A method for classification of flying birds and UAV radar targets based on motion characteristics

Abstract

The present invention discloses a method for classifying flying birds and unmanned aerial vehicle radar targets based on motion features, the steps of which include: establishing a feature recognition library: collecting data of unmanned aerial vehicles and bird targets; performing feature statistics on the data; using the above features as samples to form a Fisher feature library; real-time target classification and recognition: real-time calculation of the motion features of the detected targets; judging the motion features according to the feature library, classifying and identifying the targets as bird targets or unmanned aerial vehicle targets. The new target classification method based on the Fisher classification and discrimination method of the present invention is mainly used for classifying and identifying unmanned aerial vehicle targets and bird targets.

Description

Technical Field

[0001] The present invention relates to the field of radar information processing, and in particular to a method for classifying flying birds and unmanned aerial vehicle radar targets based on motion characteristics. Background Art

[0002] In recent years, with the gradual opening of the country's low-altitude airspace and the country's active support for the development of general aviation and the UAV industry, traditional physical protection is mainly aimed at ground threats, resulting in increasingly prominent threats and harassment from aerial targets. At present, the fastest-growing and most prominent problems in prevention and control are small civilian consumer drones. In particular, multi-rotor drones are small in size, easy to carry, low in purchase cost, easy to obtain, simple to operate, and popularized quickly, which will bring real safety risks to key areas. Low-altitude flying objects are of various types and large in number, especially bird targets are easily confused with small drone targets, which will bring unnecessary false alarms.

[0003] For low-altitude targets, the more obvious the multipath effect is, the greater the fluctuation of the signal amplitude, the wider the notch width of the signal amplitude fluctuation with distance, and the amplitude fluctuation phenomenon of bird echoes. When birds vibrate their wings, the amplitude of the echo will change with the vibration of the wings due to the change in flight posture. The flight posture of drone targets is relatively stable, and the amplitude change of the echo is also stable. Based on its motion characteristics, the Fisher discriminant analysis method can be used to derive a set of effective target classification methods under specific conditions. Summary of the invention

[0004] In view of the problems existing in the prior art, the present invention provides a new target classification method based on the Fisher classification and discrimination method, which is mainly used for classifying and identifying UAV targets and bird targets.

[0005] The purpose of the present invention is achieved through the following technical solutions.

[0006] A method for classifying flying birds and unmanned aerial vehicle radar targets based on motion characteristics, comprising the following steps:

[0007] 1) Establish a feature recognition library: collect drone data and bird target data; perform feature statistics on the data; use the above features as samples to form a Fisher feature library;

[0008] 2) Real-time target classification and recognition: Calculate the motion characteristics of the detected target in real time; judge the motion characteristics based on the feature library, classify and identify the target as a bird target or a drone target.

[0009] Perform feature statistics on the data including amplitude features and speed features.

[0010] The amplitude characteristics include: RCS characteristics, RCS standard deviation, coefficient of variation, amplitude difference mean, smoothing difference mean, period, difference variance, dB difference mean, range, standard deviation, mean absolute deviation, and standard mean deviation.

[0011] The speed characteristics include: speed standard deviation, speed difference mean, smoothing difference mean, speed difference variance, speed range, speed mean absolute deviation, and coefficient of variation.

[0012] Compared with the prior art, the advantages of the present invention are:

[0013] 1) Accurate classification: Traditional methods of studying target information detected by radar mostly focus on RCS feature research. RCS feature is the radar cross-section feature of the target, which more intuitively reflects the size of the target. In actual work, bird and drone targets are greatly affected by environmental factors, including different types of birds, bird formations, radar bands, etc. It is not feasible to identify targets simply by RCS features.

[0014] 2) High real-time performance: The target can be classified and judged in real time, leaving enough time for anti-no processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a process flow chart of the present invention.

[0016] Figure 2 This is the distribution map of the collected UAV target and bird target characteristics.

[0017] Figure 3 In order to back-substitute the test data, a statistical table of the correct recognition rates of drone targets and birds is provided.

