A bird flock target recognition method based on range-Doppler integrated features

Abstract

The present invention discloses a method for identifying flocks of birds based on comprehensive range-Doppler features, which relates to the field of radar target identification technology and includes the following steps: S1, group target detection, and determining the target type based on the detection points; S2, making a preliminary determination of the target based on the range-Doppler dispersion characteristics of the detection points; S3, determining the fluctuation between detection points for the target determined by the range-Doppler dispersion characteristics; and S4, determining the power ratio of the detection points for the target determined by the fluctuation between detection points. The present invention adopts the above-mentioned method for identifying flocks of birds based on comprehensive range-Doppler features, and uses features such as the range-Doppler dispersion of detection points, the power fluctuation between detection points, and the power ratio of detection points to determine the target signal in a cascaded manner and output the identification result, thereby improving the accuracy of flock target identification.

Description

Technical Field

[0001] The present invention relates to the technical field of radar target recognition, in particular to a bird flock target recognition method based on range-Doppler comprehensive characteristics. Background Art

[0002] Bird strikes are a significant threat to aviation safety in civil aviation airports. Compared to individual birds, flocks of birds, due to their sheer numbers, significantly increase the probability of multiple birds simultaneously striking critical parts of an aircraft, such as the fuselage and engine, upon encountering it during flight. This poses a serious threat to the lives of both the crew and the aircraft. However, the highly maneuverable flight posture of a flock of birds, coupled with the dense density of individual birds within, results in intertwined echo signals, making it difficult for radar systems to accurately identify flock targets, resulting in high false alarms and missed alerts. Therefore, improving radar's ability to identify bird flocks and building a solid defense for aviation safety has become a pressing issue in radar signal processing.

[0003] Range-Doppler dispersion, as a target characterization method, provides a new approach for identifying flocks of birds. During radar detection, flocks of birds form a unique dispersion pattern in the range-Doppler domain due to differences in flight speed, distance, and relative motion with the radar. In contrast, individual birds typically exhibit a concentrated distribution. Furthermore, the ever-changing relative positions of individual birds within a flock, as well as inter-individual occlusion and scattering characteristics, also significantly differ from those of individual birds, and can serve as a key indicator for identifying flock targets.

[0004] Current bird flock recognition relies primarily on traditional single-dimensional feature analysis methods, including threshold detection based on echo amplitude, velocity estimation based on Doppler frequency, and cluster analysis based on statistical models. Threshold detection filters suspected targets by setting an amplitude threshold, but is susceptible to clutter and cannot distinguish target types. Doppler analysis uses frequency shift information to determine direction and velocity, but underutilizes velocity distribution within a flock. Cluster analysis classifies objects based on statistical feature similarity, but its performance is highly dependent on feature selection and prior knowledge. Furthermore, while deep learning-based methods have demonstrated certain advantages in complex scenarios, they suffer from insufficient training data and high computational resource requirements.

[0005] Therefore, there is an urgent need for a bird flock recognition method that combines distance-Doppler comprehensive features to improve the recognition accuracy and robustness in complex scenarios through multi-dimensional feature fusion. Summary of the Invention

[0006] The purpose of the present invention is to provide a bird flock target recognition method based on the comprehensive characteristics of range-Doppler. By using the characteristics of the range-Doppler dimension detection point dispersion, the power fluctuation between the detection points, the power ratio of the detection points and so on, the target signal is judged step by step in a cascade manner and the recognition result is output, thereby improving the accuracy of bird flock target recognition.

[0007] To achieve the above object, the present invention provides a method for bird flock target recognition based on range-Doppler comprehensive characteristics, comprising the following steps:

[0008] S1, group target detection, judging the target type based on the detection points;

[0009] S2. Make a preliminary judgment on the target based on the distance-Doppler dispersion characteristics of the detection point;

[0010] S3: For targets identified by range-Doppler dispersion characteristics, fluctuations between detection points are determined. Detection points that meet the requirements are marked as suspected bird flock targets and proceed to S4. Other detection points are marked as non-bird flock targets, and the identification process is terminated.

[0011] S4. For the targets identified by the fluctuations between detection points, the detection point power ratio is judged. The detection points that meet the requirements are marked as suspected bird flock targets and enter the subsequent judgment process. Other detection points are marked as non-bird flock targets and the recognition process is terminated.

