A method for detecting small low-altitude targets on the sea surface based on characteristic clutter images
By using the characteristic clutter graph method in radar detection, small targets and sea clutter are extracted and distinguished from low-altitude sea surfaces, the problem of low detection performance of radar in sea surface environments is solved, and more efficient target detection and false alarm control are achieved.
Patent Information
- Application Number
- CN202210194463.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-03-01
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2042-03-01
AI Technical Summary
In the detection of low-altitude small targets on the sea surface, the radar faces the clutter interference caused by the small RCS, the slow flight speed, the low flight altitude and the complex sea surface environment, resulting in a degradation of detection performance. The traditional frequency domain filtering method is not applicable, and the machine learning method is not subject to samples. The influence of balanced and singular characteristics, the false alarm rate is high.
Using a detection method based on feature clutter graph, the radar echo data under the background of pure sea clutter, extract the feature matrix and iteratively update it, establish the feature clutter graph, and combine the feature clutter graph judgment criteria to distinguish the target and sea clutter wave, control detection probability and false alarm probability.
It improves the accuracy and performance of low-altitude small target detection on the sea surface, reduces the problem of missing target detection and false alarm, and the dimensions of the characteristic clutter graph are controllable, with high storage space efficiency, avoiding the problem of unbalanced sample.
Smart Images

Figure CN114518564B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of radar signal processing, and in particular relates to a method for detecting small low-altitude targets on the sea surface based on a characteristic clutter map. Background Art
[0002] Radar faces many problems when detecting small targets at low altitudes on the sea surface: 1. The targets detected by radar have the characteristics of small RCS, slow flight speed, and low flight altitude; 2. The sea surface environment is complex, and small targets at low altitudes will be submerged in strong sea clutter, resulting in a serious decline in the radar's detection performance of the target; 3. The real-time requirements for radar detection are very high, requiring the radar to have real-time detection capabilities within each CPI (coherent processing period).
[0003] In traditional radar signal processing, target detection in the sea clutter background is generally achieved through frequency domain filtering. However, due to the non-stationary, non-Gaussian, and time-varying characteristics of sea clutter, the Doppler spectrum of sea clutter is severely broadened, and the measurement of the center of the clutter spectrum and the estimation of its spectrum width become a difficult problem. Therefore, traditional frequency domain filtering methods such as adaptive moving target indication (AMTI) and moving target detection (MTD) are no longer applicable.
[0004] The traditional clutter map is a clutter intensity distribution map within the radar power range stored in the memory. A complete clutter map is formed after one antenna scan. Generally, it takes several to dozens of antenna scan cycles to establish a relatively stable clutter map. The clutter map method is suitable for target detection in the clutter background of fixed objects, but it is no longer applicable to sea clutter with severe Doppler broadening and clutter spectrum center offset. In addition, due to the influence of the clutter map dimension and storage space, the intensity clutter map has high requirements for storage space and is no longer applicable to high-resolution and high-data-volume radars.
[0005] Peng-Lang Shui et al. published a paper titled "Tri-feature-based detection of floating small targets in sea clutter" in IEEE Transactions on Aerospace and Electronic Systems in 2014. The paper disclosed a method for extracting sea surface targets based on a three-feature detector. The method transforms the target detection problem under the background of sea clutter into a classification problem, introduces the three-dimensional convex hull algorithm into the feature classification of targets and sea clutter, and forms a discrimination space between pure sea clutter and target plus sea clutter by extracting effective three-dimensional features of sea clutter and targets, and then discriminates whether the echo to be detected belongs to the pure sea clutter class or the target plus sea clutter class. This type of method has obvious advantages in detecting small targets at low altitude on the sea surface, but the method can only be used in a three-dimensional feature space. In order to expand the feature dimension, the team introduced machine learning methods such as support vector machines, K-nearest neighbors, and neural networks into the classification problem of high-dimensional feature space. The method has better robustness and detection effect, but the use of machine learning methods first requires obtaining a large number of training samples. For small targets at low altitude on the sea surface detected by radar, the number of target samples that can be obtained is small, and sample acquisition is difficult. At the same time, when training feature detectors, machine learning methods will be affected by individual singular sample characteristics and cannot effectively control the false alarm rate, which reduces the radar's detection probability of small low-altitude targets on the sea surface and increases the false alarm probability. Summary of the invention
[0006] The technical purpose of the present invention is to provide a method for detecting small low-altitude targets on the sea surface based on characteristic clutter patterns to solve the problems of the prior art.
