Millimeter wave water surface flow velocity measurement radar CFAR detection method and system

Abstract

The invention discloses a millimeter wave water surface flow velocity measurement radar CFAR detection method and system, and the method comprises the steps: eliminating an outlier in advance through a local outlier factor, carrying out the estimation of the quartile distance of a sample through the screened data, obtaining a cut-off threshold value of the sample, and finally obtaining a background clutter power estimation value through the maximum likelihood estimation. According to the method, the selection of the quartile optimization truncation threshold is introduced, and the reference window sample is screened in advance by using the local outlier factor, so that the robustness of the quartile and the truncation threshold is enhanced; according to the method, the environmental adaptability of the algorithm is enhanced, the target detection performance of the algorithm in a multi-target and clutter edge environment is effectively improved, and a feasible new scheme is provided for target detection of the millimeter wave water surface flow velocity measurement radar.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar target detection, and in particular relates to a millimeter wave water surface flow velocity measurement radar CFAR detection method and system. Background Art

[0002] Target detection, a critical step in millimeter-wave surface velocity measurement radar signal processing, is crucial for overall radar performance. Its primary goal is to accurately extract key parameters such as the target's position and velocity from the echo signal. However, non-uniform clutter, both natural and artificial, often exists in the background environment. These factors can cause dramatic fluctuations in background power, severely impacting target detection reliability.

[0003] In radar signal processing, target detection algorithms are a core component that determines radar performance, with constant false alarm (CFAR) algorithms playing a crucial role. Over the years, researchers at home and abroad have proposed a variety of CFAR algorithms based on the calculation of background clutter power. These algorithms can be broadly categorized into three types: mean-level (ML), ordered statistics (OS), and other adaptive CFAR algorithms. Mean-level algorithms estimate background clutter power by calculating the average value of samples within a reference window. This method is simple and offers good target detection performance in uniform environments. However, they are susceptible to strong interference in multi-target environments, significantly degrading detection performance. To address the poor environmental adaptability of mean-level CFAR algorithms, the ordered statistics CFAR (OS-CFAR) algorithm was proposed. This algorithm sorts samples within a reference window and selects the kth smallest value as the background clutter power estimate. This algorithm effectively suppresses the influence of strong interference within the reference window, maintaining good detection performance in multi-target environments. However, its detection performance is significantly affected by the k value. If the k value is too small, the false alarm probability in strong clutter areas may increase; if the k value is too large, the target detection probability may be reduced in multi-target environments. A single type of CFAR algorithm often fails to achieve excellent detection performance in all scenarios. Therefore, adaptive CFAR algorithms that take into account the characteristics of different scenarios have been widely used. Among them, the Truncated Statistics CFAR (TS-CFAR) algorithm based on truncated statistics eliminates interference values by setting a truncation threshold and using maximum likelihood estimation to estimate the background clutter power of the remaining samples. This method can obtain an adaptive threshold value, improving detection performance in multi-target situations. Summary of the Invention

[0004] The object of the present invention is to provide a millimeter wave water surface flow velocity measurement radar CFAR detection method and system.

[0005] The technical solution for achieving the purpose of the present invention is: a millimeter wave water surface flow velocity measurement radar CFAR detection method, comprising the following steps:

[0006] Step 1: Remove outliers through local outlier factors to eliminate the influence of strong interference samples on the estimated values of sample parameters;

[0007] Step 2: Obtain sample parameter estimates based on the remaining sample data;

[0008] Step 3: Use the interquartile range to perform unilateral truncation of the sample data to control strong interference data and remove potential outliers;

[0009] Step 4: Perform maximum likelihood estimation on the truncated data and use the obtained scale parameter estimate as the background clutter power estimate.

[0010] A millimeter-wave water surface flow velocity measurement radar CFAR detection system is used to implement the above method, and the system includes:

[0011] The first module removes outliers through local outlier factors and eliminates the influence of strong interference samples;

[0012] The second module obtains the estimated value of the sample parameters based on the remaining sample data after elimination;

[0013] In the third module, the interquartile range is used to truncate the sample data unilaterally, control the interference data, and remove potential outliers;

[0014] The fourth module is used to perform maximum likelihood estimation on the truncated data and use the obtained scale parameter estimation value as the background clutter power estimation value.

