Sea state classification method and device based on radar echo characteristics

Abstract

The present invention provides a sea condition classification method and device based on radar echo characteristics, the method comprising: acquiring sea clutter data to be classified; preprocessing the sea clutter data to obtain preprocessing parameters, and extracting target features from the preprocessing results according to preset feature extraction rules; inputting the sea clutter data based on the target features into a prestored classifier for classification to obtain classification results; the method of the present invention improves the way of preprocessing the sea clutter data, obtains more effective features from the sea clutter data to realize the classification of high / low sea condition sea clutter, and improves the classification accuracy.

Description

Technical Field

[0001] The invention belongs to the field of radar clutter characteristics research, and in particular relates to a sea condition classification method and device based on radar echo characteristics. Background Art

[0002] Sea conditions are important background information for radar detection of maritime targets, and accurate judgment of them plays an important role in the refined detection of radar targets.

[0003] In order to achieve robust target detection in sea clutter, it is necessary to perform refined dynamic identification of the background information of the entire detection scene, and then divide the detection scene into multiple sub-scenes. For different sub-scenes, the target detection method matching them is used for detection, thereby improving the overall target detection performance. The level of sea conditions, as an important background information for radar detection of marine targets, greatly affects the division of detection scenes and the selection of detectors. The data preprocessing method used in the existing technology for identifying high / low sea conditions is relatively single, and it is impossible to stably highlight the difference between high / low sea conditions, resulting in poor generalization ability of the classifier, which cannot meet the requirements of the radar working in scanning mode. Summary of the invention

[0004] The present invention provides a sea condition classification method and device based on radar echo characteristics, which are used to solve the problem that the prior art has a relatively simple data preprocessing process when classifying sea conditions, and thus cannot effectively screen out information that can stably highlight the difference between high / low sea conditions, resulting in a low generalization ability of the classifier, thereby improving the accuracy of sea condition classification.

[0005] The present invention provides a sea condition classification method based on radar echo characteristics, the method comprising:

[0006] Acquire sea clutter data to be classified; preprocess the sea clutter data to obtain preprocessing parameters, and extract target features from the preprocessing results according to preset feature extraction rules; input the sea clutter data based on the target features into a pre-stored classifier for classification to obtain classification results.

[0007] According to a sea condition classification method based on radar echo characteristics provided by the present invention, the method further includes:

[0008] The sea clutter data is converted from a one-dimensional random sequence in the time domain to a two-dimensional random sequence in the time-frequency domain by using a time-frequency transformation tool, and preprocessing parameters representing the two-dimensional random sequence are obtained, wherein the preprocessing parameters include a mean function and a standard deviation function.

[0009] The preprocessing parameters are quantized and summed based on the ridge integral theory to obtain the target feature, wherein the target feature includes the ridge integral of the mean function and the ridge integral of the standard deviation function.

[0010] The regional peak power and the noise mean power of the sea clutter data are calculated respectively; the ratio of the regional peak power to the noise mean power is calculated, and the target feature is constructed based on the ridge integral of the mean function, the ridge integral of the standard deviation function and the ratio.

[0011] The classifier includes any one of short-time Fourier transform, Wigner-Ville distribution and smoothed pseudo WVD.

[0012] The classifier includes any one of the machine learning or deep learning algorithms suitable for target binary classification, such as SVM, random forest and deep neural network.

[0013] The present invention also provides a sea condition classification device based on radar echo, the device comprising:

[0014] A data acquisition module is used to acquire sea clutter data to be classified; a feature extraction module is used to preprocess the sea clutter data to obtain preprocessing parameters, and extract target features from the preprocessing results according to preset feature extraction rules; and a classification module is used to input the sea clutter data based on the target features into a pre-stored classifier for classification to obtain classification results.

[0015] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the sea condition classification based on radar echo as described in any one of the above is realized.

[0016] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the sea condition classification method based on radar echo characteristics as described in any one of the above is implemented.

[0017] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements any of the above-mentioned sea condition classification methods based on radar echo characteristics.