[0007] To solve the above problems, the technical solution of the present invention is:
[0008] A method for detecting small low-altitude targets on the sea surface based on characteristic clutter images comprises the following steps:
[0009] S1: Acquire radar echo data within several coherent processing periods CPI under pure sea clutter background, perform rearrangement, pulse compression and data feature extraction in sequence according to the pulse repetition period PRT, and iterate the feature matrices of several CPIs to obtain the feature clutter map matrix under pure sea clutter background;
[0010] S2: Acquire the echo data to be measured within a coherent processing period CPI, perform rearrangement, pulse compression and data extraction features in sequence according to PRT, and obtain the echo feature matrix to be measured;
[0011] S3: Perform clutter map judgment processing based on the characteristic clutter map matrix and the characteristic matrix of the echo to be measured, and extract the distance unit information where the target to be measured is located;
[0012] S4: Measure the speed and angle of the target based on the distance unit information.
[0013] Preferably, before step S1, the following steps are also included:
[0014] The forgetting factor of the characteristic clutter map is preset to w, the number of iterations of the characteristic clutter map is preset to X, and K features are preset to distinguish the target to be measured from the clutter. For the first Q features, the eigenvalue of the target to be measured is greater than the eigenvalue of the clutter, and for the remaining KQ features, the eigenvalue of the target to be measured is less than the eigenvalue of the clutter, and the characteristic clutter map matrix C is initialized based on the above preset parameters. K×N It is a zero matrix with K rows and N columns, where N is the number of sampling points in one pulse repetition period PRT of the radar echo data, corresponding to N different distance units of the target to be measured.
[0015] Wherein, step S1 specifically includes the following steps:
[0016] S11: Obtain radar echo data S within a coherent processing cycle CPI under pure sea clutter background r (t), t is time;
[0017] S12: radar echo data S r (t) Rearrange according to the pulse repetition period PRT to obtain the pure sea clutter echo data matrix S with M rows and N columns r [t r (m),t a (n)], where t r is the slow time dimension, m=1,2,…,M, M is the number of PRTs in a CPI, t a is the fast time dimension, n=1,2,…,N;
[0018] S13: Pure sea clutter echo data matrix S r [t r (m),t a (n)] Pulse compression is performed along the fast time dimension to obtain the pure sea clutter echo data matrix after pulse compression.
[0019] S14: Pure sea clutter echo data matrix after pulse compression Extract K features from the slow time dimension data corresponding to each distance unit, sort the corresponding eigenvalues of the K features in sequence, and obtain a feature matrix with K rows and N columns Based on the feature matrix For the characteristic clutter matrix C K×NPerform iterative update, the formula is
[0020]
[0021] S15: Determine whether the number of iterations is less than X. If so, repeat steps S11 to S15. Otherwise, obtain the characteristic clutter map matrix C under the pure sea clutter background. K×N .
[0022] Wherein, step S2 specifically includes the following steps:
[0023] S21: Acquire the echo data S to be measured within a coherent processing period CPI d (t), t is time;
[0024] S22: echo data to be measured S d (t) Rearrange according to the pulse repetition period PRT to obtain the measured echo data matrix S with M rows and N columns d [t r (m),t a (n)];
[0025] S23: Echo data matrix S to be measured d [t r (m),t a (n)] Pulse compression is performed along the fast time dimension to obtain the echo data matrix to be measured after pulse compression.
[0026] S24: Echo data matrix to be measured after pulse pressure Extract K features from the slow time dimension data corresponding to each distance unit, sort the corresponding eigenvalues of the K features in sequence, and obtain the K-row and N-column echo feature matrix F K×N .
