Low false alarm rate detection method and device suitable for low slow small target detection

Abstract

The invention discloses a low false alarm rate detection method and device suitable for low slow small target detection, and the method comprises the steps: collecting trace point data containing low slow small targets, and selecting a trace point set of unmanned plane targets from the trace point data; fitting the relationship between the actual amplitude and the actual distance of each trace point in the trace point set by using a least square method to obtain a fitting curve; in the fitting process, the actual amplitude of each trace point in the trace point set serves as a dependent variable, and the actual distance of each trace point in the trace point set serves as an independent variable; calculating the residual error of each trace point in the trace point set according to the fitting curve and the actual curve; the distribution of all residual errors meets normal distribution; calculating the coefficient of the detection threshold curve according to the normal distribution characteristics of all residual errors; designing a detection threshold curve according to coefficients of the fitting curve and the detection threshold curve; and performing constant false alarm detection by using the detection threshold curve. According to the invention, the target points and miscellaneous points in the low-slow small target can be distinguished more accurately, and the false alarm probability of the radar is reduced.

Description

Technical Field

[0001] The present invention belongs to the technical field of radar signal processing, and in particular relates to a low false alarm rate detection method and device suitable for detecting slow and small targets. Background Art

[0002] As a new type of low-altitude aircraft, drones are small, highly concealed, and highly flexible. Such low-altitude, slow-moving small targets are often difficult to capture with traditional radars. In addition, the low-altitude environment has strong interference and complex application scenarios. Therefore, when using radar to detect and track such targets, some noise is often misjudged as targets, which affects the accuracy of the radar system and increases the workload of the radar system, thereby affecting the real-time performance of the radar system.

[0003] The task of radar signal detection is to process the mixed signal output by the receiver, which is often composed of target signals, noise and other interference, and accurately identify the target signal with a certain detection probability, while keeping the misjudgment probability of noise and other interference signals at a low level. The size of the detection probability and false alarm probability depends on the amplitude distribution of the target signal and the characteristics of the noise and interference signals. However, the traditional constant false alarm detection method usually has difficulty in effectively distinguishing the noise generated by low, slow and small targets and clutter noise, and it is also difficult to remove noise and noise, which affects the subsequent radar data processing and tracking of targets.

[0004] The definition of false alarm probability is the probability that the radar believes there is a target in the received signal when there is no target in the signal. Constant false alarm detection technology adjusts the detection threshold to ensure that the false alarm probability remains constant in different signal environments. However, if the detection threshold is set too low, noise may be misjudged as a target; if it is set too high, the real target may be missed. The most basic constant false alarm processing is to estimate the noise and interference in a signal that needs to be detected, and set the threshold according to the estimated value to determine whether there is a target in the signal. The common constant false alarm detection methods are:

[0005] (1) Cell Averaging CFAR (CA-CFAR)

[0006] CA-CFAR uses multiple noise reference cells on both sides of the target cell to calculate the mean of the background noise to determine the detection threshold. Specifically, CA-CFAR first selects a protection cell interval on the left and right of the target cell to ensure that the noise cannot be directly affected by the target cell. Then, several reference cells are selected outside the protection cell, and the signal amplitudes of these reference cells are averaged to generate an estimate of the background noise. Next, the estimated noise level is multiplied by a constant factor and set as the target detection threshold. If the signal amplitude of the target cell exceeds this threshold, it is determined that the target exists.

[0007] (2) Greatest Of CFAR (GO-CFAR)

[0008] GO-CFAR sets the detection threshold by selecting the one with a larger noise level from multiple reference groups. In the specific steps, the two sides of the target unit will be divided into two sets of reference units, and the signal strength means in each set will be calculated respectively. Subsequently, when calculating the threshold, GO-CFAR selects the maximum value of the two sets of signal strength means as the noise background estimate to set the detection threshold. This strategy ensures that when detecting isolated targets, confusion from neighboring strong interfering targets can be effectively suppressed. Although GO-CFAR has good target isolation capabilities in environments with uneven backgrounds, it may have the risk of missing weak targets in dense target groups.