[0018] The present invention provides a sea condition classification method and device based on radar echo characteristics, which first obtains sea clutter data to be classified, then preprocesses the sea clutter data to obtain preprocessing parameters, extracts target features from the preprocessing results according to preset feature extraction rules, and finally inputs the sea clutter data based on the target features into a prestored classifier for classification to obtain classification results; the method of the present invention improves the way of preprocessing sea clutter data, obtains more effective features from the sea clutter data to realize the classification of high / low sea condition sea clutter, and improves the classification accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 It is a flow chart of a sea condition classification method based on radar echo characteristics provided by an embodiment of the present invention;

[0021] Figure 2 is a flow chart of a classification method using two features combined with SVM provided by another embodiment of the present invention;

[0022] Figure 3 It is a flowchart of a classification method using three features combined with SVM provided by another embodiment of the present invention;

[0023] Figure 4 is a schematic structural diagram of a sea condition classification device based on radar echo provided in an embodiment of the present invention;

[0024] Figure 5 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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.

[0026] Combine the following Figure 1 The sea condition classification method based on radar echo characteristics provided by an embodiment of the present invention includes:

[0027] Step 101: Acquire sea clutter data to be classified.

[0028] It can be understood that sea clutter is the backscattered echo of the electromagnetic wave emitted by the radar on the sea surface. The characteristics of sea clutter are highly complex. Under the high-resolution radar system, the non-stationary, non-uniform and non-Gaussian characteristics of sea clutter are more obvious. The same distribution characteristics of sea clutter in the entire detection scene are significantly reduced, and it is difficult to achieve the independent and identically distributed prerequisite required by the classic CFAR method and the adaptive detection method. Therefore, extracting some feature information that can stably highlight the difference between high / low sea conditions from the sea clutter data is the key to studying the sea state classification under sea clutter; this embodiment obtains a large amount of sea clutter data from the designated "Radar Sea Detection Data Sharing Plan" data set as an input sample set for the subsequent classifier. The data set mainly includes sea clutter and target echo data under different sea state levels. During data acquisition, the radar mainly works in the staring mode, the polarization mode is HH polarization, the range sampling rate is 60MHz, the radar frame height is 80m, and the pulse repetition frequency (PRF) is 1.6kHz.

[0029] Step 102: preprocess the sea clutter data to obtain preprocessing parameters, and extract target features from the preprocessing results according to preset feature extraction rules.

[0030] Specifically, sea clutter is the backscattered echo of the sea surface to the electromagnetic waves emitted by the radar. In this embodiment, the sea clutter is first modeled as a non-stationary random process, and time-frequency transformation is performed on it to conveniently reflect its essential characteristics, and the transformation parameters of the sea clutter data after the time-frequency transformation are obtained. Then, based on the theory of ridge integral, the parameters are quantified, summed, etc. to obtain the required target features. The target features are used as input features of the subsequent classifier, and contain information that can stably highlight the difference of different sea conditions.

[0031] Step 103: input the sea clutter data based on the target features into a pre-stored classifier for classification to obtain a classification result.

[0032] It can be understood that in order to verify the accuracy of the target feature in characterizing the classification information of the sea condition under the sea clutter number, the present embodiment uses the extracted target as the input feature of the classifier. After completing the training and testing of the classifier, the classification result of the sea clutter data can be obtained. Specifically, the present embodiment first obtains part of the sea clutter data from the "Radar Sea Detection Data Sharing Program" data set as the input sample set, and then uses the short-time Fourier transform (STFT) as a time-frequency transformation tool to perform time-frequency transformation on the training set of the input sample to obtain the transformation parameter, which can be the mean function or standard deviation function of the random process. Then, according to the ridge integral theory, the difference of the sea clutter data in the time-frequency domain is quantified, and the target feature is obtained according to the quantification result. Finally, the sea clutter data based on the target feature is sent to the pre-stored support vector machine (SVM) for testing to obtain the binary classification result of high / low sea conditions under sea clutter.

[0033] The method described in this embodiment improves the way of preprocessing sea clutter data, obtains more effective features from the sea clutter data to realize the classification of high / low sea condition sea clutter, and improves the classification accuracy.

[0034] Optionally, a time-frequency transformation tool is used to convert the sea clutter data from a one-dimensional random sequence in the time domain to a two-dimensional random sequence in the time-frequency domain, and preprocessing parameters representing the two-dimensional random sequence are obtained, wherein the preprocessing parameters include a mean function and a standard deviation function.