[0015] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the steps of the above method are implemented when the processor executes the program.

[0016] A computer-readable storage medium stores a computer program, which implements the steps of the above method when executed by a processor.

[0017] A computer program product comprises a computer program, which implements the steps of the above method when executed by a processor.

[0018] Compared with the prior art, the present invention has the following significant advantages: (1) The present invention optimizes the selection of the truncation threshold by using the interquartile range. Thanks to the support of statistical theory, the threshold setting of the algorithm has a clearer physical meaning, overcomes the shortcoming that the performance of the traditional algorithm is limited by the selection of the truncation factor, improves the environmental adaptability of the truncation threshold, and enhances the detection performance in multi-target and clutter edge environments; (2) The present invention uses the local outlier factor to eliminate some outliers in advance, enhances the robustness of the interquartile range and the truncation threshold, and further improves the detection performance of the algorithm in non-uniform environments; (3) The present invention performs maximum likelihood estimation on the remaining samples after truncation, improves the credibility of subsequent clutter power estimation, and reduces the constant false alarm loss of the algorithm.

[0019] The present invention is described in further detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 This is the structural block diagram of the LQTS-CFAR algorithm.

[0021] Figure 2 This is a comparison chart of LQTS-CFAR detection probabilities for three interference situations on one side of the target.

[0022] Figure 3 This is a comparison chart of LQTS-CFAR detection probabilities for three interference situations on both sides of the target.

[0023] Figure 4 This is a comparison chart of LQTS-CFAR detection probability in the case of clutter edge.

[0024] Figure 5 This is a comparison chart of the false alarm probability of LQTS-CFAR in the case of clutter edge. DETAILED DESCRIPTION

[0025] The present invention discloses a millimeter-wave water surface velocity measurement radar CFAR detection method and system, which can realize real-time and accurate water surface velocity measurement. Specifically, outliers are eliminated in advance by using local outlier factors, and the sample parameters are estimated using the screened sample data to obtain the interquartile range and truncation threshold of the sample distribution, and finally the background clutter power estimation value is obtained by maximum likelihood estimation. The improved truncated statistical CFAR algorithm based on the interquartile range proposed in the present invention introduces the interquartile range to optimize the selection of the truncation threshold, and uses the local outlier factor to screen the reference window samples in advance, thereby enhancing the robustness of the interquartile range and the truncation threshold. The algorithm enhances the environmental adaptability of the truncation threshold, further improves the target detection capability of the algorithm in a non-uniform environment, and provides a feasible new solution for target detection of millimeter-wave water surface velocity measurement radar.

[0026] like Figure 1As shown, a millimeter wave water surface flow velocity measurement radar CFAR detection method includes the following steps:

[0027] Step 1: Eliminate outliers through local outlier factors to effectively eliminate the influence of strong interference samples on the sample parameter estimates. The specific steps are:

[0028] Step 1-1: For the reference window sample data x i , whose k-distance is defined as point x i The distance to its kth nearest neighbor is:

[0029] k-distance(x i )=distance(x i ,o k )

[0030] Among them, k It is point x i The kth nearest neighbor of i ,o k ) represents point x i With dot o k The distance between them, based on k-distance, can construct point x i The k-neighborhood of is defined as:

[0031] N k (x i )={o∈(Dx i )|distance(x i ,o)≤k-distance(x i )}

[0032] Among them, D represents the entire data set, N k (x i ) is point x i The set of all points in the neighborhood of whose distance is less than or equal to k-distance;

[0033] Step 1-2: In order to reduce the impact of isolated points or outliers on density estimation, the local outlier factor introduces the concept of reachable distance to correct the direct distance between points. i The reachable distance to point o is defined as:

[0034] Reachability Distance k (x i ,o)=max(k-distance(o),distance(x i ,o))

