Unmanned aerial vehicle detection and identification method and system

The images are denoised and meshed through convolutional networks, combined with Gabor filters, and solved the problem of drone identification, realized effective detection, identification and countermeasures of drones, and protected personal privacy and aviation security.

CN119068364AInactive Publication Date: 2024-12-03CHONGQING VOCATIONAL COLLEGE OF TRANSPORTATION +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411148639.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-21
Publication Date
2024-12-03
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The prior art is difficult to effectively identify and prevent unauthorized drones, ensuring personal privacy and aviation security, especially in cases where unauthorized drones may invade privacy, interfere with aviation security, or be used for illegal activities.

Method used

The image is denoised and meshed by using convolutional networks. The convolutional module and residual network module are combined with Gabor filters to predict the position of the drone, and the data set is constructed and trained to achieve effective detection of the drone. Test and identify.

Benefits of technology

It has realized the effective detection and identification of drones, provided corresponding countermeasures, guaranteed personal privacy and aviation security, and cracked down on the abuse of drones.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119068364A_ABST
    Figure CN119068364A_ABST
Patent Text Reader

Abstract

The embodiment of the invention discloses an unmanned aerial vehicle detection and identification method and system. The method comprises the following steps: acquiring a to-be-detected image through image acquisition equipment for a monitoring area; each to-be-detected image at least comprises an unmanned aerial vehicle; performing denoising processing on the to-be-detected image to obtain an original image; inputting the original image into a pre-trained convolutional network for grid division to obtain a plurality of grid images; and for each grid image, performing position prediction of the target object based on the first frame and the second frame to obtain position information of the unmanned aerial vehicle. According to the embodiment of the invention, the convolutional network is adopted to perform target detection on the image acquired by the image acquisition equipment, so that the position information of the unmanned aerial vehicle is predicted, and effective detection and identification of the unmanned aerial vehicle are realized; based on the obtained position information of the unmanned aerial vehicle, corresponding countering measures can be taken for the unmanned aerial vehicle, guarantee is provided for personal privacy protection, aviation safety and the like, and the phenomenon of abuse of the unmanned aerial vehicle can be effectively attacked.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of target detection, and in particular to a method and system for detecting and identifying unmanned aerial vehicles. Background Art

[0002] Unmanned aerial vehicles (UAVs) are widely used in aerial photography, express delivery, news reporting, disaster relief, and wildlife observation. For example, in the field of wildlife protection, manual census is dangerous and costly, but the use of drones for detection can better determine the location and number of wild animals, facilitate the understanding of wild animal habits, and play a certain role in the protection of wild animals. In terms of monitoring and early warning, the real-time transmission function of drones and computer technology knowledge are used to design an anti-trampling monitoring and early warning system. The drone obtains video data, transmits the density of human traffic to the system in real time, compares it with the established early warning mathematical model, calculates the early warning signal, and provides a basis for on-site decision-making, guidance, and emergency plan formulation. In terms of field rescue, in the field environment with poor network conditions and bad weather, drones can provide the best path planning for trapped people by identifying roads and detecting road conditions, and can also identify emergency situations and potential dangerous conditions.

[0003] Although the application of drones is becoming more and more widespread, it also brings some potential security threats. For example, unauthorized drones may violate privacy, interfere with aviation safety, and be used for illegal activities. Summary of the invention

[0004] In view of the deficiencies of the prior art, an object of the embodiments of the present invention is to provide a method and system for detecting and identifying a drone.

[0005] To achieve the above objectives, in a first aspect, an embodiment of the present invention provides a method for detecting and identifying a drone, comprising:

[0006] For the monitoring area, an image to be detected is acquired by an image acquisition device; each image to be detected includes at least one target object, and the target object includes at least a drone;

[0007] Performing denoising on the image to be detected to obtain an original image;

[0008] Inputting the original image into a pre-trained convolutional network for grid division to obtain a plurality of grid images;

[0009] For each grid image, the position of the target object is predicted based on the first frame and the second frame to obtain the position information of the drone.

[0010] Furthermore, as a preferred implementation of the present application, the method further includes constructing and training a convolutional network, specifically:

[0011] Constructing a convolutional network; the convolutional network includes a convolutional module and a residual network module, the convolutional module includes a plurality of convolutional kernels, and each convolutional kernel is newly added with a Gabor filter;

[0012] Acquire sample images; the sample images include a first image and a second image, in which the drone maintains the same flying posture in the first image, and in which the drone adopts a different flying posture in the second image;

[0013] Performing denoising and filtering processing on the sample image to construct a first data set;

[0014] Performing clustering and expansion processing on the first data set to obtain a second training set;

[0015] The convolutional network is trained based on the second training set.