[0035] Specifically, this embodiment uses STFT as a time-frequency transformation tool and uses SPWVD for verification. The cross term has a certain influence on the distinguishing effect of the proposed feature. Assume that the sea radar transmits multiple coherent pulses in a certain direction, and a sea clutter sequence with a length of N is received at a certain distance unit, that is, r(n)=c(n), n=1,2,...,N. In a smaller range, the sea state levels of the surrounding distance units can be regarded as the same level, and the echo pulses received by the surrounding distance units r s (n) is expressed as:

[0036] r s (n) = c s (n), n=1,2,…,N, s=1,2,…,S

[0037] Among them, c s(n) represents the sea clutter sequence received at the surrounding S range units, and S represents the number of surrounding range units. STFT is performed on the sea clutter of the range unit to obtain the time-frequency spectrum STFT(t,f|r). Here, only the preprocessing of the time-frequency transformation is performed, and the difference between the high / low sea conditions sea clutter in the time-frequency domain cannot be stably highlighted. In this embodiment, the sea clutter sequence is converted from a one-dimensional random process in the time domain to a two-dimensional random process in the time-frequency domain through STFT. The mean function and standard deviation function of the random process are expressed as μ(t,f) and σ(t,f), which respectively characterize the average power level and fluctuation degree of the sea clutter at the point (t,f) of the two-dimensional time-frequency grid; while μ(t,f) and σ(t,f) are unknown. In this embodiment, μ(t,f) and σ(t,f) are estimated through the range units around the sea clutter unit, and the calculation formula is as follows:

[0038]

[0039]

[0040] Both μ(t,f) and σ(t,f) can stably and prominently reflect the differences in time-frequency domain between high and low sea conditions.

[0041] This embodiment provides a preprocessing method for performing time-frequency transformation on sea clutter data, which can extract the difference information of sea clutter in the time-frequency domain that can stably and prominently reflect high / low sea conditions from the sea clutter data, and facilitate the subsequent process of quantifying the difference information to obtain target features.

[0042] Optionally, the preprocessing parameters are quantized and summed based on ridge integral theory to obtain the target features, wherein the target features include the ridge integral of the mean function and the ridge integral of the standard deviation function.

[0043] Specifically, this embodiment quantifies the difference information of sea clutter data of high / low sea conditions in the time-frequency domain based on ridge integration (RI), and the ridge integration represents the sum of all ridge points; the ridges of the mean function μ(t, f) and the standard deviation function σ(t, f) are defined as follows:

[0044]

[0045]

[0046] The above formula represents the set of maximum values ​​of μ(t,f) or σ(t,f) at each time slice t; the ridge integral of mean function (RIMF) and the ridge integral of standard deviation function (RISD) are the sum of all ridge point values, which are defined as follows:

[0047]

[0048]

[0049] Among them, RIMF and RISD respectively represent a power level and fluctuation degree of sea clutter. The statistical distribution characteristics of RIMF and RISD are closely related to the actual radial distance corresponding to the sea clutter range unit. Figure 2 As shown, this embodiment first obtains part of the sea clutter data from the "Radar Sea Detection Data Sharing Program" data set as an input sample set, and then uses short-time Fourier transform (STFT) as a time-frequency transformation tool to perform time-frequency transformation on the training set of the input samples to obtain transformation parameters, which can be a mean function or a standard deviation function of a random process. Then, based on the ridge integral theory, the difference of the sea clutter data in the time-frequency domain is quantified, and RIMF features and RISD features are obtained according to the quantification results. Finally, the sea clutter data based on the RIMF features and RISD features are sent to a pre-stored support vector machine (SVM) for testing to obtain a binary classification result of high / low sea conditions under sea clutter.

[0050] This embodiment provides a method for quantifying and summing sea clutter data after time-frequency transformation based on the ridge integral theory to obtain RIMF features and RISD features. The purpose is to quantify the difference information of high / low sea clutter in the time-frequency domain and provide input features for the subsequent classification process.

[0051] Optionally, the regional peak power and the noise mean power of the sea clutter data are calculated respectively; the ratio of the regional peak power to the noise mean power is calculated, and the target feature is constructed based on the ridge integral of the mean function, the ridge integral of the standard deviation function and the ratio.