[0009] (3) Smallest Of CFAR (SO-CFAR)

[0010] SO-CFAR calculates the detection threshold by selecting the signal with smaller strength in the noise reference group. During operation, the reference units on both sides of the target unit are divided into two groups, and the signal strength mean is calculated for each group of units. After that, the minimum of the two means is selected as the background noise estimate. Compared with GO-CFAR, SO-CFAR is more suitable for use in multi-target environments, because selecting the minimum value as the threshold can effectively reduce the probability of false alarms caused by adjacent targets. However, when the clutter background is strong or the noise is uneven, this choice may lead to incorrect background estimation, which in turn affects the overall detection performance and needs to be adjusted according to actual needs.

[0011] (4) Ordered Statistics CFAR (OS-CFAR)

[0012] OS-CFAR is a sorting-based method that aims to improve adaptability to environmental noise changes and the ability to suppress isolated noise points. The principle is to sort the reference unit signal amplitudes outside the target unit area from small to large. A fixed position in the sorting, such as the median or the front position signal value, is selected as the noise background estimate. The sorting operation makes the noise estimate more stable, and can effectively filter out isolated strong noise points or outliers without affecting the final detection threshold. OS-CFAR is often used in environments with uneven background noise, and its calculation amount is large.

[0013] However, since environmental noise and clutter will affect the propagation of radar signals, resulting in a decrease in signal quality, the above-mentioned existing constant false alarm detection method can remove some noise points, but it still cannot remove noise points well when facing low, slow and small targets in complex environments, and it usually sets an absolute threshold. Such an absolute threshold cannot remove most of the noise points when detecting low, slow and small targets, resulting in a large number of noise points being reported to the radar system, which requires the radar system to process a large number of points, consumes a lot of computing resources and leads to insufficient real-time performance. Summary of the invention

[0014] In order to solve the above problems existing in the prior art, the present invention provides a low false alarm rate detection method and device suitable for low, slow and small target detection. The technical problem to be solved by the present invention is achieved by the following technical solutions:

[0015] In a first aspect, an embodiment of the present invention provides a low false alarm rate detection method suitable for detecting slow and small targets, including:

[0016] Collecting point trace data containing low, slow and small targets, and selecting a point trace set of the UAV target from all the point trace data; each point trace in the point trace set includes an actual amplitude and an actual distance;

[0017] Using the least square method, fitting the relationship between the actual amplitude and the actual distance to obtain a fitting curve; wherein, during the fitting process, the actual amplitude of each point trace in the point trace set is used as a dependent variable, and the actual distance of each point trace in the point trace set is used as an independent variable to construct a fitting curve;

[0018] Calculating the residual of each point in the point trace set according to the fitting curve and the actual curve corresponding to the point trace set; wherein the distribution of all residuals satisfies the normal distribution;

[0019] Calculate the coefficients of the detection threshold curve based on the normal distribution characteristics of all residuals;

[0020] Designing a detection threshold curve according to the fitting curve and the coefficients of the detection threshold curve;

[0021] The detection threshold curve is used to perform constant false alarm detection.

[0022] In one embodiment of the present invention, the coefficients of the detection threshold curve are calculated according to the normal distribution characteristics of all residuals, including:

[0023] Calculating the fitting amplitude of each point trace in the point trace set according to the fitting curve;

[0024] Calculate the residual of each point trace of the point trace set according to the fitting amplitude and the actual amplitude;

[0025] Calculating an average residual according to the residual of each point trace in the point trace set;

[0026] Calculate the standard deviation of the residual according to the residual of each point trace in the point trace set and the average residual;

[0027] According to the preset expected threshold, the probability distribution table of the standard normal distribution is queried to obtain the width of the confidence interval;

[0028] The coefficient of the detection threshold curve is calculated according to the standard deviation of the residual and the width of the confidence interval.