[0016] As a specific implementation of the present application, the second training set is obtained as follows:

[0017] Setting an initial membership degree according to the first training set;

[0018] According to the initial membership, setting the initial cluster center;

[0019] Obtaining a clustering loss function value through the initial cluster center;

[0020] Adjusting the membership degree according to the clustering loss function value to obtain the best cluster center;

[0021] Performing fuzzy classification on the first training set according to the optimal cluster center to obtain a classification result;

[0022] A second training set is obtained according to the classification result.

[0023] As a specific implementation method of the present application, the location information of the drone is obtained as follows:

[0024] Performing cluster analysis on the borders of the sample images in the first training set or the second training set to obtain a first border;

[0025] Performing translation and scaling processing on the first frame to obtain a second frame;

[0026] For each grid image, position prediction is performed based on the second frame to obtain the position information of the drone.

[0027] In the second aspect, an embodiment of the present application also provides a drone detection and identification system, including an image acquisition device and a data processing device; the data processing device includes a processor, an input device, an output device and a memory, and the processor, input device, output device and memory are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the method described in the first aspect above.

[0028] In a third aspect, an embodiment of the present application provides another drone detection and identification system, including an image acquisition device and a data processing device; wherein the data processing device includes:

[0029] An image acquisition unit is used to acquire images to be detected by an image acquisition device for the monitored area; each image to be detected includes at least one target object, and the target object includes at least a drone;

[0030] A denoising unit, used for performing denoising processing on the image to be detected to obtain an original image;

[0031] A division unit, used for inputting the original image into a pre-trained convolutional network for grid division to obtain a plurality of grid images;

[0032] The position prediction unit is used to predict the position of the target object based on the first frame and the second frame for each grid image, so as to obtain the position information of the drone.

[0033] Furthermore, as a preferred implementation of the present application, the system further includes a network training unit for constructing and training a convolutional network, specifically:

[0034] Constructing a convolutional network; the convolutional network includes a convolutional module and a residual network module, the convolutional module includes a plurality of convolutional kernels, and each convolutional kernel is newly added with a Gabor filter;

[0035] Acquire sample images; the sample images include a first image and a second image, in which the drone maintains the same flying posture in the first image, and in which the drone adopts a different flying posture in the second image;

[0036] Performing denoising and filtering processing on the sample image to construct a first data set;

[0037] Performing clustering and expansion processing on the first data set to obtain a second training set;

[0038] The convolutional network is trained based on the second training set.

[0039] In implementing the embodiments of the present invention, a convolutional network is used to perform target detection on images acquired by an image acquisition device, thereby predicting the location information of the drone, thereby achieving effective detection and identification of the drone; based on the obtained drone location information, corresponding countermeasures can be taken against the drone, providing protection for personal privacy protection, aviation safety, etc., and can also effectively combat the abuse of drones. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the specific implementation of the present invention or the technical solution in the prior art, the drawings required for use in the specific implementation or the description of the prior art are briefly introduced below.

[0041] Figure 1 is a schematic flow chart of a drone detection and identification method provided by a first embodiment of the present invention;

[0042] Figure 2 is a structural diagram of a drone detection and identification system provided by an embodiment of the present invention;

[0043] Figure 3 yes Figure 2 A structural diagram of the data processing device shown;

[0044] Figure 4 yes Figure 2 Another structural diagram of the data processing device shown. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0047] Please refer to Figure 1 , is a flow chart of a method for detecting and identifying a drone provided by a first embodiment of the present invention, comprising the following steps:

[0048] S1, build and train convolutional network.

[0049] Object detection algorithm is an important research direction in the field of computer vision, which is mainly used to locate and identify specific target objects in images or videos. The main types of object detection algorithms are:

[0050] (1) Traditional target detection algorithm: based on sliding window method, this method has large computational complexity and low efficiency;

[0051] (II) Object detection algorithms based on deep learning: two-stage detection algorithms and one-stage detection algorithms. Among them, two-stage detection algorithms include R-CNN, etc., which have high detection accuracy but slow speed; one-stage detection algorithms include YOLO, which has fast detection speed, but its accuracy is slightly lower than that of the two-stage detection algorithm for detection of some complex scenes.