[0052] It can be understood that, since there are obvious differences in the Doppler power spectrum between high and low sea conditions, on the one hand, the clutter main lobe spectral width (3dB spectral width) of the Doppler power spectrum of the sea clutter in high sea conditions is significantly greater than the clutter main lobe spectral width of the Doppler power spectrum of the sea clutter in low sea conditions. On the other hand, the clutter-to-noise ratio (CNR) of the sea clutter in high sea conditions is greater than that of the sea clutter in low sea conditions. Using the spectral width or CNR alone to distinguish between high and low sea conditions cannot achieve stable results. Therefore, after obtaining the two features of RIMF feature and RISD feature, this embodiment combines the two features of spectral width and CNR to propose a third feature, i.e., a new Doppler power spectrum feature, i.e., the ratio of regional peak power to noise mean power (RPNM), to characterize the difference in the Doppler power spectrum between high and low sea conditions. RPNM is defined as follows:

[0053]

[0054] X1(p)p=1,2,...,P represents the power spectrum of the peak area, and its position can be determined by the position of the Doppler power spectrum peak. The peak area size P is determined by the 3dB spectrum width of the Doppler power spectrum of sea clutter in high sea conditions; X2(k)k=1,2,...,K represents the noise area of ​​the power spectrum, and its range is roughly the area outside the range of 3 times the peak area of ​​the Doppler power spectrum of sea clutter in high sea conditions. It should be noted that in order to reduce the variance of the Doppler power spectrum, this embodiment needs to use the AR model to estimate the power spectrum. In addition, in order to increase the stability and effectiveness of the feature, it is necessary to use the data of several distance units around r(n) in combination to obtain the average Doppler power spectrum, and then calculate the corresponding RPNM feature; specifically, according to Figure 3 As shown, this embodiment first obtains part of the sea clutter data from the "Radar Sea Detection Data Sharing Program" data set as an input sample set, and then uses short-time Fourier transform (STFT) as a time-frequency transformation tool to perform time-frequency transformation on the training set of the input samples to obtain transformation parameters, which can be the mean function or standard deviation function of the random process. Then, based on the ridge integral theory, the difference of the sea clutter data in the time-frequency domain is quantified, and the RIMF feature and RISD feature are obtained according to the quantization result. Then, the spectrum width and CNR features are combined to construct a new Doppler power spectrum feature RPNM. Finally, the sea clutter data based on the three types of features of RIMF, RISD and RPNM are sent to the pre-stored support vector machine (SVM) for testing to obtain the binary classification results of high / low sea conditions under sea clutter.

[0055] This embodiment provides another method for characterizing the difference information of high / low sea clutter, which is combined with RIMF features and RISD features as three input features of the classifier, and can effectively improve the ability to identify high / low sea conditions.

[0056] Optionally, the classifier includes any one of short-time Fourier transform, Wigner-Ville distribution and smoothed pseudo WVD.

[0057] It can be understood that the purpose of performing video transformation on sea clutter data in this embodiment is to convert the sea clutter sequence from a one-dimensional random process in the time domain to a two-dimensional random process in the time-frequency domain, so that the mean function and standard deviation function representing the sequence can be extracted, so as to extract the difference information of the high / low sea clutter in the time-frequency domain that stably highlights the sea clutter. The time-frequency transformation tool used in this embodiment is short-time Fourier transform (STFT), and other embodiments can also use Wigner-Ville distribution (WVD) or smoothed pseudo WVD (Smoothed Pseudo WVD, SPWVD), etc. They each have their own advantages and disadvantages, and the corresponding time-frequency transformation method can be flexibly selected according to the input data type.

[0058] The method described in this embodiment provides several transformation tools for performing time-frequency transformation on sea clutter data, which are used to transform the sea clutter sequence from a one-dimensional random process in the time domain to a two-dimensional random process in the time-frequency domain, so as to facilitate the extraction of mean function and standard deviation function in the subsequent process.

[0059] Optionally, the classifier includes any one of the machine learning or deep learning algorithms suitable for target binary classification, such as SVM, random forest and deep neural network.