[0012] Preferably, S1 comprises the following steps:

[0013] S11, the echo signal is pulse compressed and coherently integrated to obtain the range-Doppler matrix. CFAR is used to detect targets in the range and Doppler dimensions respectively. The threshold of CFAR detection is calculated as follows:

[0014] ;

[0015] in, is the CFAR detection threshold, is the scaling factor, is the reference window length, is the sample value of the range-Doppler unit in the reference window;

[0016] S12: traverse all detection points and perform subsequent level-by-level judgment on each detection point.

[0017] Preferably, in S12, the group targets present a cluster distribution, and the single targets present a discrete distribution.

[0018] Preferably, S2 comprises the following steps:

[0019] S21. Taking the current traversal point as the center, extract the monitoring point with the largest Doppler amplitude corresponding to the distance unit where the current point is located as the main peak point, which is expressed as:

[0020] ;

[0021] in, is the distance unit to the main peak, It is the main peak point Doppler unit;

[0022] S22. Cluster the main peak point and surrounding detection points as follows:

[0023] ;

[0024] in, represents the set of detection points obtained after clustering, is the number of detection points in the set; and represents the distance cluster range, and represents the Doppler dimension clustering range;

[0025] S23. Calculate the distance dimension and Doppler dimension scatter characteristics of the detection point set where the main peak point is located as follows:

[0026] ;

[0027] ;

[0028] ;

[0029] in, is the distance dimension scatter vector, is the Doppler spread vector, is the scatter eigenvector;

[0030] S24. Determine distance dispersion characteristics Whether it meets:

[0031] ;

[0032] in, and represent the range dimension and Doppler dimension scatter thresholds respectively.

[0033] Preferably, S3 comprises the following steps:

[0034] S31. Extract all detection points in the Doppler dimension corresponding to the distance unit where the main peak is located, and arrange them according to the size of the Doppler unit:

[0035] ;

[0036] in, is the number of Doppler dimension detection points corresponding to the distance unit where the main peak is located. If , it is marked as a non-bird flock target, otherwise continue to judge;

[0037] S32, calculate the average power of adjacent non-main peak detection points, and obtain the average power of all signal points between the adjacent non-main peak detection points of the Doppler signal of the current traversal point, calculate the difference between the two and compare it with the fluctuation threshold:

[0038] ;

[0039] ;

[0040] in, is the fluctuation eigenvector, represents the signal power corresponding to the range-Doppler unit, is the number of Doppler units between adjacent non-main peak detection points, is the target Doppler detection point index, is the signal point index between target Doppler detection points, It is the fluctuation threshold.

[0041] Preferably, S4 comprises the following steps:

[0042] S41. For the detection point set extracted in S31, extract the point with the maximum signal power among the detection points other than the main peak point, and record it as ;

[0043] S42. Calculate the main peak point and the second maximum power point The power ratio is compared with the power ratio threshold;

[0044] ;

[0045] ;

[0046] in, is the power ratio eigenvector, is the power ratio threshold, and the detection points that meet the threshold requirement are marked as bird flock targets, and other detection points are marked as non-bird flock targets.

[0047] Therefore, the present invention adopts the aforementioned bird flock target recognition method based on comprehensive range-Doppler features. This method combines the characteristic differences between individual targets during flight, the two-dimensional range-Doppler dispersion of the target's reflected echo, and comprehensively utilizes the fluctuation characteristics and power ratio characteristics of the signal in the clustering results to identify the target. This method fully utilizes multidimensional feature fusion to improve recognition accuracy.

[0048] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of a bird flock target recognition method based on range-Doppler comprehensive characteristics of the present invention;

[0050] Figure 2 This is the bird flock target RD graph of the present invention;

[0051] Figure 3 This is the RD diagram of a single bird target in the present invention;

[0052] Figure 4 This is the Doppler signal diagram of the bird flock target of the present invention;

[0053] Figure 5 This is the Doppler signal diagram of a single bird target in the present invention. DETAILED DESCRIPTION

[0054] The technical solution of the present invention is further described below with reference to the accompanying drawings and embodiments.

[0055] Unless otherwise defined, the technical or scientific terms used in the present invention shall have the usual meanings understood by persons of ordinary skill in the field to which the present invention belongs. The words "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Words such as "include" or "comprise" mean that the elements or objects preceding the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connect" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0056] Example

[0057] See also Figure 1-Figure 5 The present invention provides a method for identifying bird flock targets based on range-Doppler comprehensive features, which specifically includes the following steps:

[0058] S1. Group target detection.

[0059] S11. After pulse compression and coherent integration, the echo signal is converted into a range-Doppler matrix. CFAR (constant false alarm rate) is used to detect targets in the range and Doppler dimensions. The threshold of CFAR detection is calculated as follows:

[0060] ;

[0061] in, is the CFAR detection threshold, is the scaling factor, is the reference window length, is the sample value of the range-Doppler unit in the reference window.