[0035] This definition ensures that the distance value will not be lower than the k-distance of point o, thus avoiding the significant impact of low-density points on local density calculation;

[0036] Steps 1-3: Local reachability density is used to measure point x i The density level in its neighborhood is calculated as:

[0037]

[0038] Among them, |N k (x i )| is the number of points in the k-neighborhood. The larger the LRD value, the closer the point x i The higher the density of the surrounding area, the less likely it is an outlier; conversely, the smaller the LRD value, the more likely point x is to be an outlier. i The lower the local density, the more likely it is an outlier;

[0039] Steps 1-4: Local outlier factor passes through point x i The ratio of the local reachability density of the point to the average local reachability density of its neighborhood points is calculated. i The degree of outlier is expressed as:

[0040]

[0041] If LOF k (x i )≈1, then point x i The density of LOF is close to the average density of its neighborhood, which is a normal point. k (x i )>1, then point x i The density of LOF is significantly lower than the neighborhood density and is an outlier. k (x i )=1.4 as the decision threshold, if LOF k (x i )≥1.4, it is considered an abnormal value, otherwise it is considered a normal value.

[0042] Step 2: Use the sample set X after filtering by the local outlier factor H , perform maximum likelihood estimation on the scale parameter of the clutter distribution, that is in is the estimated value of the scale parameter, H is the number of remaining samples, x i is the value of the i-th reference sample.

[0043] Step 3: Use the interquartile range to perform unilateral truncation of the sample data, which can focus on controlling strong interference data and effectively remove potential outliers. The specific steps are as follows:

[0044] Step 3-1, calculate the quantile function of the sample, the formula is:

[0045]

[0046] Where q is the position of the quantile and c is the shape parameter;

[0047] Step 3-2, obtain the quartiles of the sample;

[0048] 1) First quartile Q1: quantile function corresponding to q = 0.25;

[0049] 2) Second quartile Q2: quantile function corresponding to q = 0.50;

[0050] 3) The third quartile Q3: corresponds to the quantile function with q = 0.75;

[0051] Step 3-3, calculate the interquartile range (IQR) of the sample;

[0052] IQR=Q3-Q1

[0053] Step 3-4: Use the interquartile range to construct the threshold boundary of the sample data. Points exceeding the boundary are judged as outliers. The threshold boundary is defined as:

[0054] T h =Q3+β·IQR

[0055] Where β is the interquartile range empirical coefficient.

[0056] Step 4: Use the remaining samples after truncation to perform maximum likelihood estimation to obtain the estimated value of background clutter power, that is:

[0057]

[0058] in is the sample truncated by the interquartile range, I is the number of retained samples, c is the shape parameter, T q =(-ln(1-q1)) 1 / c +β[(-ln(1-q1)) 1 / c -(-ln(1-q2)) 1 / c ],q1=0.75,q2=0.25.

[0059] Based on the same inventive concept, the present invention also provides a millimeter-wave water surface flow velocity measurement radar CFAR detection system, comprising:

[0060] The first module removes outliers through local outlier factors, which can effectively eliminate the influence of strong interference samples;

[0061] The second module obtains the estimated value of the sample parameters based on the remaining sample data after elimination;

[0062] The third module uses the interquartile range to perform unilateral truncation of the sample data, which can focus on controlling strong interference data and effectively remove potential outliers;

[0063] The fourth module is used to perform maximum likelihood estimation on the truncated data and use the obtained scale parameter estimation value as the background clutter power estimation value.

[0064] The specific implementation methods of the above-mentioned first to fourth modules are the same as the steps in the aforementioned millimeter-wave water surface flow velocity measurement radar CFAR detection method, and will not be repeated here.

[0065] The present invention uses the interquartile range to optimize the selection of the truncation threshold. Thanks to the support of statistical theory, the threshold setting of the algorithm has a clearer physical meaning. The local outlier factor is used to screen the reference window samples in advance, which enhances the robustness of the interquartile range and truncation threshold, further improving the algorithm's target detection capability in non-uniform environments, and providing a feasible new solution for target detection in millimeter-wave water surface velocity measurement radar.