[0027] Wherein, step S3 specifically includes the following steps:
[0028] S31: Based on the detection probability and false alarm probability of the radar detecting the target to be detected, a clutter map detection feature number threshold L is set, L≤K, and threshold factors T1 and T2 are set, wherein the threshold factor T1 corresponds to the first Q rows of the characteristic clutter map matrix, T1≥1, and the threshold factor T2 corresponds to the KQ rows of the characteristic clutter map matrix, T2≤1;
[0029] S32: Detecting feature number threshold L, threshold factors T1 and T2 and feature clutter map matrix C based on clutter map K×N and the echo characteristic matrix F K×N Establish target decision criteria for characteristic clutter images;
[0030] S33: Comparison of characteristic clutter matrix C K×N and the echo characteristic matrix FK×N The corresponding value of is extracted based on the target decision criterion of the characteristic clutter map to obtain the distance unit information where the target is located.
[0031] Among them, the target judgment criterion of the characteristic clutter map is as follows:
[0032] Characteristic clutter matrix C K×N and the echo characteristic matrix F K×N , among the K features corresponding to each column, for the first Q features, the feature number whose value extracted from the echo data to be measured is greater than the product of the corresponding value of the echo data to be measured and the threshold factor T1 is set as A, and for the remaining KQ features, the feature number whose value extracted from the echo data to be measured is less than the product of the corresponding value of the echo data to be measured and the threshold factor T2 is set as B. If A+B≥L, it is considered that the echo feature matrix F K×N There is a target in the distance cell corresponding to this column.
[0033] Specifically, the features used to distinguish the target and clutter include: relative average amplitude, relative Doppler entropy, relative Doppler spectrum variation coefficient, relative Doppler peak height, relative phase linearity, relative energy concentration, relative DC component proportion, relative Doppler spectrum width, relative waveform factor, relative margin factor, Doppler spectrum kurtosis, Doppler spectrum skewness, peak factor, ridge accumulation, number of connected areas, and maximum connected area size.
[0034] Due to the adoption of the above technical solution, the present invention has the following advantages and positive effects compared with the prior art:
[0035] 1) The present invention extracts effective distinguishing features between targets and sea clutter and establishes a characteristic clutter map under a pure sea clutter background. Through inter-frame accumulation, the clutter map gradually tends to be stable, so that the discrimination between targets and sea clutter is not affected by singular eigenvalues. Moreover, compared with the traditional intensity clutter map, the dimension of the characteristic clutter map is controllable and the storage space occupied is small.
[0036] 2) Compared with the existing three-feature detector, the feature dimension extracted by the present invention can be expanded to multiple dimensions, and by establishing a clutter map target decision criterion, the detection probability and false alarm probability are effectively controlled, the detection performance is guaranteed, and the target missed detection and false alarm problems caused by the different sensitivities of different data to different features are reduced.
[0037] 3) The present invention establishes a characteristic clutter map by extracting only the radar echo features under the pure sea clutter background. Compared with the method of using a machine learning algorithm to extract a large number of target and sea clutter sample features and train the decision space of the target and sea clutter, the pure sea clutter samples are relatively easy to obtain. The method of the present invention does not have the problem of imbalance of target and sea clutter training samples, and improves the detection probability of low-altitude small targets under the sea clutter background. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the following detailed description of the preferred embodiment.The drawings are only for the purpose of illustrating the preferred embodiments and are not to be construed as limiting the invention.
[0039] Figure 1 It is a schematic diagram of the process of the method for detecting small targets at low altitude on the sea surface based on characteristic clutter images of the present invention;
[0040] Figure 2 is a schematic diagram for comparing relative Doppler spectrum variation coefficients obtained by implementing the present invention;
[0041] Figure 3 It is a schematic diagram for comparing relative margin factors obtained based on the implementation of the present invention;
[0042] Figure 4 is a schematic diagram of the comparison of relative Doppler spectrum widths obtained based on the implementation of the present invention;
[0043] Figure 5 is a schematic diagram of the comparison of relative Doppler spectral entropy obtained based on the implementation of the present invention;
[0044] Figure 6 The figure is a schematic diagram for comparing the relative DC component proportions obtained based on the implementation of the present invention. DETAILED DESCRIPTION
[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the specific implementation methods of the present invention will be described below with reference to the accompanying drawings. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings and other implementation methods can be obtained based on these drawings without creative work.