[0062] S12: traverse all detection points and perform subsequent level-by-level judgment on each detection point.

[0063] When a group of targets is flying, it will show an overall moving trend. On the premise that the radar range and velocity resolution meet the requirements, multiple detection points will appear on the RD diagram after CFAR detection. The group of targets will show a cluster distribution, and the single target will show a discrete distribution. The detection points are the main basis for the subsequent judgment process.

[0064] S2. Make a preliminary judgment on the target based on the distance-Doppler dispersion characteristics of the detection point.

[0065] S21. Taking the current traversal point as the center, extract the detection point with the largest Doppler amplitude corresponding to the distance unit where the current point is located as the main peak point, which is expressed as:

[0066] ;

[0067] in, is the distance unit to the main peak, It is the main peak point Doppler unit;

[0068] S22. Cluster the main peak point and surrounding detection points as follows:

[0069] ;

[0070] in, represents the set of detection points obtained after clustering, is the number of detection points in the set; and represents the distance cluster range, and Indicates the Doppler clustering range.

[0071] S23. Calculate the distance dimension and Doppler dimension scatter characteristics of the detection point set where the main peak point is located as follows:

[0072] ;

[0073] ;

[0074] ;

[0075] in, is the distance dimension scatter vector, is the Doppler spread vector, is the scatter eigenvector;

[0076] S24. Determine distance dispersion characteristics Whether it meets:

[0077] ;

[0078] in, and represent the range dimension and Doppler dimension scatter thresholds respectively.

[0079] The detection points whose scatter characteristics meet the threshold are marked as suspected bird flock targets and enter the subsequent judgment process. Other detection points are identified as non-bird flock targets and the recognition process is terminated.

[0080] In a clutter environment, the echo signal of a single bird target aliases with static clutter in the range and Doppler dimensions, resulting in multiple detection points. Individual targets in a flock of birds also vary in range and speed, similarly leading to multiple detection points in both the range and Doppler dimensions. By clustering these detection points and analyzing their dispersion characteristics, we can initially distinguish between single targets and flocks.

[0081] S3. For the target determined by the range-Doppler dispersion characteristics, the fluctuation between detection points is determined.

[0082] S31. Extract all detection points in the Doppler dimension corresponding to the distance unit where the main peak is located, and arrange them according to the size of the Doppler unit:

[0083] ;

[0084] in, is the number of Doppler dimension detection points corresponding to the distance unit where the main peak is located. If , it is marked as a non-bird flock target, otherwise continue to judge.

[0085] S32, calculate the difference between the two and compare it with the fluctuation threshold:

[0086] ;

[0087] ;

[0088] in, is the fluctuation eigenvector, represents the signal power corresponding to the range-Doppler unit, is the number of Doppler units between adjacent non-main peak detection points, is the target Doppler detection point index, is the signal point index between target Doppler detection points, It is the fluctuation threshold.

[0089] The detection points that meet the threshold are marked as suspected bird flock targets and enter the subsequent process judgment. Other detection points are marked as non-bird flock targets and the recognition process is terminated.

[0090] The power of a single bird target in a static clutter environment usually varies greatly between multiple detection points in the Doppler spectrum. The speed distribution of individual bird targets in a flock of birds is random, and there is no distribution pattern between adjacent detection points in the Doppler spectrum. The spectrum superposition of multiple targets after coherent accumulation results in the power between two adjacent detection points being stronger than the noise floor level.

[0091] S4. For the target determined by the fluctuation between detection points, the detection point power ratio is determined.

[0092] S41, for the detection point set extracted in step S31, extract the point with the maximum signal power among the detection points other than the main peak point, and record it as .

[0093] S42. Calculate the main peak point and the second maximum power point The power ratio is compared with the power ratio threshold.

[0094] ;

[0095] ;

[0096] in, is the power ratio eigenvector, is the power ratio threshold. Detection points that meet the threshold are marked as flock targets, and other detection points are marked as non-flock targets. The flock identification process ends.

[0097] Bird flocks are often composed of individual targets of similar size, resulting in relatively small differences in echo intensity. However, the signal intensity of a single bird detection point is significantly different from that of the clutter signal. This difference can be exploited to better distinguish between flocks and single birds.

[0098] In this example, an X-band phased array radar is used to identify low-altitude bird flocks. The radar parameters are as follows.

