Unmanned aerial vehicle identification and positioning method, electronic equipment, system and storage medium

Abstract

The invention relates to the technical field of anti-unmanned aerial vehicles, and discloses an unmanned aerial vehicle identification and positioning method, and the method comprises the steps: obtaining a time-frequency graph, a distance Doppler frequency spectrum and a visual image of a target unmanned aerial vehicle; respectively carrying out feature extraction on the time-frequency graph, the distance Doppler frequency spectrum and the visual image to obtain a time-frequency feature, a distance Doppler frequency feature and an image feature; performing feature fusion on the time-frequency feature, the distance Doppler frequency feature and the image feature to obtain an initial fusion feature; performing feature fusion on the initial fusion feature and the time-frequency feature to obtain a first fusion feature; performing feature fusion on the initial fusion feature and the distance Doppler frequency feature to obtain a second fusion feature; performing feature extraction on the first fusion feature and the second fusion feature to obtain first feature data and second feature data of the target unmanned aerial vehicle; and identifying and positioning the target unmanned aerial vehicle according to the first feature data and the second feature data. In this way, the missing report of the unmanned aerial vehicle is reduced.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of anti-UAV technology, and specifically to a method, electronic device, system, and storage medium for identifying and locating a UAV. Background Art

[0002] In recent years, as the cost of small consumer drones has continued to decline, these drones have been widely used in a variety of fields, including security, education and training, and agriculture. However, unauthorized drone access to the airspace of key areas poses a serious threat to security, including but not limited to airports, government agencies, and international conference venues. Identification and positioning technologies can effectively prevent drones from interfering with normal flight operations or infringing on the privacy of others. Therefore, accurate drone identification and positioning is crucial.

[0003] In the counter-drone field, core drone detection equipment primarily includes radio detection devices, radar, and optoelectronic devices. To reliably identify and locate all types of drones, the detection information from these devices must be fused. Currently, most fusion solutions utilize target-level fusion technology. This involves filtering the drone's position, speed, and category after the radio, radar, and optoelectronic devices detect it. However, target-level fusion technology places high demands on individual devices for target detection. If a single device fails to detect a drone, there's a lack of valid detection data to update the fusion results, potentially leading to underreporting of drones. Summary of the Invention

[0004] In view of the above problems, the embodiments of the present application provide a method, electronic device, system and storage medium for identifying and locating drones to reduce the underreporting of drones.

[0005] According to one aspect of an embodiment of the present application, a method for identifying and locating a drone is provided, the method comprising: obtaining a time-frequency graph, a range Doppler frequency spectrum, and a visual image of a target drone; performing feature extraction on the time-frequency graph, the range Doppler frequency spectrum, and the visual image, respectively, to obtain time-frequency features, range Doppler frequency features, and image features; performing feature fusion on the time-frequency features, the range Doppler frequency features, and the image features to obtain initial fusion features; performing feature fusion on the initial fusion features and the time-frequency features to obtain first feature data of the target drone; performing feature fusion on the initial fusion features and the range Doppler frequency features to obtain second feature data of the target drone; and identifying and locating the target drone based on the first feature data and the second feature data.

[0006] In an optional manner, the first feature data includes spectrum information, category information and a first existence confidence, and the second feature data includes position information and speed information; the initial fusion feature and the time-frequency feature are feature fused to obtain the first feature data of the target UAV, and the initial fusion feature and the range Doppler frequency feature are feature fused to obtain the second feature data of the target UAV, further including: feature fusion of the initial fusion feature and the time-frequency feature to obtain spectrum information, category information and a first existence confidence; feature fusion of the initial fusion feature and the range Doppler frequency feature to obtain position information and speed information; identifying and locating the target UAV based on the first feature data and the second feature data, further including: identifying the target UAV based on the spectrum information, category information and the first existence confidence; locating the target UAV based on the position information and speed information.