[0029] In one embodiment of the present invention, the formula for calculating the standard deviation of the residual is:

[0030]

[0031] Among them, σ represents the standard deviation of the residual, n represents the number of points in the point set, and r i Represents the residual of the ith point in the point set, r i =y i -f(x i ), y i Indicates the actual amplitude of the i-th point in the point set, x i represents the actual distance of the ith point in the point set, f(x i ) represents the actual distance x calculated according to the fitting curve i The fitting amplitude at Represents the mean of the residual errors of all points in the set of points.

[0032] In one embodiment of the present invention, the formula for calculating the coefficient of the detection threshold curve is:

[0033]

[0034] Among them, A represents the coefficient of the detection threshold curve, n represents the number of points in the point set, and x i represents the actual distance of the ith point in the point set, f(x i) represents the actual distance x calculated according to the fitting curve i , σ represents the standard deviation of the residuals, and z represents the width of the confidence interval.

[0035] In one embodiment of the present invention, when the preset expected threshold is 98%, the width of the confidence interval is -2.05.

[0036] In one embodiment of the present invention, the designed detection threshold curve formula is expressed as:

[0037] f′(x)=A·f(x);

[0038] Where f′(x) represents the detection threshold curve, A represents the coefficient of the detection threshold curve, f(x) represents the fitting curve, x represents the set of actual distances of the points in the point set, and x = {x1,…,x i ,…,x n}, x i Indicates the actual distance of the i-th point in the point set.

[0039] In one embodiment of the present invention, using the detection threshold curve to perform constant false alarm detection includes:

[0040] For each point in the point set, the execution steps include: using the detection threshold curve to calculate the estimated amplitude of the point, judging whether the actual amplitude of the point is greater than the estimated amplitude, if so, judging that the point is a target point, if not, judging that the point is a noise point, and removing the point from the point set.

[0041] In a second aspect, an embodiment of the present invention provides a low false alarm rate detection device suitable for detecting slow and small targets, the device comprising:

[0042] The acquisition module is used to collect point trace data containing low, slow and small targets, and select a point trace set of the UAV target from all the point trace data; each point trace in the point trace set includes an actual amplitude and an actual distance;

[0043] A fitting module, used to fit the relationship between the actual amplitude and the actual distance using the least square method to obtain a fitting curve; wherein, during the fitting process, the actual amplitude of each point trace in the point trace set is used as a dependent variable, and the actual distance of each point trace in the point trace set is used as an independent variable to construct a fitting curve;

[0044] A first calculation module, used for calculating the residual of each point trace in the point trace set according to the fitting curve and the actual curve corresponding to the point trace set; wherein the distribution of all residuals satisfies the normal distribution;

[0045] A second calculation module is used to calculate the coefficient of the detection threshold curve according to the normal distribution characteristics of all residuals;

[0046] A design module, used for designing a detection threshold curve according to the fitting curve and the coefficients of the detection threshold curve;

[0047] The detection module is used to perform constant false alarm detection using the detection threshold curve.

[0048] Beneficial effects of the present invention:

[0049] The low false alarm rate detection method suitable for detecting slow and small targets proposed by the present invention is simple in implementation but can detect the correct target while reducing the probability of noise points passing through the detection, thereby reducing the false alarm probability of the radar, and specifically: by statistically analyzing a large number of measured point trace data of unmanned aerial vehicle targets, the relationship between the distance and the amplitude is fitted using the least square method; in the constant false alarm detection, the absolute threshold calculated according to the noise mean value is replaced with a detection threshold curve designed according to the distance and amplitude characteristics of the unmanned aerial vehicle target, and the design idea of ​​the detection threshold curve is to use the relationship between the residual and normal distribution of the actual value of the unmanned aerial vehicle target point and the fitting value, so as to ensure the preset expected threshold value, such as a target detection probability of more than 98%, to obtain the coefficient of the detection threshold curve, and the target detection curve is designed by the fitting curve and the coefficient of the detection threshold curve, so as to use the target detection curve for constant false alarm detection. In summary, the detection method proposed in the present invention can more accurately distinguish between low, slow and small target points and noise, reduce the false alarm probability of the radar, and have a good inhibitory effect on radar clutter; reduce the influence of environmental clutter on radar detection of low, slow and small targets, reduce the load of the radar system, and enhance the real-time and accuracy of the system. It has great application prospects for the detection and tracking of low, slow and small targets such as drones in low-altitude environments, such as urban monitoring and airspace management.