[0046] In order to simplify the drawings, only the parts related to the present invention are schematically shown in each figure, and they do not represent the actual structure of the product. In addition, in order to simplify the drawings and facilitate understanding, in some figures, only one of the parts with the same structure or function is schematically drawn or marked. In this article, "one" not only means "only one", but also means "more than one".
[0047] The following is a further detailed description of a method for detecting small targets at low altitudes over the sea surface based on characteristic clutter patterns proposed by the present invention in conjunction with the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description and claims.
[0048] Example
[0049] See also Figure 1This embodiment provides a method for detecting small low-altitude targets on the sea surface based on a characteristic clutter map, comprising the following steps:
[0050] First, before executing the steps of this embodiment, the following steps are also included
[0051] The forgetting factor of the characteristic clutter map is preset to w, and the number of iterations of the characteristic clutter map is preset to X. The number of iterations X of the wave map can be selected according to the actual application scenario. The larger the X, the more stable the clutter map formed, but the time consumed to form the clutter map increases. K features are preset to distinguish the target to be measured from the clutter. For the first Q features, the characteristic value of the target to be measured is larger than the characteristic value of the clutter. For the remaining KQ features, the characteristic value of the target to be measured is smaller than the characteristic value of the clutter. The optional feature range includes but is not limited to the following features: relative average amplitude, relative Doppler entropy, relative Doppler spectrum variation coefficient, relative Doppler peak height, relative phase linearity, relative energy concentration, relative DC component proportion, relative Doppler spectrum width, relative waveform factor, relative margin factor, Doppler spectrum kurtosis, Doppler spectrum skewness, peak factor, ridge accumulation, number of connected regions, maximum connected region size, etc.
[0052] Then, the characteristic clutter map matrix C is initialized based on the above preset parameters K×N It is a zero matrix with K rows and N columns, where N is the number of sampling points in one pulse repetition period PRT of the radar echo data, corresponding to N different distance units of the target to be measured.
[0053] In this embodiment, taking the following scheme as an example, the clutter map forgetting factor w=7 / 8 is set, the number of iterations of the characteristic clutter map of each wave position is X=20, and K=5 features are selected to distinguish the target and clutter, which are relative Doppler variation coefficient, relative margin factor, relative Doppler spectrum width, relative Doppler spectrum entropy and relative DC component proportion. The eigenvalues of the target of the first two features are larger than the eigenvalues of the clutter, and the eigenvalues of the target of the remaining three features are smaller than the eigenvalues of the clutter, that is, Q=2, and the characteristic clutter map matrix C is initialized. K×N It is a 5-row, N-column all-zero matrix, where N represents the number of fast time dimension sampling points of the radar echo signal, corresponding to different distance units. The definition and specific description of the distinguishing features of the selected target and sea clutter are as follows:
[0054] 1) Relative Doppler spectrum variation coefficient:
[0055] The relative Doppler variation coefficient can distinguish targets and sea clutter by the sharpness of the Doppler spectrum peak. The Doppler spectrum peak of the target range unit is relatively concentrated, while the Doppler spectrum peak of the sea clutter range unit is relatively scattered. Suppose the number of range units in a CPI of radar echo data is N, the number of pulses of coherent accumulation is M, and the radar echo data of each range unit is xm (m=1,2,…,M), after discrete Fourier transform, the Doppler amplitude spectrum X m =DFT(x m ), the coefficient of variation of the Doppler spectrum is calculated as follows:
[0056]
[0057] The relative Doppler spectrum variation coefficient is defined as:
[0058]
[0059] R is the number of reference units. To eliminate the influence of main lobe broadening and side lobes, a protection unit may be added between the unit to be tested and the reference unit. In this embodiment, the number of reference units R=5 is selected.