[0099] Table 1 X-band phased array radar parameters

[0100] ;

[0101] For different targets, the target data of bird flocks and single bird targets are collected at fixed points to obtain multiple sets of measured data, and the single bird targets and bird flock targets are processed separately to obtain Figure 2-Figure 5 By performing steps S1-S4 of this patent on the flock target data, the accuracy of bird flock recognition was found to be over 80%, while the single bird target data was not recognized as a flock. This proves the accuracy and effectiveness of this method.

[0102] Therefore, the present invention adopts the above-mentioned bird flock target recognition method based on the range-Doppler comprehensive feature, combines the characteristic differences between individual targets during the flight of the group target, combines the range-Doppler two-dimensional dispersion characteristics of the target reflection echo, and comprehensively utilizes the fluctuation characteristics and power ratio characteristics of the signal in the clustering result to judge the target, and makes full use of multi-dimensional feature fusion to improve the recognition accuracy.

[0103] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit the same. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solutions of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for identifying bird flock targets based on range-Doppler integrated features, characterized in that: The following steps are involved: S1, group target detection, judging the target type based on the detection points; S2. Make a preliminary judgment on the target based on the distance-Doppler dispersion characteristics of the detection point; S3: For targets identified by range-Doppler dispersion characteristics, fluctuations between detection points are determined. Detection points that meet the requirements are marked as suspected bird flock targets and proceed to S4. Other detection points are marked as non-bird flock targets, and the identification process is terminated. S4. For the targets identified by the fluctuations between the detection points, the detection point power ratio is determined. The detection points that meet the requirements are marked as bird flock targets, and the other detection points are marked as non-bird flock targets. The bird flock identification process ends. S3 includes the following steps: S31. Extract all detection points in the Doppler dimension corresponding to the distance unit where the main peak is located, and arrange them according to the size of the Doppler unit: ； in, is the number of Doppler dimension detection points corresponding to the distance unit where the main peak is located. If , it is marked as a non-bird flock target, otherwise continue to judge; S32, calculate the average power of adjacent non-main peak detection points, and obtain the average power of all signal points between the adjacent non-main peak detection points of the Doppler signal of the current traversal point, calculate the difference between the two and compare it with the fluctuation threshold: ； ； in, is the fluctuation eigenvector, represents the signal power corresponding to the range-Doppler unit, is the number of Doppler units between adjacent non-main peak detection points, i is the target Doppler detection point index, j is the signal point index between target Doppler detection points, It is the fluctuation threshold.

2. The bird flock target recognition method based on range-Doppler comprehensive characteristics according to claim 1, characterized in that: S1 includes the following steps: S11, the echo signal is pulse compressed and coherently integrated to obtain the range-Doppler matrix. CFAR is used to detect targets in the range and Doppler dimensions respectively. The threshold of CFAR detection is calculated as follows: ； in, is the CFAR detection threshold, is the scaling factor, is the reference window length, is the sample value of the range-Doppler unit in the reference window; S12: traverse all detection points and perform subsequent level-by-level judgment on each detection point.

3. The method for bird flock target recognition based on range-Doppler integrated features according to claim 2, characterized in that: In S12, group targets present a cluster distribution, while single targets present a discrete distribution.

4. The bird flock target recognition method based on range-Doppler comprehensive characteristics according to claim 3, characterized in that: S2 includes the following steps: S21. Taking the current traversal point as the center, extract the monitoring point with the largest Doppler amplitude corresponding to the distance unit where the current point is located as the main peak point, which is expressed as: ； in, is the distance unit to the main peak, It is the main peak point Doppler unit; S22. Cluster the main peak point and surrounding detection points as follows: ； in, represents the set of detection points obtained after clustering, is the number of detection points in the set; and represents the distance cluster range, and represents the Doppler dimension clustering range; S23. Calculate the distance dimension and Doppler dimension scatter characteristics of the detection point set where the main peak point is located as follows: ； ； ； in, is the distance dimension scatter feature, is the Doppler scatter characteristic, is the dispersion characteristic; S24. Determine the distribution characteristics Whether it meets: ； in, and represent the range dimension and Doppler dimension scatter thresholds respectively.

5. The bird flock target recognition method based on range-Doppler comprehensive characteristics according to claim 4, characterized in that: S4 includes the following steps: S41. For the detection point set extracted in S31, extract the point with the maximum signal power among the detection points other than the main peak point, and record it as ; S42. Calculate the main peak point and the second maximum power point and compare it with the power ratio threshold; ； ； in, is the power ratio eigenvector, is the power ratio threshold, and the detection points that meet the threshold requirement are marked as bird flock targets, and other detection points are marked as non-bird flock targets.

Images