[0060] It can be understood that the SVM used in this embodiment is a machine learning classifier widely used in pattern recognition problems. It can realize the binary classification of medium and high dimensional vectors with a small number of training samples, and the trained classifier has a high generalization ability. Specifically, this embodiment first extracts three features of RIMF, RISD and RPNM from sea clutter to form the required feature sample set. For any sample i in the training set, this paper constructs a RIMF (f i,1 )、RISD(f i,2 ) and the RPNM of the Doppler spectrum (f i,3 ) is an ordered three-dimensional feature vector F i , F i The expression is as follows:

[0061] F i =[fi,1 ,f i,2 ,f i,3 ] T

[0062] And use label y i ∈{+1,-1}, marking the categories of high sea clutter (+1) and low sea clutter (-1), so the training set with a sample size of M is {(F i ,y i ),i=1,2,...,M}. In fact, the feature vectors of high / low sea clutter in the constructed feature space are not necessarily linearly separable, and may even be essentially linearly inseparable; in order to solve the problem of essentially linear inseparability, SVM adopts a generalized linearization method, that is, by introducing a kernel function, the nonlinear separable problem in the low-dimensional feature space is converted into a linear classification problem in a high-dimensional feature space. At the same time, the introduction of the kernel function also avoids the sudden increase in the amount of calculation caused by the dimensional conversion; commonly used kernel functions include Gaussian kernel function, radial basis kernel function, linear kernel function and polynomial kernel function, etc. The general expression of the kernel function is as follows:

[0063] k(F1,F2)=Φ T (F1)Φ(F2)

[0064] Among them, Φ(F1) and Φ(F2) represent the mapping of feature vectors F1 and F2 in the high-dimensional feature space respectively; after mapping to the high-dimensional feature space, the key to the problem is to solve the hyperplane of the linear SVM, that is, ω T F+b=0, where the weight vector ω and the intercept b are unknown parameters, which can be solved by solving the optimization problem in the following formula:

[0065]

[0066] sty i [k(ω,F i )+b]≥1-ξ i ,ξ i ≥0,i=1,2,...,M

[0067] ξ i is a slack variable, an unknown variable, used to convert the classification interval into a soft interval; C is a penalty factor, which is greater than 0 and indicates the tolerance for the existence of outliers in the classifier. This optimization problem can be solved by constructing a Lagrangian function and converting it into a dual problem, using a sequential minimum optimization algorithm to obtain the hyperplane ω used for classification T After F+b=0, for the input feature vector F j , its category can be determined by the following principles:

[0068] yj =+1 if f(F j )=ω T F+b>0

[0069] y j =-1 if f(F j )=ω T F+b≤0

[0070] In the above formula, f(F j )=ω T F+b is the decision function obtained by training. So far, this paper has constructed a new three-feature joint classifier based on nonlinear SVM, which can classify high / low sea conditions and sea clutter at the unit level; in fact, since the applicability of the three features mentioned in this paper is closely related to the actual radial distance corresponding to the sea clutter distance unit, in the actual classification process, it is necessary to train different classifiers at different distance segments, and then classify high / low sea conditions at different distance segments; it should be noted that SVM, as the classifier used in this embodiment, performs well in the classification of high / low sea conditions, and other embodiments can also use random forests or deep neural networks and other machine learning or deep learning algorithms suitable for target binary classification as classifiers to achieve the classification of high / low sea conditions under sea clutter.

[0071] This embodiment provides a classifier for classifying sea clutter data, which is used to classify the sea clutter data based on target features to obtain a classification result of the sea clutter data.

[0072] Combination Figure 4 A sea condition classification device based on radar echo provided in an embodiment of the present invention is described. The sea condition classification device based on radar echo described below and the sea condition classification method based on radar echo characteristics described above can correspond to each other.

[0073] The present invention provides a sea condition classification device based on radar echo, the device comprising:

[0074] The data acquisition module 401 is used to acquire sea clutter data to be classified; the feature extraction module 402 is used to preprocess the sea clutter data to obtain preprocessing parameters, and extract target features from the preprocessing results according to preset feature extraction rules; the classification module 403 inputs the sea clutter data based on the target features into a pre-stored classifier for classification to obtain classification results.

[0075] The present invention provides a sea condition classification device based on radar echo. The data acquisition module 401 is used to first acquire sea clutter data to be classified. The feature extraction module 402 is used to preprocess the sea clutter data to acquire preprocessing parameters. The target feature is extracted from the preprocessing result according to a preset feature extraction rule. Finally, the sea clutter data based on the target feature is input into a prestored classifier for classification through a classification module 403 to acquire a classification result. The device of the present invention improves the preprocessing method of the sea clutter data, acquires more effective features from the sea clutter data to realize the classification of high / low sea condition sea clutter, and improves the classification accuracy.