[0007] In an optional manner, identifying a target UAV based on spectrum information, category information and a first existence confidence further includes: obtaining environmental target information of a dynamic environment in which the target UAV is located; performing feature extraction on the environmental target information to obtain environmental target features; inputting the environmental target features into a false alarm confidence model to obtain a second existence confidence of the target UAV, wherein the false alarm confidence model is trained by sample environmental target information in a dynamic environment; determining a final existence confidence of the target UAV based on the first existence confidence and the second existence confidence; and identifying the target UAV based on the spectrum information, category information and the final existence confidence.

[0008] In an optional manner, the environmental target information includes a noise spectrum and radar false alarm target information, and the false alarm confidence model is obtained by training sample noise spectra and sample radar false alarm target information in a dynamic environment; feature extraction is performed on the environmental target information to obtain environmental target features, and the environmental target features are input into the false alarm confidence model to obtain a second existence confidence of the target UAV, further including: feature extraction on the noise spectrum and radar false alarm target information to obtain noise spectrum features and radar false alarm target information features; and the noise spectrum features and radar false alarm target information features are input into the false alarm confidence model to obtain a second existence confidence.

[0009] In an optional manner, the method further includes: querying whether there is preset category information identical to the category information in the feature database, wherein the feature database includes preset category information and time-frequency features of multiple drones, and the category information of each drone corresponds to the time-frequency features one-to-one; if the preset category information identical to the category information does not exist in the feature database, updating the category information and the time-frequency features of the target drone to the feature database; if the preset category information identical to the category information exists in the feature database, determining whether the time-frequency features corresponding to the preset category information are identical to the time-frequency features of the target drone; if the time-frequency features corresponding to the preset category information are different from the time-frequency features of the target drone, using the time-frequency features of the target drone to replace the time-frequency features corresponding to the preset category information in the feature database.

[0010] In an optional manner, obtaining a time-frequency diagram, a range Doppler frequency spectrum, and a visual image of a target UAV further includes: obtaining a time-frequency diagram, a range Doppler frequency spectrum, and a visual image of the target UAV at a first time; identifying the target UAV based on the spectrum information, the category information, and the first existence confidence, further includes: obtaining a new visual image of the target UAV at a second time, the new visual image being taken by an optoelectronic device after locating the target UAV based on the position information, wherein the second time and the first time belong to the same detection time period; updating the first existence confidence according to the new visual image to obtain an updated first existence confidence; and identifying the target UAV based on the spectrum information, the category information, and the updated first existence confidence.

[0011] In an optional manner, locating the target UAV based on the position information and speed information further includes: acquiring a first new visual image and a second new visual image of the target UAV at a third time, the first new visual image and the second new visual image being respectively taken by different optoelectronic devices at different positions after locating the target UAV based on the position information, wherein the third time belongs to the same detection time period as the first time and the second time; determining the new position information of the target UAV based on the first new visual image and the second new visual image; updating the position information based on the new position information to obtain the updated position information; and locating the target UAV based on the updated position information and speed information.

[0012] According to another aspect of an embodiment of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the method for identifying and locating a drone provided in any of the above embodiments.

[0013] According to another aspect of the embodiments of the present application, a drone identification and positioning system is provided, comprising a radio detection device, a radar, an optoelectronic device, and the electronic device provided in the above embodiments, wherein the electronic device is communicatively connected to the radio detection device, the radar, and the optoelectronic device, respectively; the radio detection device, the radar, and the optoelectronic device are respectively used to obtain a time-frequency map, a range Doppler frequency spectrum, and a visual image of a target drone, and respectively send the time-frequency map, the range Doppler frequency spectrum, and the visual image to the electronic device; the electronic device is used to receive the time-frequency map, the range Doppler frequency spectrum, and the visual image, and identify and locate the target drone based on the time-frequency map, the range Doppler frequency spectrum, and the visual image.

[0014] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for identifying and locating a drone provided in any of the above embodiments is implemented.