[0050] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a flow chart of a low false alarm rate detection method suitable for detecting slow and small targets provided by an embodiment of the present invention;

[0052] Figure 2 It is a schematic diagram showing all the points in the point trace set of the drone target provided by an embodiment of the present invention in terms of azimuth and distance;

[0053] Figure 3 It is a schematic diagram showing all the points in the point trace set provided by an embodiment of the present invention in terms of amplitude and distance;

[0054] Figure 4is a schematic diagram of a fitting curve provided by an embodiment of the present invention;

[0055] Figure 5 is a schematic diagram of the distribution of all residuals corresponding to the point trace set provided in an embodiment of the present invention;

[0056] Figure 6 is a schematic diagram of a detection threshold curve provided by an embodiment of the present invention;

[0057] Figure 7 (a)~ Figure 7 (b) is a comparison diagram of the detection method using an absolute threshold and the detection method proposed by the present invention in a scenario where a drone is vertically flying upwards provided by an embodiment of the present invention;

[0058] Figure 8 (a)~ Figure 8 (b) is a comparison diagram of the detection method using an absolute threshold and the detection method proposed by the present invention in the S-shaped direct flight scenario of the drone provided by an embodiment of the present invention;

[0059] Fig. 9 (a)~ Fig. 9 (b) is a comparative diagram of the track display on the radar terminal of the same UAV provided by an embodiment of the present invention at different times and in different flight postures in the same environment, using the absolute threshold detection method and the detection method proposed by the present invention;

[0060] Fig.10 It is a structural schematic diagram of a low false alarm rate detection device suitable for detecting slow and small targets provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0061] The present invention is further described in detail below with reference to specific embodiments, but the embodiments of the present invention are not limited thereto.

[0062] First, see Figure 1 The embodiment of the present invention provides a low false alarm rate detection method suitable for detecting slow and small targets, which specifically includes the following steps:

[0063] S10, collecting point trace data containing low, slow and small targets, and selecting a point trace set of the UAV target from all the point trace data; each point trace in the point trace set includes an actual amplitude and an actual distance.

[0064] Usually, after the radar transmission signal encounters the target in space, the reflected echo signal is received by the receiving antenna, and multiple continuous echo signals are usually combined into a frame of data for signal processing. For example, when using radar to detect drone targets during flight, the radar echo signal is collected by limiting the drone to a certain distance and speed range, and then the collected radar echo signal is processed. The signal processing includes pulse compression, moving target detection, constant false alarm detection, angle measurement, etc., to collect point data containing low, slow and small targets. Here, low, slow and small targets are low-altitude and slow small targets, usually referring to a type of small aircraft with a flight altitude below 1,000 meters, a flight speed below 200 kilometers per hour (56 meters per second), and a radar reflection cross-section RCS of less than 2 square meters.

[0065] Since the movement of the UAV is coherent and the distance and speed information of the UAV target are available during the point trace data collection, the point trace set of the UAV target can be obtained from the point trace data of the radar echo signal after signal processing. Figure 2 As shown, the information of the drone's target point can be obtained according to the continuity of the drone's motion.

[0066] S20. Use the least squares method to fit the relationship between the actual amplitude and the actual distance to obtain a fitting curve; wherein, during the fitting process, the actual amplitude of each point trace in the point trace set is used as the dependent variable, and the actual distance of each point trace in the point trace set is used as the independent variable to construct the fitting curve.

[0067] After obtaining the point set of the UAV target from S10, the relationship between the signal amplitude and distance of the UAV target is fitted using the least squares method. The least squares method is a mathematical optimization technique commonly used for data fitting. It finds the optimal function by minimizing the sum of squares of the errors between the actual points and the fitting curve.