[0060] 2) Relative Doppler spectrum width:
[0061] Affected by the sea environment, wind, gravity and other factors, the sea surface scatterers are always in motion, and the motion states of each scattering unit are not completely consistent, which makes the Doppler spectrum of sea clutter severely broadened. For moving targets, their motion characteristics remain basically unchanged within the resolution unit, and the Doppler broadening is small. Therefore, the relative Doppler spectrum width can be used as a feature to distinguish targets from sea clutter.
[0062] The normalized Doppler shift obtained using the centroid method can be expressed as:
[0063]
[0064] where f m (m=1,2,…,M) is the normalized Doppler frequency. The normalized Doppler bandwidth can be calculated by the following formula:
[0065]
[0066] The relative Doppler spectral width is defined as:
[0067]
[0068] 3) Relative margin factor:
[0069] Margin factor is often used to detect impact signals in signals and can describe the extreme degree of the signal peak in the waveform. Therefore, the margin factor can distinguish the target to be measured from the sea clutter by the extreme degree of the Doppler spectrum peak. According to the scattering characteristics of the target to be measured and the sea clutter, the margin factor of the target to be measured is usually greater than the margin factor of the sea clutter.
[0070] The margin factor is the ratio of the signal peak value to the RMS amplitude, and is calculated as follows:
[0071]
[0072] The relative margin factor is defined as:
[0073]
[0074] 4)Relative Doppler entropy:
[0075] By analyzing the characteristics of sea clutter, the Doppler amplitude spectrum of pure sea clutter is more dispersed than the Doppler amplitude spectrum of the target plus sea clutter. Entropy can describe the degree of chaos of the system or data, and the Doppler vector entropy can be used to determine whether the target exists.
[0076] Doppler entropy is defined as:
[0077]
[0078]
[0079] represents the normalized Doppler amplitude spectrum, and the relative Doppler entropy is defined as:
[0080]
[0081] 5)Relative DC component ratio:
[0082] The Doppler spectrum of sea clutter received by the radar will be greatly broadened, and due to wind and wave movement, the center frequency of the spectrum often deviates from the zero frequency, that is, the DC component in the sea clutter is relatively small, while the Doppler of the target echo is relatively concentrated, and its DC component is relatively large. Calculate the DC power and average power of each distance unit in the radar echo data:
[0083]
[0084]
[0085] Thus, the proportion of DC component of each resolution unit is obtained:
[0086]
[0087] The relative DC component ratio is defined as:
[0088]
[0089] Then see Figure 1, enter step S1, obtain radar echo data within several coherent processing periods CPI under the pure sea clutter background, rearrange, pulse compress and extract data features according to the pulse repetition period PRT in turn, and iteratively update the feature matrices of several CPIs to obtain the feature clutter map matrix under the pure sea clutter background.
[0090] The details are as follows
[0091] S11: Obtain radar echo data S within a coherent processing cycle CPI under pure sea clutter background r (t), t is time.
[0092] S12: radar echo data S within a CPI r (t) Rearrange according to the pulse repetition period PRT to obtain the pure sea clutter echo data matrix S with M rows and N columns r [t r (m),t a (n)], where t r is the slow time dimension, m=1,2,…,M, M is the number of PRTs in a CPI, t a is the fast time dimension, n=1,2,…,N, corresponding to different distance units respectively.
[0093] S13: Pure sea clutter echo data matrix S r [t r (m),t a (n)] Pulse compression is performed along the fast time dimension to obtain the pure sea clutter echo data matrix after pulse compression.
[0094] S14: Pure sea clutter echo data matrix after pulse compression Extract K features from the slow time dimension data corresponding to each distance unit to obtain a feature matrix with K rows and N columns Sort the corresponding eigenvalues of the K features in sequence, based on the feature matrix For the characteristic clutter matrix C K×N Perform iterative update, the formula is
[0095]
[0096] In this embodiment, each column of the echo data matrix after pulse compression is the slow time dimension data corresponding to a distance unit. The values of the five features described above are extracted from the slow time dimension data corresponding to each distance unit to obtain a feature matrix with 5 rows and N columns. The superscript x represents the characteristic matrix of the xth CPI, x≤X, and the characteristic matrix describes the characteristics of radar echo units at different distances in a CPI under the background of pure sea clutter. The characteristic clutter map matrix is iteratively updated. In this embodiment, the characteristic clutter map forgetting factor w=7 / 8 is selected. Therefore, the iterative formula of the characteristic clutter map is as follows:
[0097]
[0098] This formula indicates that the characteristic clutter map matrix of the current frame is obtained by multiplying a weight coefficient by the characteristic clutter map matrix formed in the previous frame and adding the characteristic matrix of the current frame multiplied by a weight coefficient. With the iterative update of the characteristic clutter map, the characteristic clutter map gradually tends to be stable.