[0052] Based on some defects of the existing target detection algorithm, the embodiment of the present invention provides a new target detection algorithm. The method first constructs and trains a convolutional network. The specific steps are as follows:

[0053] (1) Constructing a convolutional network

[0054] The convolution network includes a convolution module and a residual network module. The convolution module includes multiple convolution kernels, and each convolution kernel is newly added with a Gabor filter.

[0055] In this embodiment, the convolution kernel is improved from the Gabor filter, that is, the Gabor filter is added on the basis of the ordinary convolution kernel. Then, a convolution layer is constructed based on the obtained convolution kernel.

[0056] It should be noted that the addition of Gabor filters can encode directional information into ordinary convolution kernels to enhance the robustness of the learned features to directions, and integrate frequency information into different network layers to enhance the robustness of the learned features to scale changes.

[0057] In addition, the introduction of the residual network module can solve the training difficulties caused by the depth of the network.

[0058] (2) Obtaining sample images

[0059] During the flight of the drone, the distance between the drone and the image acquisition device can capture distant images (the drone looks smaller in the image) and close-up images (the drone looks larger in the image). The drone may also change its posture constantly. Therefore, in order to obtain more comprehensive samples, the convolutional network trained subsequently is more conducive to target prediction. Therefore, the sample images in this embodiment include a first image and a second image. In the first image, the drone maintains the same posture when flying, and in the second image, the drone adopts a different posture when flying.

[0060] (3) De-noising and filtering the sample image to construct a first data set

[0061] There will be some noise in the sample image obtained. In order to ensure the subsequent effect, denoising and filtering are usually required. Image denoising methods can adopt one or more of the following:

[0062] Mean filtering: Replace the value of each pixel in the image with the average value of its neighboring pixels. This method can effectively remove Gaussian noise, but it will make the image blurry.

[0063] Median filter: replaces the value of each pixel in the image with the median value of its neighboring pixels. Median filter has a good effect on removing salt and pepper noise, while better preserving the edges and details of the image.

[0064] Gaussian filtering: Use the Gaussian function as a weighting function to smooth the image. Gaussian filtering can effectively remove Gaussian noise while causing relatively little blur to the image.

[0065] Low-pass filtering: It removes high-frequency components from an image to achieve the purpose of denoising. Common low-pass filters include ideal low-pass filters and Butterworth low-pass filters. Low-pass filtering can effectively remove noise, but it will blur the edges and details of the image.

[0066] High-pass filtering: removes low-frequency components from an image and retains high-frequency components. High-pass filtering can enhance the edges and details of an image, but has limited effect on removing noise.

[0067] (4) Clustering and expanding the first data set to obtain a second training set

[0068] Furthermore, the first data set may be clustered and expanded, as follows:

[0069] (41) According to the first training set, set the initial membership.

[0070] Randomly initialize the membership matrix U = [U il ],U il Indicates that sample i belongs to the cluster with sample l as the cluster center, and 0 il <1, k represents the number of cluster centers.

[0071] (42) According to the initial membership, the initial cluster center is set.

[0072] Specifically, according to the initial membership, the initial cluster center is set to satisfy the following formula:

[0073] ​

[0074] Among them, ρ c represents the probability of taking sample C as the initial cluster center, n represents the number of samples, x i represents the i-th sample, Ω(x i ) represents the region formed by the combination of the i-th sample and the k similar samples, d(·) represents the Euclidean distance, U il represents the membership degree of the i-th sample to the l-th cluster center, m represents the fuzzy index, specifically a constant of (1,∞). In this embodiment, m=2, x c represents the cluster center, f(·) represents the value function, satisfying

[0075] Sample C according to ρ c To sort, ρ c The larger the value, the greater the probability that sample C is the initial cluster center, and the top k corresponding samples are used as the initial cluster centers.

[0076] (43) The clustering loss function value is obtained through the initial cluster center.

[0077] According to the initial cluster center, the clustering loss function value is calculated to satisfy the following formula:

[0078]

[0079] Among them, J c represents the clustering loss function value of the c-th cluster center, θ represents the clustering hesitation, which can be adjusted according to the clustering effect and actual requirements. In this embodiment, θ=0.5.

[0080] (44) Adjust the membership degree according to the clustering loss function value to obtain the best cluster center

[0081] (45) Fuzzy classification is performed on the first training set according to the optimal cluster center to obtain the classification result

[0082] If the clustering loss function value is less than the preset value or reaches the convergence condition, the iteration is stopped, the current clustering result is used as the final clustering result, the best clustering center is obtained, and the training set is fuzzy classified.

