the valid sampling point labeled as sea clutter as the first sampling point, and the valid sampling point labeled as precipitation echo as the second sampling point; Calculating a specific value of each characteristic quantity of the first sampling point according to the first sampling point of the first sample data set, and calculating a specific value of each characteristic quantity of the second sampling point according to the second sampling point of the first sample data set; wherein the characteristic quantity is a parameter capable of distinguishing sea clutter from precipitation echo; Generate a logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the corresponding feature quantity according to the specific value of each feature quantity at the first sampling point and the specific value of each feature quantity at the second sampling point; Acquire test sample volume scan data, and calculate specific values ​​of each feature quantity of the actual sampling point according to the actual sampling point in the test sample volume scan data; Determining actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity at the actual sampling point according to the specific value of each feature quantity at the actual sampling point and the logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity; Sea clutter is filtered out according to actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of all characteristic quantities of the actual sampling points; The characteristic quantities include the standard deviation of the differential phase shift, the mean value of the correlation coefficient and the probability of precipitation echo of the echo block; The specific value of the standard deviation of the differential phase shift is calculated as follows: ； Among them, σ(Φ dp ) represents the specific value of the standard deviation of the differential phase shift of the first sampling point or the specific value of the standard deviation of the differential phase shift of the second sampling point, n represents the window size of the first sampling point or the second sampling point along the azimuth and radial distance library, Φ dpj represents the differential phase shift of the jth valid sampling point in the window of the first sampling point or the second sampling point, The mean of the differential phase shifts of all valid sampling points within the window representing the first sampling point or the second sampling point; The calculation formula for the specific value of the correlation coefficient mean is: ； in, represents the specific value of the mean value of the correlation coefficient of the first sampling point or the specific value of the mean value of the correlation coefficient of the second sampling point; CC j represents the correlation coefficient of the jth valid sampling point in the window of the first sampling point or the second sampling point; The calculation formula for the specific value of the probability of precipitation echo in the echo block is: R pre =A P / A va ; Among them, R pre Indicates the specific value of the probability of precipitation echo in the echo block, A P A represents the area of ​​precipitation echo in the echo block in the sea area. va Indicates the echo block area in the sea surface area.

2. The sea clutter filtering method according to claim 1, characterized in that: When calculating the specific value of the standard deviation of the differential phase shift, the differential phase shift of the effective sampling point is also unfolded, and the specific process includes: Copying the differential phase shifts of n valid sampling points within the window of the first sampling point or the second sampling point to obtain a first differential phase shift array; Adding the negative value in the first differential phase shift array to the radar folding period to obtain a second differential phase shift array; Calculating a first specific value of the standard deviation of the differential phase shift of the first sampling point or the second sampling point according to the first differential phase shift array; calculating a second specific value of the standard deviation of the differential phase shift of the first sampling point or the second sampling point according to the second differential phase shift array; The smaller one of the first specific value and the second specific value is selected as the specific value of the standard deviation of the differential phase shift of the current first sampling point or the current second sampling point.

3. The sea clutter filtering method according to claim 1, characterized in that: A logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the corresponding feature quantity is generated according to the specific value of each feature quantity of the first sampling point and the specific value of each feature quantity of the second sampling point, including: Determine a value range of the corresponding feature quantity according to the specific value of each feature quantity of each first sampling point or the second sampling point, and divide the value range into N intervals; According to the specific value of each characteristic quantity of each first sampling point, the conditional probability of sea clutter of the corresponding characteristic quantity in the i-th interval of the value range is calculated; according to the specific value of each characteristic quantity of each second sampling point, the conditional probability of precipitation echo of the corresponding characteristic quantity in the i-th interval of the value range is calculated; According to the sea clutter conditional probability and precipitation echo conditional probability of each characteristic quantity in the i-th interval, the logarithmic data of the ratio of the sea clutter conditional probability and the precipitation echo conditional probability of the corresponding characteristic quantity in the i-th interval is calculated; A logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of the corresponding feature quantity is generated according to the logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity in N intervals.

4. The sea clutter filtering method according to claim 3, characterized in that: The sea clutter conditional probability of the characteristic quantity in the i-th interval of the value range is equal to the ratio of the number of specific values ​​of the characteristic quantity of the first sampling point belonging to the i-th interval of the value range to the number of specific values ​​of the characteristic quantity of all first sampling points; The conditional probability of precipitation echo of the characteristic quantity in the i-th interval of the value range is equal to the ratio of the number of specific values ​​of the characteristic quantity of the second sampling points belonging to the i-th interval of the value range to the number of specific values ​​of the characteristic quantity of all second sampling points.

5. The sea clutter filtering method according to any one of claims 1 to 4, characterized in that: Sea clutter filtering is performed according to actual logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of all characteristic quantities of the actual sampling points, including: Calculate the probability of sea clutter relative to precipitation echo at each actual sampling point. The calculation formula is: ； Where T represents the probability of sea clutter relative to precipitation echo at the actual sampling point, ω k represents the weight of the kth feature, K represents the number of features, The specific value of the kth feature is X k The actual logarithm of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo at the time, P((X k ∈R ki ) |wave) indicates that the specific value of the kth feature is X k The conditional probability of sea clutter in the ith interval is P((X k ∈R ki ) |pre) indicates that the specific value of the kth feature is X k The conditional probability of precipitation echo in the ith interval when R ki Represents the i-th interval of the value range; When the probability of sea clutter at an actual sampling point relative to precipitation echo is ≤ the probability threshold, the actual sampling point is a precipitation echo; when the probability of sea clutter at an actual sampling point relative to precipitation echo is > the probability threshold, the actual sampling point is sea clutter and is filtered out.

6. The sea clutter filtering method according to claim 5, characterized in that: The process of obtaining the probability threshold and the weight of each feature quantity includes: Constructing a second sample data set; wherein the second sample data set includes each valid sampling point and its label in each radar volume scan data, and the label includes sea clutter and precipitation echo; Calculate specific values ​​of all feature quantities of each valid sampling point in the second sample data set; Determine the logarithmic data of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity at each valid sampling point in the second sample data set according to the specific values ​​of all feature quantities at each valid sampling point in the second sample data set and the logarithmic data histogram of the ratio of the sea clutter conditional probability to the precipitation echo conditional probability of each feature quantity; Initialize the probability threshold and the weight of each feature; Calculating the probability of sea clutter relative to precipitation echo at each valid sampling point in the second sample data set according to the logarithm data of the ratio of the conditional probability of sea clutter to the conditional probability of precipitation echo of each feature quantity at each valid sampling point in the second sample data set and the initial value of the weight of each feature quantity; Determining the category of each valid sampling point in the second sample data set according to the probability of sea clutter relative to precipitation echo at each valid sampling point in the second sample data set and an initial value of the probability threshold; Calculate the classification result metric according to the category and corresponding label of each valid sampling point in the second sample data set; The probability threshold and the weight of each feature quantity are changed, and the probability calculation of the sea clutter relative to the precipitation echo for each valid sampling point in the second sample data set, the category determination of each valid sampling point in the second sample data set, and the calculation of the classification result metric index are repeated until the termination condition is met; The probability threshold corresponding to the optimal classification result measurement indicator and the weight of each feature quantity are selected as the optimal result.

7. An electronic device comprising a memory, a processor, and a computer program / instruction stored in the memory, characterized in that: The processor executes the computer program / instructions to implement the sea clutter filtering method according to any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instruction is executed by a processor, the sea clutter filtering method as described in any one of claims 1 to 6 is implemented.

9. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the sea clutter filtering method as described in any one of claims 1 to 6 is implemented.

Images