[0068] During the fitting process, the actual amplitude of each point trace in the point trace set of the embodiment of the present invention is used as the dependent variable, and the actual distance of each point trace in the point trace set is used as the independent variable to obtain a set of points (x1, y1), (x2, y2), ..., (x n ,y n ), {x1,…,x i ,…,x n} constitutes the set of actual distances of the points in the point set, {y1,…,y i ,…,y n} constitutes the set of actual amplitudes of the points in the point set, such as Figure 3 Assume that the function expression of the fitting curve is:

[0069] f(x)=p1x n +p2xn-1 +p3x n-2 +…+p n x+p n+1 (1);

[0070] Where x represents the set of actual distances of the points in the point set, x = {x1,…,x i ,…,x n}, f(x) represents the fitting curve. According to the fitting curve f(x), the fitting amplitude at each actual distance in x can be calculated, {p1, p2, …, p n+1} represents the polynomial coefficient, and n represents the number of points in the point set.

[0071] For any point in the point set, the residual calculation between it and the fitting curve is expressed as:

[0072] r i =y i -f(x i ) (2);

[0073] Among them, r i represents the residual of the ith point in the point set, x i Indicates the actual distance of the i-th point in the point set, y i represents the actual amplitude of the ith point in the point set, f(x i ) represents the actual distance x calculated according to the fitting curve i The fitting amplitude at .

[0074] The goal of the least squares method is to minimize the sum of squares S of all residuals:

[0075]

[0076] Next, the fitting curve f(x) is solved. Here, we take f(x) as a cubic polynomial as an example: According to the number of traces n in the trace set, and the actual amplitude and actual distance of each trace in the trace set, the process matrix X and process matrix Y are constructed:

[0077]

[0078]

[0079] To minimize the sum of squares S of all residuals, the least squares solution can be obtained by solving the following normal equation:

[0080] X T Xp=X T Y (6);

[0081] Where p is the vector of polynomial coefficients, p = [p1 p2 p3 p4] T , [] T represents the transposition operation; by solving formula (6), the polynomial coefficients can be obtained, and then the fitting curve represented by formula (1) can be determined as follows: Figure 4 shown.

[0082] S30, calculating the residual of each point in the point trace set according to the fitting curve and the actual curve corresponding to the point trace set; wherein the distribution of all residuals satisfies the normal distribution.

[0083] In the process of determining the fitting curve in S20, the residual of each point in the point trace set can also be calculated according to the fitting curve and the actual curve corresponding to the point trace set. The specific residual calculation method is shown in formula (2), which will not be repeated here.

[0084] The residual is the deviation between the actual curve and the fitted curve. These target points are obtained by processing the radar echo signal of the drone by the same radar, and there are certain errors in them. These errors may be caused by multiple independent noise sources such as measurement error, system error and environmental noise. According to the central limit theorem, as the number of samples increases, the distribution of residuals will approach the normal distribution. Figure 5 Therefore, the embodiment of the present invention considers that the distribution of the residuals approximately satisfies the normal distribution.

[0085] S40, calculating the coefficients of the detection threshold curve according to the normal distribution characteristics of all residuals.

[0086] In the embodiment of the present invention, the coefficients of the detection threshold curve are calculated according to the normal distribution characteristics of all residuals, including:

[0087] The fitting amplitude of each point in the point trace set is calculated according to the fitting curve; the residual of each point in the point trace set is calculated according to the fitting amplitude and the actual amplitude; the average residual is calculated according to the residual of each point in the point trace set; the standard deviation of the residual is calculated according to the residual of each point in the point trace set and the average residual; the probability distribution table of the standard normal distribution is queried according to the preset expected threshold value to obtain the width of the confidence interval; the coefficient of the detection threshold curve is calculated according to the standard deviation of the residual and the width of the confidence interval.

[0088] The formula for calculating the standard deviation of the residual in the embodiment of the present invention is:

[0089]

[0090] Among them, σ represents the standard deviation of the residual, n represents the number of points in the point set, and r i Represents the residual of the ith point in the point set, r i =y i -f(x i), y i Indicates the actual amplitude of the i-th point in the point set, x i represents the actual distance of the ith point in the point set, f(x i ) represents the actual distance x calculated according to the fitting curve i The fitting amplitude at Represents the mean of the residual errors of all points in the set of points.