[0083] If the clustering loss function value is greater than the preset value and the convergence condition is not reached, the initialized membership matrix is ​​updated, and the update formula satisfies:

[0084] Among them, m represents the fuzzy index, U il It represents the membership degree of the i-th sample to the l-th cluster center, and d(·) represents the Euclidean distance.

[0085] (46) Obtain the second training set based on the classification results

[0086] (5) Training the convolutional network based on the second training set

[0087] S2, for the monitored area, obtain the image to be detected through the image acquisition device.

[0088] The embodiment of the present invention is to detect drones, so the acquired image is required to include at least one target object, and the target object at least includes the drone. In other words, the image to be detected cannot be a pure background image, which will bring unnecessary computing power burden to subsequent calculations.

[0089] S3, performing denoising on the image to be detected to obtain the original image.

[0090] It should be noted that the denoising method is as described above and will not be repeated here.

[0091] S4, input the original image into the pre-trained convolutional network for grid division to obtain multiple grid images.

[0092] S5, for each grid image, predict the position of the target object based on the first frame and the second frame to obtain the position information of the drone.

[0093] In specific implementation, step S5 includes:

[0094] Performing cluster analysis on the borders of the sample images in the first training set or the second training set to obtain a first border;

[0095] The first frame is translated and scaled to obtain a second frame;

[0096] For each grid image, the position is predicted based on the second frame to obtain the position information of the drone.

[0097] From the above description, it can be known that by implementing the embodiment of the present invention, a convolutional network is used to perform target detection on the image acquired by the image acquisition device, so as to predict the position information of the drone, thereby realizing effective detection and identification of the drone; based on the obtained drone position information, corresponding countermeasures can be taken against the drone, so as to provide protection for personal privacy protection, aviation safety, etc., and can also effectively combat the abuse of drones.

[0098] Based on the same inventive concept, an embodiment of the present invention provides a drone detection and identification system, such as Figure 2 As shown, the system includes an image acquisition device and a data processing device.

[0099] As a specific implementation method, a data processing device such as Figure 3As shown, the system may include: one or more processors 101, one or more input devices 102, one or more output devices 103 and a memory 104, wherein the processors 101, the input devices 102, the output devices 103 and the memory 104 are interconnected via a bus 105. The memory 104 is used to store a computer program, wherein the computer program includes program instructions, and the processor 101 is configured to call the program instructions to execute the method of the above method embodiment.

[0100] It should be understood that in the embodiment of the present invention, the processor 101 may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0101] The input device 102 may include a keyboard, etc., and the output device 103 may include a display (LCD, etc.), a speaker, etc.

[0102] The memory 104 may include a read-only memory and a random access memory, and provides instructions and data to the processor 101. A portion of the memory 104 may also include a non-volatile random access memory. For example, the memory 104 may also store information on the device type.

[0103] In a specific implementation, the processor 101, input device 102, and output device 103 described in the embodiment of the present invention can execute the implementation method described in the embodiment of the drone detection and identification method provided in the embodiment of the present invention, which will not be repeated here.

[0104] Alternatively, if Figure 4 As shown, the embodiment of the present invention further provides another data processing device, including:

[0105] Network training unit, used to build and train convolutional networks;

[0106] An image acquisition unit is used to acquire images to be detected by an image acquisition device for the monitored area; each image to be detected includes at least one target object, and the target object includes at least a drone;

[0107] A denoising unit, used for performing denoising processing on the image to be detected to obtain an original image;

[0108] A division unit, used for inputting the original image into a pre-trained convolutional network for grid division to obtain a plurality of grid images;

[0109] The position prediction unit is used to predict the position of the target object based on the first frame and the second frame for each grid image, so as to obtain the position information of the drone.

[0110] Wherein, the network training unit is specifically used for:

[0111] Constructing a convolutional network; the convolutional network includes a convolutional module and a residual network module, the convolutional module includes a plurality of convolutional kernels, and each convolutional kernel is newly added with a Gabor filter;

[0112] Acquire sample images; the sample images include a first image and a second image, in which the drone maintains the same flying posture in the first image, and in which the drone adopts a different flying posture in the second image;

[0113] Performing denoising and filtering processing on the sample image to construct a first data set;

[0114] Performing clustering and expansion processing on the first data set to obtain a second training set;

[0115] The convolutional network is trained based on the second training set.