[0066] The following is a detailed description with reference to the embodiments:

[0067] Example

[0068] The specific conditions of this embodiment are: the clutter obeys the Weibull distribution, the scale parameter b=2, the shape parameter c=2, the total number of range cells is 600, the length of the left and right protection cells is 2, the length of the left and right reference cells is N=16, and the false alarm probability is set to P fa = = 10 -4 ; The sample number of OS-CFAR is k = 24; the interquartile range coefficient β = 1.5.

[0069] First, the target is placed in the 300th distance unit. The first case is that there are three strong interferences on one side of the target. The interferences are placed in the 305th, 310th and 315th units, and the detection probability comparison chart of LQTS-CFAR is as follows: Figure 2 As shown in the figure, it can be seen that the LQTS-CFAR algorithm has the best performance. When SCR = 7dB, its detection probability can reach 87%, followed by the TS-CFAR algorithm, whose detection probability is 83.62%, which is significantly higher than other comparison algorithms; the second case is that there are three interferences on both sides of the target. The interferences are placed in the 285th, 290th, 295th, 305th, 310th, and 315th distance units, and the detection probability comparison diagram of LQTS-CFAR is as follows Figure 3 As shown in the figure, it can be seen that the LQTS-CFAR algorithm can still maintain high detection performance when the number of interferences is large.

[0070] Then, the clutter edge is set at the 300th distance unit, 0~300 is the weak clutter area with an average power of 22dB, 301~600 is the strong clutter area with an average power of 42dB. The target is placed at the 290th distance unit, and the detection probability comparison chart of LQTS-CFAR in the clutter edge environment is as follows: Figure 4 As shown in the figure, it can be seen that when SCR = 8dB, the detection probability of LQTS-CFAR is 88.36%, while that of TS-CFAR is only 6.44%, which is significantly better than the other comparison algorithms.

[0071] No target is set, and the false alarm probability is set to 10 -2 , the rest of the settings are the same as above, and each distance unit is 10 6 The Monte Carlo simulation results show that the false alarm probability of the 270-330th distance segment is as follows: Figure 5 As shown in the figure, the LQTS-CFAR algorithm maintains the best false alarm performance in the weak clutter zone before the 300th range unit, with the rate only beginning to decline significantly at the 292nd range unit, effectively avoiding the "target obscuration effect." In practical applications, clutter edges are primarily distributed along the river surface and riverbanks. Compared to the highly reflective nature of the riverbank, the river surface typically exhibits weak clutter. Therefore, optimizing detection performance in weak clutter zones plays a crucial role in improving radar target detection capabilities. It can effectively enhance target detectability in clutter-edge environments, thereby improving the system's overall detection performance.

[0072] As can be seen from the above examples, the millimeter-wave surface flow velocity radar (CFAR) detection method of the present invention utilizes the interquartile range to optimize the selection of the truncation threshold, giving the algorithm's threshold setting a clearer physical meaning. It also utilizes the local outlier factor to preemptively eliminate some outliers, enhancing the robustness of the interquartile range and truncation threshold. Simulation results demonstrate that LQTS-CFAR overcomes the shortcomings of traditional algorithms, where performance is limited by the selection of the truncation factor. This improves the environmental adaptability of the truncation threshold, further enhances the algorithm's detection performance in non-uniform environments, and mitigates the "target obscuration effect" in weak clutter areas, significantly enhancing the algorithm's environmental adaptability.

Claims

1. A millimeter wave water surface flow velocity measurement radar CFAR detection method, characterized in that: The following steps are involved: Step 1: Remove outliers through local outlier factors to eliminate the influence of strong interference samples on the estimated values of sample parameters; Step 2: Obtain sample parameter estimates based on the remaining sample data; Step 3: Use the interquartile range to perform unilateral truncation of the sample data to control strong interference data and remove potential outliers; Step 4: Perform maximum likelihood estimation on the truncated data and use the obtained scale parameter estimate as the background clutter power estimate.