[0099] S15: Determine whether the number of iterations is less than X. In this embodiment, X=20 is set. If the number of iterations is less than 20, repeat steps S11 to S15. If it is equal to 20, the characteristic clutter map matrix C under the pure sea clutter background is obtained. 5×N to save.
[0100] Then, in step S2, the echo data to be measured within a coherent processing period CPI is obtained, and PRT rearrangement, pulse compression and data feature extraction are performed in sequence to obtain a feature matrix of the echo to be measured.
[0101] The specific steps include:
[0102] S21: The radar receives the echo data S to be measured within a coherent processing period CPI d (t), t is time.
[0103] S22: echo data to be measured S d (t) Rearrange according to the pulse repetition period PRT to obtain the measured echo data matrix S with M rows and N columns d [t r (m),t a (n)].
[0104] S23: Echo data matrix S to be measured d [t r (m),t a (n)] Pulse compression is performed along the fast time dimension to obtain the echo data matrix to be measured after pulse compression.
[0105] S24: Echo data matrix to be measured after pulse pressure Extract K features from the slow time dimension data corresponding to each distance unit, sort the corresponding eigenvalues of the K features in sequence, and obtain the K-row and N-column echo feature matrix F K×N , the extracted K features and the extraction method are as described above.
[0106] Then, the process proceeds to step S3 to perform clutter map decision processing based on the characteristic clutter map matrix and the characteristic matrix of the echo to be measured, and extract the distance unit information where the target to be measured is located.
[0107] Specifically, S31: Based on the detection probability and false alarm probability of the target to be detected by radar, a clutter map detection feature number threshold L is set, L≤K. The larger the feature number threshold L is, that is, the more features of the feature clutter map that pass the threshold are required, the lower the false alarm probability is, and the smaller L is, the higher the detection probability of the target is. However, since different data have different sensitivities to different features, the value of L needs to be analyzed according to the actual application scenario. Threshold factors T1 and T2 are set, where threshold factor T1 corresponds to the first Q rows of the feature clutter map matrix, T1≥1, and threshold factor T2 corresponds to the KQ rows of the feature clutter map matrix, T2≤1. The selection of T1 and T2 values can be dynamically adjusted according to the actual scenario.
[0108] In this embodiment, the clutter pattern detection feature number threshold L is set to 3, and T1 and T2 are set to 1.5 and 0.9 respectively.
[0109] S32: Detecting feature number threshold L, threshold factors T1 and T2 and feature clutter map matrix C based on clutter map K×N and the echo characteristic matrix F K×N The characteristic clutter map target decision criterion is established. The characteristic clutter map target decision criterion is specifically the characteristic clutter map matrix C K×N and the echo characteristic matrix F K×N , among the K features corresponding to each column, for the first Q features, the feature number whose value extracted from the echo data to be measured is greater than the product of the corresponding value of the echo data to be measured and the threshold factor T1 is set as A, and for the remaining KQ features, the feature number whose value extracted from the echo data to be measured is less than the product of the corresponding value of the echo data to be measured and the threshold factor T2 is set as B. If A+B≥L, it is considered that the echo feature matrix F K×N There is a target in the distance cell corresponding to this column.
[0110] S33: Comparison of characteristic clutter matrix C K×N and the echo characteristic matrix F K×N The corresponding value of is extracted based on the target decision criterion of the characteristic clutter map to obtain the distance unit information where the target is located. According to step S1 and step S2, if the values of three features among the five features extracted from a certain distance unit meet the target decision criterion of the characteristic clutter map, it is considered that there is a target in the distance unit, and then the distance information of the target can be extracted.