[0091] The formula for calculating the coefficient of the detection threshold curve in the embodiment of the present invention is:

[0092]

[0093] Among them, A represents the coefficient of the detection threshold curve, n represents the number of points in the point set, and x i represents the actual distance of the ith point in the point set, f(x i ) represents the actual distance x calculated according to the fitting curve i The fitting amplitude at , σ represents the standard deviation of the residual, and z represents the width of the confidence interval. The width of the confidence interval z is determined by the actual detection expectation. For example, when the preset expected threshold is 98%, that is, it is expected that 98% of the target points can pass the detection threshold curve, the probability distribution table of the standard normal distribution is queried, and the width of the confidence interval is -2.05.

[0094] S50: Design a detection threshold curve according to the fitting curve and the coefficients of the detection threshold curve.

[0095] The embodiment of the present invention designs the detection threshold curve for constant false alarm detection based on the fitting curve determined in S20 and the coefficient of the detection threshold curve calculated in S40. Figure 6 As shown, the formula is expressed as:

[0096] f′(x)=A·f(x) (9);

[0097] Where f′(x) represents the detection threshold curve, A represents the coefficient of the detection threshold curve, f(x) represents the fitting curve, x represents the set of actual distances of the points in the point set, and x = {x1,…,x i ,…,x n}, x i Indicates the actual distance of the i-th point in the point set.

[0098] S60: Perform constant false alarm detection using the detection threshold curve.

[0099] In the embodiment of the present invention, the detection threshold curve is used to perform constant false alarm detection, including: for each point trace in the point trace set, the steps include: using the detection threshold curve to calculate the estimated amplitude of the point trace, judging whether the actual amplitude of the point trace is greater than the estimated amplitude, if so, judging that the point trace is a target point, if not, judging that the point trace is a noise point, and removing the point trace from the point trace set. More specifically:

[0100] The embodiment of the present invention no longer uses an absolute threshold to screen the target points, but uses a detection threshold curve f′(x) designed according to the actual scene to screen the points from the point set, such as Figure 5 As shown. Determine whether a point is a target point:

[0101] The actual distance and actual amplitude of the point trace are obtained (x i ,y i ), x i is the actual distance, y i is the actual amplitude; substitute the actual distance of the point into the detection threshold curve to obtain f′(x i );Judge the actual amplitude y i and f′(x i ) size, if y i Greater than f′(x i ) is considered as the target point, otherwise it is considered as a noise point and removed from the point set. Repeat the above process to complete the detection of all points in the point set.

[0102] In order to verify the effectiveness of the low false alarm rate detection method for low, slow and small target detection provided by the embodiment of the present invention, a 24G linear frequency modulation continuous wave radar is used to collect data on the drone. Through the analysis of the measured data, the drone is flying vertically and S-shaped straightly. The detection method with an absolute threshold of noise amplitude +12dB and the detection method proposed by the present invention are used to screen the target points. The comparison diagram of the drone vertical flight scene is shown in the figure. Figure 7 (a)~ Figure 7 As shown in (b), the comparison of the UAV S-shaped direct flight scene is as follows Figure 8 (a)~ Figure 8 (b) from Figure 7 (a)~ Figure 7 (b) Figure 8 (a)~ Figure 8 (b) It can be seen that compared with the absolute threshold detection method, the detection method proposed in the present invention has a better clutter suppression effect.

[0103] After that, the radar system is verified as a whole. The radar is used to detect the drone in real time, and the radar terminal display interface shows the following: For the same drone in the same environment at different times and different flight postures, the track display comparison diagram of the radar terminal after using the absolute threshold detection method and the detection method proposed by the present invention is as follows: Fig. 9 (a)~ Fig. 9 As shown in (b), Fig. 9 (a)~ Fig. 9 (b) It can be seen that a large number of noise points are eliminated after using the detection method proposed by the present invention.