[0018] Figure 4 A statistical table of the results of single-frame recognition (10-frame sliding window) and sequential recognition (6-frame sequential) for a new batch of radar measured data. DETAILED DESCRIPTION

[0019] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0020] like Figure 1 As shown, a method for classifying flying birds and UAV radar targets based on motion features comprises the following steps:

[0021] 1) Establish a feature recognition library: collect drone data and bird target data; perform feature statistics on the data; use the above features as samples to form a Fisher feature library;

[0022] 2) Real-time target classification and recognition: Calculate the motion characteristics of the detected target in real time; judge the motion characteristics based on the feature library, classify and identify the target as a bird target or a drone target.

[0023] Perform feature statistics on the data including amplitude features and speed features.

[0024] The amplitude characteristics include: RCS characteristics, RCS standard deviation, coefficient of variation, amplitude difference mean, smoothing difference mean, period, difference variance, dB difference mean, range, standard deviation, mean absolute deviation, and standard mean deviation.

[0025] The speed characteristics include: speed standard deviation, speed difference mean, smoothing difference mean, speed difference variance, speed range, speed mean absolute deviation, and coefficient of variation.

[0026] The present invention has clear classification: most of the traditional methods for studying target information detected by radar focus on RCS feature research.

[0027] The RCS feature is the radar cross-section feature of the target, which directly reflects the size of the target. In actual work, bird and drone targets are greatly affected by environmental factors, including different types of birds, bird formations, radar bands, etc. It is not feasible to identify targets based solely on RCS features.

[0028] like Figure 2 As shown, the collected UAV target and bird target feature distribution map: blue: bird target feature distribution curve; red: UAV target feature distribution curve.

[0029] The collected bird target data and drone target data are taken as a whole. Through the feature statistics of the data, the features used by birds and drones are

[0030] Amplitude characteristics include: RCS characteristics, RCS standard deviation, coefficient of variation, amplitude difference mean, smoothing difference mean, period, difference variance, dB difference mean, range, standard deviation, mean absolute deviation, and standard mean deviation.

[0031] Speed ​​characteristics include: speed standard deviation, speed difference mean, smoothed difference mean, speed difference variance, speed range, speed mean absolute deviation, and coefficient of variation.

[0032] like Figure 3 As shown, by back-substituting the test data, it can be found that the correct recognition rate of this classification method for drone targets is 92.1%, and the correct recognition rate for bird targets is 91.6%.

[0033] A new batch of radar measured data was selected to conduct statistical tests on single-frame recognition (10-frame sliding window) and sequential recognition (6-frame sequential) results. The statistical results are as follows: Figure 4 shown.

[0034] Most of the traditional methods for studying target information detected by radar focus on RCS characteristics, which are greatly affected by environmental factors and have a low recognition accuracy. The Fisher classification and discrimination method based on motion characteristics is reasonable and feasible for target classification and recognition of bird targets and UAV targets. The recognition accuracy of targets is basically around 90%. This classification method is simple and feasible, with high real-time performance: it can classify and judge targets in real time, leaving enough time for anti-no processing. The classification efficiency is high and has certain application prospects.

Claims

1. A method for classifying flying birds and unmanned aerial vehicle radar targets based on motion characteristics, characterized by the following steps: include: 1) Establish a feature recognition library: collect drone data and bird target data; perform feature statistics on the data; use the above features as samples to form a Fisher feature library; 2) Real-time target classification and recognition: Calculate the motion characteristics of the detected target in real time; judge the motion characteristics based on the feature library, classify and identify the target as a bird target or a drone target.

2. The method for classifying flying birds and unmanned aerial vehicle radar targets based on motion characteristics according to claim 1 is characterized in that Perform feature statistics on the data including amplitude features and speed features.

3. The method for classifying flying birds and unmanned aerial vehicle radar targets based on motion characteristics according to claim 2 is characterized in that The amplitude characteristics include: RCS characteristics, RCS standard deviation, coefficient of variation, amplitude difference mean, smoothing difference mean, period, difference variance, dB difference mean, range, standard deviation, mean absolute deviation, and standard mean deviation.

4. The method for classifying flying birds and unmanned aerial vehicle radar targets based on motion characteristics according to claim 2 is characterized in that The speed characteristics include: speed standard deviation, speed difference mean, smoothing difference mean, speed difference variance, speed range, speed mean absolute deviation, and coefficient of variation.

Images