[0076] Figure 5 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530 and a communication bus 540, wherein the processor 510, the communication interface 520 and the memory 530 communicate with each other through the communication bus 540. The processor 510 may call the logic instructions in the memory 530 to execute a sea condition classification method based on radar echo characteristics, the method comprising: obtaining sea clutter data to be classified; preprocessing the sea clutter data to obtain preprocessing parameters, and extracting target features from the preprocessing results according to preset feature extraction rules; inputting the sea clutter data based on the target features into a pre-stored classifier for classification to obtain a classification result.

[0077] In addition, the logic instructions in the above-mentioned memory 530 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0078] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it is implemented to execute a sea condition classification method based on radar echo characteristics provided by the above methods, the method comprising: acquiring sea clutter data to be classified; preprocessing the sea clutter data to obtain preprocessing parameters, and extracting target features from the preprocessing results according to preset feature extraction rules; inputting the sea clutter data based on the target features into a pre-stored classifier for classification to obtain a classification result.

[0079] The present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute a sea condition classification method based on radar echo characteristics provided by the above methods, the method comprising: acquiring sea clutter data to be classified; preprocessing the sea clutter data to obtain preprocessing parameters, and extracting target features from the preprocessing results according to preset feature extraction rules; inputting the sea clutter data based on the target features into a pre-stored classifier for classification to obtain a classification result.

[0080] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0081] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0082] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A sea condition classification method based on radar echo characteristics, characterized in that: include: Acquire sea clutter data to be classified; Preprocessing the sea clutter data to obtain preprocessing parameters, and extracting target features from the preprocessing results according to preset feature extraction rules; Extracting target features from the preprocessing result according to a preset feature extraction rule includes: Based on the ridge integral theory, the preprocessing parameters are quantized and summed to obtain the target features, wherein the target features include the ridge integral of the mean function and the ridge integral of the standard deviation function; The sea clutter data is preprocessed to obtain preprocessing parameters, and target features are extracted from the preprocessing results according to a preset feature extraction rule, followed by: Calculate the regional peak power and noise mean power of the sea clutter data respectively; Calculating a ratio of the regional peak power to the noise mean power, and constructing the target feature based on the ridge integral of the mean function, the ridge integral of the standard deviation function, and the ratio; The regional peak power is expressed as: The noise mean power is expressed as: Wherein, X1(p) represents the power spectrum of the peak area, p = 1, 2, ..., P, the position of X1(p) can be determined by the position of the peak of the Doppler power spectrum, and the size of the peak area P is determined by the 3dB spectrum width of the Doppler power spectrum of the sea clutter in high sea conditions; X2(k) represents the noise area of ​​the power spectrum, k = 1, 2, ..., K, and the range of X2(k) is the area outside the range of 3 times the peak area of ​​the Doppler power spectrum of the sea clutter in high sea conditions; The sea clutter data based on the target features is input into a pre-stored classifier for classification to obtain a classification result.

2. The sea condition classification method based on radar echo characteristics according to claim 1 is characterized in that: Preprocessing the sea clutter data to obtain preprocessing parameters includes: The sea clutter data is converted from a one-dimensional random sequence in the time domain to a two-dimensional random sequence in the time-frequency domain by using a time-frequency transformation tool, and preprocessing parameters representing the two-dimensional random sequence are obtained, wherein the preprocessing parameters include a mean function and a standard deviation function.

3. The sea condition classification method based on radar echo characteristics according to claim 1 is characterized in that: The classifier includes any one of short-time Fourier transform, Wigner-Ville distribution and smoothed pseudo WVD.

4. The sea condition classification method based on radar echo characteristics according to claim 1 is characterized in that: The classifier includes any one of the machine learning or deep learning algorithms suitable for target binary classification, such as SVM, random forest and deep neural network.

5. A sea condition classification device based on radar echo, applied to the sea condition classification method based on radar echo characteristics as claimed in claim 1, characterized in that: The device comprises: A data acquisition module, used for acquiring sea clutter data to be classified; A feature extraction module, used for preprocessing the sea clutter data to obtain preprocessing parameters, and extracting target features from the preprocessing results according to preset feature extraction rules; The classification module inputs the sea clutter data based on the target features into a pre-stored classifier for classification to obtain a classification result.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the sea condition classification method based on radar echo characteristics as described in any one of claims 1 to 4 is implemented.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the sea condition classification method based on radar echo characteristics as claimed in any one of claims 1 to 4 is implemented.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the sea condition classification method based on radar echo characteristics as claimed in any one of claims 1 to 4 is implemented.

Images