[0104] In summary, the low false alarm rate detection method suitable for low, slow and small target detection proposed in the embodiment of the present invention is simple but can detect the correct target while reducing the probability of noise passing through the detection, thereby reducing the false alarm probability of the radar, compared with the existing constant false alarm rate detection method. Specifically: by counting the measured point trace data of a large number of UAV targets, the relationship between their distance and amplitude is fitted using the least squares method; in the constant false alarm rate detection, the absolute threshold calculated according to the noise mean value is replaced by a detection threshold curve designed according to the distance and amplitude characteristics of the UAV target. The design idea of ​​the detection threshold curve is to use the relationship between the residual and normal distribution of the actual value of the UAV target point and the fitted value to ensure the preset expected threshold, such as a target detection probability of more than 98%, to obtain the coefficient of the detection threshold curve, and the target detection curve is designed by the coefficient of the fitting curve and the detection threshold curve, so as to use the target detection curve for constant false alarm rate detection. In summary, the detection method proposed in the embodiment of the present invention can more accurately distinguish between low, slow and small target points and noise, reduce the false alarm probability of the radar, and have a good suppression effect on radar clutter; reduce the influence of environmental clutter on the radar detection of low, slow and small targets, reduce the load of the radar system, and enhance the real-time and accuracy of the system. It has great application prospects for the detection and tracking of low, slow and small targets such as drones in low-altitude environments, such as urban monitoring and airspace management.

[0105] Second, see Fig.10 The embodiment of the present invention provides a low false alarm rate detection device suitable for detecting slow and small targets, the device comprising:

[0106] The acquisition module is used to collect the point trace data containing low, slow and small targets, and select the point trace set of the UAV target from all the point trace data; each point trace in the point trace set includes the actual amplitude and actual distance;

[0107] A fitting module is used to fit the relationship between the actual amplitude and the actual distance using the least square method to obtain a fitting curve; wherein, during the fitting process, the actual amplitude of each point trace in the point trace set is used as a dependent variable, and the actual distance of each point trace in the point trace set is used as an independent variable to construct a fitting curve;

[0108] A first calculation module is used to calculate the residual of each point trace in the point trace set according to the fitting curve and the actual curve corresponding to the point trace set; wherein the distribution of all residuals satisfies the normal distribution;

[0109] A second calculation module is used to calculate the coefficient of the detection threshold curve according to the normal distribution characteristics of all residuals;

[0110] A design module, used for designing a detection threshold curve according to the fitting curve and the coefficients of the detection threshold curve;

[0111] The detection module is used to perform constant false alarm detection using a detection threshold curve.

[0112] As for the device embodiment of the second aspect, since it is basically similar to the method embodiment of the first aspect, the description is relatively simple, and the relevant parts may refer to the partial description of the method embodiment of the first aspect.

[0113] In the description of the present invention, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0114] Although the present invention is described herein in conjunction with various embodiments, in the process of implementing the claimed invention, those skilled in the art may understand and implement other variations of the disclosed embodiments by viewing the specification and its drawings. In the specification, the word "comprising" does not exclude other components or steps, and "a" or "one" does not exclude multiple situations. Certain measures are recorded in different embodiments, but this does not mean that these measures cannot be combined to produce good results.

[0115] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For ordinary technicians in the technical field to which the present invention belongs, several simple deductions or substitutions can be made without departing from the concept of the present invention, which should be regarded as falling within the protection scope of the present invention.

Claims

1. A low false alarm rate detection method suitable for detecting slow and small targets, characterized in that: The method comprises: Collecting point trace data containing low, slow and small targets, and selecting a point trace set of the UAV target from all the point trace data; each point trace in the point trace set includes an actual amplitude and an actual distance; Using the least square method, fitting the relationship between the actual amplitude and the actual distance to obtain a fitting curve; wherein, during the fitting process, the actual amplitude of each point trace in the point trace set is used as a dependent variable, and the actual distance of each point trace in the point trace set is used as an independent variable to construct a fitting curve; Calculating the residual of each point in the point trace set according to the fitting curve and the actual curve corresponding to the point trace set; wherein the distribution of all residuals satisfies the normal distribution; Calculate the coefficients of the detection threshold curve based on the normal distribution characteristics of all residuals; Designing a detection threshold curve according to the fitting curve and the coefficients of the detection threshold curve; The detection threshold curve is used to perform constant false alarm detection.