[0111] Since the characteristic clutter pattern detection method can only extract the distance information of the target, for the detection of small targets at low altitude on the sea surface, it is usually necessary to extract the speed and angle information of the target. Therefore, step S4 is finally required to measure the speed and angle of the target to be measured based on the distance unit information. The speed measurement process can be performed by the Doppler filter group on the distance unit signal where the target is located, and the angle measurement process can be performed by using the sum-difference ratio method.
[0112] The technical effects of the present invention are further described below in combination with measured data.
[0113] The method described in the present invention is verified by using the measured data of a Ku-band shore-based offshore radar. First, after the radar is turned on, the echo data under the pure sea clutter background is obtained, and the characteristic values of different distance units are extracted from the echo data according to the method described in the embodiment of the present invention, and the characteristic clutter map matrix is iteratively established and stored; then the radar starts to detect the target, and for the echo data to be measured received by the radar, the characteristic value corresponding to each distance unit is extracted respectively, and compared with the corresponding position of the stored characteristic clutter map matrix. If the characteristic clutter map target judgment criterion described in the present invention is met, it is considered that there is a target in the distance unit.
[0114] See also Figures 2 to 6 , the experiment sets the clutter map forgetting factor w = 7 / 8; the number of iterations of the characteristic clutter map of each wave position is X = 20; selects K = 5 features for distinguishing targets and clutter, which are relative Doppler variation coefficient, relative margin factor, relative Doppler spectrum width, relative Doppler spectrum entropy and relative DC component proportion. The number of reference units R = 5; the threshold of the number of clutter map detection features L = 3; the threshold factors T1 = 1.5 and T2 = 0.9 are set, and the method described in this embodiment is applied. The results are as follows Figures 2 to 6 shown.
[0115] The clutter image characteristic value threshold and the characteristic value of the echo to be measured are compared. Figure 2 As shown, Figure 2 , 3 They correspond to the characteristic relative Doppler spectrum variation coefficient and the relative margin factor respectively. It can be seen from the figure that the characteristic value corresponding to the target position is greater than the characteristic clutter map threshold value, while the characteristic values corresponding to the other clutter positions are less than the characteristic clutter map threshold value, so the target can be distinguished; Figure 4 , 5 , 6 correspond to the characteristic relative Doppler spectrum width, relative Doppler spectrum entropy and relative DC component proportion respectively. It can be seen from the figure that the characteristic value corresponding to the target position is less than the characteristic clutter map threshold, while the characteristic values corresponding to the other clutter positions are greater than the characteristic clutter map threshold, and the total number of features at the target position is greater than 3, so it can be determined that there is a target at this position.
[0116] The embodiments of the present invention are described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above embodiments. Even if various changes are made to the present invention, if these changes fall within the scope of the claims of the present invention and their equivalents, they still fall within the protection scope of the present invention.