2. The millimeter wave water surface flow velocity measurement radar CFAR detection method according to claim 1 is characterized in that: The specific steps for removing outliers using the local outlier factor in step 1 are: Step 1-1: For the reference window sample data x i , whose k-distance is defined as point x i The distance to its kth nearest neighbor is: k-distance(x i )=distance(x i ,o k ) Among them, k It is point x i The kth nearest neighbor of i ,o k ) represents point x i With dot o k The distance between them, based on k-distance, constructs point x i The k-neighborhood of is defined as: N k (x i )={o∈(D-x i )|distance(x i ,o)≤k-distance(x i )} Among them, D represents the entire data set, N k (x i ) is point x i The set of all points in the neighborhood of whose distance is less than or equal to k-distance; Step 1-2: The local outlier factor introduces the concept of reachable distance to correct the direct distance between points. Point x i The reachable distance to point o is defined as: Reachability Distance k (x i ,o)=max(k-distance(o),distance(x i ,o)) Steps 1-3: Local reachability density is used to measure point x i The density level in its neighborhood is calculated as: Among them, |N k (x i )| is the number of points in the k-neighborhood; the larger the LRD value, the closer the point x i The higher the density of the surrounding area, the less likely it is an outlier; conversely, the smaller the LRD value, the more likely point x is to be an outlier. i The lower the local density, the more likely it is an outlier; Steps 1-4: Local outlier factor passes through point x i The ratio of the local reachability density of the point to the average local reachability density of its neighborhood points is calculated. i The degree of outlier is expressed as:

3. The millimeter wave water surface flow velocity measurement radar CFAR detection method according to claim 2 is characterized in that: Select LOF k (x i )=1.4 as the decision threshold, if LOF k (x i )≥1.4, it is considered an abnormal value, otherwise it is considered a normal value.

4. The millimeter wave water surface flow velocity measurement radar CFAR detection method according to claim 1, characterized in that: The sample set X after screening by the local outlier factor in step 2 H , perform maximum likelihood estimation on the scale parameter of the clutter distribution, that is in is the estimated value of the scale parameter, H is the number of remaining samples, x i is the value of the i-th reference sample.

5. The millimeter wave water surface flow velocity measurement radar CFAR detection method according to claim 4 is characterized in that: The specific steps for using the interquartile range to perform unilateral truncation of sample data in step 3 are: Step 3-1, calculate the quantile function of the sample, the formula is: Where q is the position of the quantile and c is the shape parameter; Step 3-2, obtain the quartiles of the sample; 1) First quartile Q1: quantile function corresponding to q = 0.25; 2) Second quartile Q2: quantile function corresponding to q = 0.50; 3) The third quartile Q3: corresponds to the quantile function with q = 0.75; Step 3-3, calculate the interquartile range (IQR) of the sample; IQR=Q3-Q1 Step 3-4: Use the interquartile range to construct the threshold boundary of the sample data. Points exceeding the boundary are judged as outliers. The threshold boundary is defined as: T h =Q3+β·IQR Where β is the interquartile range empirical coefficient.

6. The millimeter wave water surface flow velocity measurement radar CFAR detection method according to claim 5, characterized in that: Step 4 uses the remaining samples after truncation to perform maximum likelihood estimation to obtain the estimated value of background clutter power, that is, in is the sample truncated by the interquartile range, I is the number of retained samples, c is the shape parameter, T q =(-ln(1-q1)) 1 / c +β[(-ln(1-q1)) 1 / c -(-ln(1-q2)) 1 / c ],q1=0.75,q2=0.

25.

7. A millimeter wave water surface flow velocity measurement radar CFAR detection system, used to implement the method according to any one of claims 1 to 6, characterized in that the system include: The first module removes outliers through local outlier factors and eliminates the influence of strong interference samples; The second module obtains the estimated value of the sample parameters based on the remaining sample data after elimination; In the third module, the interquartile range is used to truncate the sample data unilaterally, control the interference data, and remove potential outliers; The fourth module is used to perform maximum likelihood estimation on the truncated data and use the obtained scale parameter estimation value as the background clutter power estimation value.

8. 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 steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images