[0015] In the embodiment of the present application, the time-frequency graph, range Doppler frequency spectrum, and visual image of the target drone are extracted to obtain time-frequency features, range Doppler frequency features, and image features. Then, the time-frequency features, range Doppler frequency features, and image features are fused to obtain initial fused features. The initial fused features combine the main features of the time-frequency graph, range Doppler frequency spectrum, and visual image. The initial fused features can more comprehensively determine the characteristics of the target drone, thereby distinguishing the target drone from other targets and reducing the underreporting of the target drone. Furthermore, by fusion of the initial fused features with the time-frequency features to obtain first feature data of the target drone, and by fusion of the initial fused features with the range Doppler frequency features to obtain second feature data of the target drone, the dynamic characteristics of frequency and time in the time-frequency features can be fully utilized to improve the accuracy of the first feature data of the target drone, and the information such as distance and speed in the range Doppler frequency features can be used to improve the accuracy of the second feature data. Therefore, when the target drone is identified and located using the first feature data and the second feature data, the target drone can be more accurately identified and located.

[0016] The above description is only an overview of the technical solutions of the embodiments of the present application. In order to more clearly understand the technical means of the embodiments of the present application, they can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiments of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present application. In addition, the same reference symbols are used to represent the same components throughout the drawings. In the drawings:

[0018] Figure 1 A schematic diagram of the structure of a drone identification and positioning system provided in an embodiment of the present application is shown;

[0019] Figure 2 A schematic diagram showing a flow chart of a method for identifying and locating a drone provided in an embodiment of the present application is shown;

[0020] Figure 3 The following is a flowchart of the network architecture provided by the embodiment of the present application;

[0021] Figure 4 Shown Figure 2 Schematic diagram of the flow of sub-steps of step 160;

[0022] Figure 5 Shown Figure 2 Flow chart of the steps after step 160;

[0023] Figure 6 A schematic diagram showing a flow chart of a method for identifying and locating a drone provided in an embodiment of the present application is shown;

[0024] Figure 7 A schematic diagram of the structure of a drone identification and positioning device provided in an embodiment of the present application is shown;

[0025] Figure 8 A schematic structural diagram of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0026] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0027] Currently, most fusion solutions in the anti-drone field use target-level fusion technology. This involves fusing information such as the drone's position, velocity, and category after radio detection equipment, radar, and optoelectronic devices detect it using filtering techniques such as Kalman Filter (KF), Extended Kalman Filter (EKF), or Particle Filter (PF). While target-level fusion technology offers real-time performance and low computing power requirements, it places high demands on target detection on individual devices. If a single device fails to detect a drone, there's a lack of valid detection data to update the fusion results, potentially leading to underreporting of drones.

[0028] To address these issues, feature extraction can be performed on the raw data from drones detected by radar, radio detection equipment, and optoelectronic devices to obtain the raw data features of different devices. These raw data features are then fused to produce a fusion result (i.e., fusion features), which can then be used to identify and locate drones. Because the fusion result retains the key features of the raw data from different devices, it can fully leverage the advantages of each device, comprehensively grasp the characteristics of drones, and improve the accuracy and robustness of drone positioning and identification. Furthermore, even if the raw data from one device is lost or noisy, the raw data from other devices can still be supplemented, reducing false positives and false negatives for drones.

[0029] However, when fusing raw data features from different devices, while the primary characteristics of the original data are preserved, the lower-level physical features of the raw data, such as the harmonic details of radio signals and the Doppler texture of radar, may be smoothed out. These low-level physical features are crucial for accurately identifying and localizing drones. Furthermore, during the fusion process, high-dimensional modal features, such as the image characteristics of optoelectronic devices, may dominate the fusion results, weakening the raw data features of radio detection devices and radars. This will reduce the accuracy of drone identification and location.

[0030] Based on this, the fused features are fused twice with the original data features of the radio detection equipment and radar respectively, and then the UAV is identified and located through the features after the secondary fusion. While taking advantage of the advantages of the fused features, the integrity of the original data features of the radio detection equipment and radar is retained. The fused features and the original data features of the radio detection equipment and radar can be fully utilized to more accurately identify and locate the UAV.