Claims
1. A method for detecting small low-altitude targets on the sea surface based on characteristic clutter patterns, characterized in that: The steps include: S1: Acquire radar echo data within several coherent processing periods CPI under pure sea clutter background, perform rearrangement, pulse compression and data feature extraction in turn according to the pulse repetition period PRT, and iteratively update the feature matrices of several CPIs to obtain the feature clutter map matrix under pure sea clutter background; S2: Obtain the echo data to be measured within a coherent processing period CPI , and then rearrange, compress and extract data features according to PRT to obtain the echo feature matrix to be measured. ; S3: Based on the characteristic clutter matrix and the characteristic matrix of the echo to be measured Perform clutter map judgment processing to extract the distance unit information where the target to be measured is located; S4: performing speed and angle measurement processing on the target to be measured based on the distance unit information; Wherein, the step S1 also includes the following steps: The forgetting factor of the characteristic clutter map is preset to , the number of iterations of the characteristic clutter map is preset to , preset selection The features used to distinguish the target from the clutter are features, the eigenvalue of the target to be measured is greater than the clutter eigenvalue, and the remaining The characteristic value of the target to be measured is less than the clutter characteristic value, and the characteristic clutter map matrix is initialized based on the above preset parameters for OK A matrix with all zero columns, is the number of sampling points in a pulse repetition period PRT of the radar echo data, corresponding to the target to be measured Different distance units; Wherein, the step S1 specifically includes the following steps: S11: Acquire radar echo data within one coherent processing period CPI under pure sea clutter background , For time; S12: The radar echo data Rearrange according to the pulse repetition period PRT, and we get OK The pure sea clutter echo data matrix ,in, For the slow time dimension, , is the number of PRTs within a CPI, For fast time maintenance, ; S13: The pure sea clutter echo data matrix Pulse compression processing is performed along the fast time dimension to obtain the pure sea clutter echo data matrix after pulse compression ; S14: The pure sea clutter echo data matrix after pulse compression The slow time dimension data corresponding to each distance unit is extracted Features, yes The corresponding eigenvalues of the features are sorted in sequence, and we get OK The characteristic matrix of the columns , based on the feature matrix The characteristic clutter matrix Perform iterative update, the formula is ; S15: Determine whether the number of iterations is less than If yes, repeat steps S11 to S15; otherwise, obtain the characteristic clutter map matrix under the pure sea clutter background. .
2. The method for detecting small low-altitude targets on the sea surface based on characteristic clutter patterns according to claim 1, characterized in that: The step S2 specifically includes the following steps S21: Acquire the echo data to be measured within a coherent processing period CPI , For time; S22: The echo data to be measured Rearrange according to the pulse repetition period PRT, and we get OK Columns of echo data matrix to be measured ; S23: The echo data matrix to be measured Perform pulse compression processing along the fast time dimension to obtain the echo data matrix to be measured after pulse compression ; S24: the echo data matrix to be measured after pulse pressure The slow time dimension data corresponding to each distance unit is extracted Features, yes The corresponding eigenvalues of the features are sorted in sequence, and we get OK The measured echo characteristic matrix .
3. The method for detecting small low-altitude targets on the sea surface based on characteristic clutter patterns according to claim 2 is characterized in that: The step S3 specifically includes the following steps S31: Setting the clutter image detection feature number threshold based on the detection probability and false alarm probability of the radar detecting target , set the threshold factor and , where the threshold factor Corresponding to the characteristic clutter map matrix OK, , the threshold factor Corresponding to the characteristic clutter map matrix OK, ; S32: Detect feature number threshold based on clutter map , the threshold factor and And the characteristic clutter map matrix and the echo characteristic matrix to be measured Establish target decision criteria for characteristic clutter images; S33: Compare the characteristic clutter map matrix and the echo characteristic matrix to be measured The corresponding value of is extracted based on the target decision criterion of the characteristic clutter image to obtain the distance unit information where the target is located.
4. The method for detecting small low-altitude targets on the sea surface based on characteristic clutter patterns according to claim 3 is characterized in that: The characteristic clutter image target decision criterion is specifically: The characteristic clutter map matrix and the echo characteristic matrix to be measured , each column corresponds to Among the features, the former A feature is set, and the value of the feature extracted from the echo data to be measured is set to be greater than the corresponding value of the echo data to be measured and the threshold factor The characteristic number of the product is , the remaining The value of the feature extracted from the echo data to be measured is set to be smaller than the corresponding value of the echo data to be measured and the threshold factor The characteristic number of the product is ,like , then it is considered that the characteristic matrix of the echo to be measured There is a target in the distance cell corresponding to this column.
5. The method for detecting small low-altitude targets on the sea surface based on characteristic clutter patterns according to claim 1, characterized in that: The features used to distinguish the target to be measured from the clutter include: relative average amplitude, relative Doppler entropy, relative Doppler spectrum variation coefficient, relative Doppler peak height, relative phase linearity, relative energy concentration, relative DC component proportion, relative Doppler spectrum width, relative waveform factor, relative margin factor, Doppler spectrum kurtosis, Doppler spectrum skewness, peak factor, ridge accumulation, number of connected areas, and maximum connected area size.
Citation Information
Cited By
Target detection method based on background noise estimation self-updating
CN121703779A