2. The low false alarm rate detection method suitable for low, slow and small target detection according to claim 1 is characterized in that: The coefficients of the detection threshold curve are calculated based on the normal distribution characteristics of all residuals, including: Calculating the fitting amplitude of each point trace in the point trace set according to the fitting curve; Calculate the residual of each point trace of the point trace set according to the fitting amplitude and the actual amplitude; Calculating an average residual according to the residual of each point trace in the point trace set; Calculate the standard deviation of the residual according to the residual of each point trace in the point trace set and the average residual; According to the preset expected threshold, the probability distribution table of the standard normal distribution is queried to obtain the width of the confidence interval; The coefficient of the detection threshold curve is calculated according to the standard deviation of the residual and the width of the confidence interval.

3. The low false alarm rate detection method suitable for low, slow and small target detection according to claim 2 is characterized in that: The formula for calculating the standard deviation of the residuals is: Among them, σ represents the standard deviation of the residual, n represents the number of points in the point set, and r i Represents the residual of the ith point in the point set, r i =y i -f(x i ), y i Indicates the actual amplitude of the i-th point in the point set, x i represents the actual distance of the ith point in the point set, f(x i ) represents the actual distance x calculated according to the fitting curve i The fitting amplitude at Represents the mean of the residual errors of all points in the set of points.

4. The low false alarm rate detection method suitable for low, slow and small target detection according to claim 2 is characterized in that: The formula for calculating the coefficient of the detection threshold curve is: Among them, A represents the coefficient of the detection threshold curve, n represents the number of points in the point set, and x i represents the actual distance of the ith point in the point set, f(x i ) represents the actual distance x calculated according to the fitting curve i , σ represents the standard deviation of the residuals, and z represents the width of the confidence interval.

5. The low false alarm rate detection method suitable for low, slow and small target detection according to claim 2, characterized in that: When the preset expected threshold is 98%, the width of the confidence interval is -2.

05.

6. The low false alarm rate detection method suitable for low, slow and small target detection according to claim 1, characterized in that: The designed detection threshold curve formula is expressed as: f′(x)=A·f(x); Where f′(x) represents the detection threshold curve, A represents the coefficient of the detection threshold curve, f(x) represents the fitting curve, x represents the set of actual distances of the points in the point set, and x = {x1,…,x i ,…,x n }, x i Indicates the actual distance of the i-th point in the point set.

7. The low false alarm rate detection method suitable for low, slow and small target detection according to claim 1, characterized in that: Using the detection threshold curve to perform constant false alarm detection includes: For each point in the point set, the execution steps include: using the detection threshold curve to calculate the estimated amplitude of the point, judging whether the actual amplitude of the point is greater than the estimated amplitude, if so, judging that the point is a target point, if not, judging that the point is a noise point, and removing the point from the point set.

8. A low false alarm rate detection device suitable for detecting slow and small targets, characterized in that: The device comprises: The acquisition module is used to collect point trace data containing low, slow and small targets, and select a point trace set of the UAV target from all the point trace data; each point trace in the point trace set includes an actual amplitude and an actual distance; A fitting module, used to fit the relationship between the actual amplitude and the actual distance using the least square method to obtain a fitting curve; wherein, during the fitting process, the actual amplitude of each point trace in the point trace set is used as a dependent variable, and the actual distance of each point trace in the point trace set is used as an independent variable to construct a fitting curve; A first calculation module, used for calculating the residual of each point trace in the point trace set according to the fitting curve and the actual curve corresponding to the point trace set; wherein the distribution of all residuals satisfies the normal distribution; A second calculation module is used to calculate the coefficient of the detection threshold curve according to the normal distribution characteristics of all residuals; A design module, used for designing a detection threshold curve according to the fitting curve and the coefficients of the detection threshold curve; The detection module is used to perform constant false alarm detection using the detection threshold curve.

Images