[0116] The prediction unit is specifically used for:

[0117] Performing cluster analysis on the borders of the sample images in the first training set or the second training set to obtain a first border;

[0118] Performing translation and scaling processing on the first frame to obtain a second frame;

[0119] For each grid image, position prediction is performed based on the second frame to obtain the position information of the drone.

[0120] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0121] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.

[0122] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of the present invention.

[0123] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0124] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0125] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for detecting and identifying a drone, characterized in that: include: For the monitoring area, the image to be detected is obtained through the image acquisition device; Each image to be detected includes at least one target object, and the target object includes at least a drone; Performing denoising on the image to be detected to obtain an original image; Inputting the original image into a pre-trained convolutional network for grid division to obtain a plurality of grid images; For each grid image, the position of the target object is predicted based on the first frame and the second frame to obtain the position information of the drone.

2. The drone detection and identification method according to claim 1, characterized in that: The method also includes constructing and training a convolutional network, specifically: Constructing a convolutional network; the convolutional network includes a convolutional module and a residual network module, the convolutional module includes a plurality of convolutional kernels, and each convolutional kernel is newly added with a Gabor filter; Acquire sample images; the sample images include a first image and a second image, in which the drone maintains the same flying posture in the first image, and in which the drone adopts a different flying posture in the second image; Performing denoising and filtering processing on the sample image to construct a first data set; Performing clustering and expansion processing on the first data set to obtain a second training set; The convolutional network is trained based on the second training set.

3. The drone detection and identification method according to claim 2, characterized in that: The second training set is obtained as follows: Setting an initial membership degree according to the first training set; According to the initial membership, setting the initial cluster center; Obtaining a clustering loss function value through the initial cluster center; Adjusting the membership degree according to the clustering loss function value to obtain the best cluster center; Performing fuzzy classification on the first training set according to the optimal cluster center to obtain a classification result; A second training set is obtained according to the classification result.

4. The method for detecting and identifying a drone according to claim 2, wherein: The specific location information of the drone is: Performing cluster analysis on the borders of the sample images in the first training set or the second training set to obtain a first border; Performing translation and scaling processing on the first frame to obtain a second frame; For each grid image, position prediction is performed based on the second frame to obtain the position information of the drone.

5. A drone detection and identification system, comprising an image acquisition device and a data processing device, characterized in that: The data processing device includes a processor, an input device, an output device and a memory, which are interconnected, wherein the memory is used to store a computer program, the computer program includes program instructions, and the processor is configured to call the program instructions to execute the method according to any one of claims 1 to 4.

6. A drone detection and identification system, comprising an image acquisition device and a data processing device, characterized in that: The data processing device comprises: An image acquisition unit is used to acquire an image to be detected by an image acquisition device for the monitored area; each image to be detected includes at least one target object, and the target object includes at least a drone; A denoising unit, used for performing denoising processing on the image to be detected to obtain an original image; A division unit, used for inputting the original image into a pre-trained convolutional network for grid division to obtain a plurality of grid images; The position prediction unit is used to predict the position of the target object based on the first frame and the second frame for each grid image, so as to obtain the position information of the drone.

7. The drone detection and identification system according to claim 6, characterized in that: The system also includes a network training unit for constructing and training a convolutional network, specifically: Constructing a convolutional network; the convolutional network includes a convolutional module and a residual network module, the convolutional module includes a plurality of convolutional kernels, and each convolutional kernel is newly added with a Gabor filter; Acquire sample images; the sample images include a first image and a second image, in which the drone maintains the same flying posture in the first image, and in which the drone adopts a different flying posture in the second image; Performing denoising and filtering processing on the sample image to construct a first data set; Performing clustering and expansion processing on the first data set to obtain a second training set; The convolutional network is trained based on the second training set.

8. The drone detection and identification system according to claim 7, characterized in that: The second training set is obtained as follows: Setting an initial membership degree according to the first training set; According to the initial membership, setting the initial cluster center; Obtaining a clustering loss function value through the initial cluster center; Adjusting the membership degree according to the clustering loss function value to obtain the best cluster center; Performing fuzzy classification on the first training set according to the optimal cluster center to obtain a classification result; A second training set is obtained according to the classification result.

9. The drone detection and identification system according to claim 8, characterized in that: The prediction unit is specifically used for: Performing cluster analysis on the borders of the sample images in the first training set or the second training set to obtain a first border; Performing translation and scaling processing on the first frame to obtain a second frame; For each grid image, position prediction is performed based on the second frame to obtain the position information of the drone.