[0031] Figure 1 The structure diagram of the drone identification and positioning system provided in the embodiment of the present application is shown as follows: Figure 1 As shown, the drone identification and positioning system 10 includes a radio detection device 11, a radar 12, an optoelectronic device 13, and an electronic device 14. The electronic device 14 is connected to the radio detection device 11, the radar 12, and the optoelectronic device 13, respectively, for communication, for example, via a network cable or Wi-Fi connection. The radio detection device 11, the radar 12, and the optoelectronic device 13 are respectively used to obtain a time-frequency map, a range-Doppler frequency spectrum, and a visual image of a target drone 15, and transmit the time-frequency map, the range-Doppler frequency spectrum, and the visual image to the electronic device 14. The electronic device 14 is used to receive the time-frequency map, the range-Doppler frequency spectrum, and the visual image, and identify and locate the target drone 15 based on the time-frequency map, the range-Doppler frequency spectrum, and the visual image.

[0032] The radio detection device 11 can receive communication signals between the target drone 15 and the remote controller controlling the target drone 15, and process the communication signals to obtain a time-frequency map of the target drone 15. The radar 12 can be a two-dimensional phased array radar, which can obtain Doppler frequency information, length information, altitude information, and speed information of the target drone 15. It has advantages such as high-precision target search, strong anti-interference capabilities, and fast response time. The optoelectronic device 13 uses an optical sensor (such as a camera) to obtain a visual image of the target drone 15. The electronic device 14 can be a smartphone, tablet computer, computer, server, or other device.

[0033] Figure 2 The flowchart of the method for identifying and locating a drone provided in an embodiment of the present application is shown. Figure 3 The network architecture flow chart provided by the embodiment of the present application is shown, wherein the method is composed of Figure 1 The electronic device 14 shown in the embodiment is executed. Figure 2 and Figure 3 As shown, the method includes the following steps:

[0034] Step 110: Acquire the time-frequency map, range Doppler frequency spectrum, and visual image of the target UAV.

[0035] A time-frequency diagram can provide information about how the frequency of a signal changes over time.

[0036] The range Doppler frequency spectrum reflects the range and Doppler frequency (speed) information of the target UAV 15. Specifically, after receiving the echo signal, the radar 12 first performs a Fast Fourier Transform (FFT) on the echo signal of a single pulse in the pulse dimension of the echo signal, converting the time domain signal into the frequency domain. Next, an FFT is performed on the multiple pulse echo sequences in the same range unit in the range dimension of the echo signal to obtain the Doppler frequency spectrum of the range unit. Finally, the Doppler frequency spectrum of each range unit is arranged in columns to obtain the range Doppler frequency spectrum.

[0037] The visual image reflects information such as the appearance and position of the target UAV 15. The visual image can be a visible light image or an infrared image of the target UAV.

[0038] Step 120: extract features from the time-frequency graph, the range Doppler frequency spectrum, and the visual image to obtain time-frequency features, range Doppler frequency features, and image features.

[0039] Time-frequency features include the target drone's signal frequency and energy distribution. Range-Doppler frequency features include motion characteristics such as the target drone's distance and speed. Image features include its appearance, such as its shape, size, and color. Time-frequency and image features can be used to determine information such as the target drone's spectrum and category, while range-Doppler frequency features can be used to determine information such as its location and speed.

[0040] Different feature extraction models can be used to extract features from the time-frequency graph, range-Doppler frequency spectrum, and visual image, respectively. The feature extraction model can be obtained by training a convolutional neural network (CNN), such as ResNet18.

[0041] Specifically, the time-frequency graphs, range Doppler frequency spectra and visual images of a large number of drones can be used to train convolutional neural networks 1, 2 and 3 respectively to obtain feature extraction models 1, 2 and 3, and then the time-frequency graphs, range Doppler frequency spectra and visual images are input into feature extraction models 1, 2 and 3 respectively, and feature extraction models 1, 2 and 3 output time-frequency features, range Doppler frequency features and image features respectively.

[0042] Step 130: Fusing the time-frequency features, the range-Doppler frequency features, and the image features to obtain initial fused features.

[0043] The time-frequency features, range Doppler frequency features and image features can be spliced first, and then the spliced features are input into the feature fusion CNN network to obtain the initial fusion features.

[0044] By splicing the data features of each device, the features from radio detection equipment, radar and optoelectronic equipment can be merged together. In this way, the feature fusion CNN network can use the initial fusion features to more comprehensively determine the characteristics of the target UAV, so that these features can be used to distinguish the target UAV from other targets, thereby improving the accuracy of target UAV identification and positioning.

[0045] Step 140: Perform feature fusion on the initial fusion features and the time-frequency features to obtain first feature data of the target UAV.

[0046] The first feature data may include spectrum information, category information, and a first existence confidence level of the target UAV, wherein the first existence confidence level is the probability of the target UAV existing.

[0047] The initial fused features and time-frequency features can be concatenated, and then the concatenated initial fused features and time-frequency features can be input into the category and spectrum CNN network. The category and spectrum CNN network then predicts the spectrum and category of the target drone based on the concatenated initial fused features and time-frequency features, outputting the target drone's spectrum information, category information, and a first confidence level. When the category and spectrum CNN network outputs the target drone's category (for example, if it outputs the target drone as a first-category drone or a second-category drone), it also outputs the probability that the target drone belongs to "drone." This probability is the first confidence level.

[0048] Based on the initial fusion features, the category and spectrum CNN network can preliminarily predict the spectrum, category, and first existence confidence of the target UAV. The time-frequency features capture the dynamic characteristics of the frequency and time of the target UAV. These dynamic features can further predict the spectrum and category of the target UAV. Therefore, the initial fusion features and time-frequency features can improve the prediction accuracy of the category and spectrum CNN network for the spectrum, category, and existence confidence of the target UAV.

[0049] Step 150: Perform feature fusion on the initial fusion feature and the range Doppler frequency feature to obtain second feature data of the target UAV.

[0050] The second feature data includes the location information and speed information of the target UAV, wherein the location information can be represented by two-dimensional coordinates.

[0051] The initial fusion features and the range Doppler frequency features can be spliced first, and then the spliced initial fusion features and the range Doppler frequency features can be input into the motion CNN network, so that the motion CNN network can predict the position and speed of the target UAV based on the spliced initial fusion features and the range Doppler frequency features, and output the position information and speed information of the target UAV.

[0052] The motion CNN network can initially predict the target drone's position and velocity based on the initial fused features. However, since the range-Doppler frequency features may be partially lost in the initial fused features, this can lead to large errors in the initial predicted position and velocity. The range-Doppler frequency features contain information such as the target drone's distance and velocity. By combining the initial fused features with the range-Doppler frequency features, the motion CNN network can more accurately predict the target drone's position and velocity.

[0053] Step 160: Identify and locate the target UAV based on the first feature data and the second feature data.

[0054] Specifically, the target drone is identified based on its spectrum information, category information and first existence confidence, and is located based on its position information and speed information to determine the spectrum, category, position and speed of the target drone, thereby tracking and striking the drone.

[0055] In the embodiment of the present application, the time-frequency graph, range Doppler frequency spectrum, and visual image of the target drone are extracted to obtain time-frequency features, range Doppler frequency features, and image features. Then, the time-frequency features, range Doppler frequency features, and image features are fused to obtain initial fused features. The initial fused features combine the main features of the time-frequency graph, range Doppler frequency spectrum, and visual image. The initial fused features can more comprehensively determine the characteristics of the target drone, thereby distinguishing the target drone from other targets and reducing the underreporting of the target drone. Furthermore, by fusion of the initial fused features with the time-frequency features to obtain first feature data of the target drone, and by fusion of the initial fused features with the range Doppler frequency features to obtain second feature data of the target drone, the dynamic characteristics of frequency and time in the time-frequency features can be fully utilized to improve the accuracy of the first feature data of the target drone, and the information such as distance and speed in the range Doppler frequency features can be used to improve the accuracy of the second feature data. Therefore, when the target drone is identified and located using the first feature data and the second feature data, the target drone can be more accurately identified and located.

[0056] In some complex, high-noise, and poorly consistent data scenarios, there are often other objects or flying objects besides drones, including but not limited to trees, balloons, birds, kites, etc. In addition, there may also be noise spectrum and communication signals between other devices in the scene, which will affect the identification of target drones by radio detection equipment, radar, or optoelectronic equipment, resulting in false alarms of target drones. Therefore, in order to reduce false alarms of target drones, this application also provides an embodiment, optionally, Figure 4 A flowchart of the sub-steps of step 160 is shown, as shown in FIG. Figure 4 As shown, step 160 also includes the following steps:

[0057] Step 161: Obtain environmental target information of the dynamic environment in which the target UAV is located.

[0058] Dynamic environments are those monitored by radio detection equipment, radar, and optoelectronic devices. Environmental target information can include noise spectrum and radar false alarm target information. The noise spectrum can include smartphone communication signals, remote control signals from other drones, and other signals with frequencies close to the target drone, received by the radio detection equipment. Radar false alarm target information can include ground clutter, treetop motion echoes, and other signals.

[0059] Step 162: Extract features of the environmental target information to obtain environmental target features.

[0060] Specifically, feature extraction is performed on the noise spectrum and radar false alarm target information to obtain noise spectrum features and radar false alarm target information features.

[0061] Step 163: Input the environmental target feature into a false alarm confidence model to obtain a second existence confidence of the target UAV, wherein the false alarm confidence model is trained by sample environmental target information in a dynamic environment.

[0062] Specifically, the false alarm confidence model is trained using sample noise spectra and sample radar false alarm target information in a dynamic environment. During operation, radio detection equipment, radar, and optoelectronic devices regularly collect a large amount of sample environmental target information in the dynamic environment. This sample environmental target information can include sample noise spectra and sample radar false alarm target information. By extracting features from these sample noise spectra and radar false alarm target information and training a deep learning network with these extracted features, a false alarm confidence model can be derived. By inputting the noise spectrum features and radar false alarm target information features into the false alarm confidence model, the false alarm confidence model can determine the probability that the target drone is not a "drone" based on these noise spectrum features and radar false alarm target information features. This allows the second confidence level of the target drone's presence to be determined based on the probability that the target drone is not a "drone." For example, if the probability that the target drone is not a "drone" is 20%, the second confidence level of the target drone's presence is 80%.

[0063] Step 164: Determine a final existence confidence level of the target UAV based on the first existence confidence level and the second existence confidence level.

[0064] Specifically, the first existence confidence level and the second existence confidence level may be weighted to obtain the final existence confidence level, or the first existence confidence level and the second existence confidence level may be subjected to Bayesian inference to obtain the final existence confidence level.

[0065] Step 165: Identify the target UAV based on the spectrum information, category information, and final existence confidence.

[0066] Figure 4In the illustrated embodiment, a false alarm confidence model is obtained by training a deep learning network with sample environmental target information, and environmental target features are obtained by extracting features from the environmental target information of the target UAV. The environmental target features are input into the false alarm confidence model so that the false alarm confidence model can output a second existence confidence, that is, the probability that the target UAV is a "drone". In this way, a more accurate existence confidence of the target UAV can be obtained through the first existence confidence and the second existence confidence, so that the target UAV can be identified more accurately and false alarms of the target UAV can be reduced.

[0067] In some cases, a feature database can be set up on a cloud server or radio detection device to store the category information of various drones and the time-frequency features corresponding to the category information. The radio detection device compares the time-frequency features of the target drone 1 with the time-frequency features stored in the feature database to determine the category information of the target drone 1. By combining the category information of the target drone 1 obtained by the radio detection device with the category information of the target drone 1 output by the drone identification and positioning system, the ability to identify the target drone 1 can be improved.

[0068] However, when the time-frequency characteristics of the target drone 1 are not stored in the feature database, the radio detection device cannot determine the category information of the target drone 1. Due to the existence of the range Doppler frequency characteristics and image characteristics, the drone identification and positioning system can still determine the category information of the target drone 1 based on the range Doppler frequency characteristics and image characteristics. Therefore, in order to improve the recognition capability of the target drone, the present application also provides an embodiment, optionally, Figure 5 A flowchart of the steps following step 160 is shown. Figure 5 As shown, after step 160, the following steps are also included:

[0069] Step 171 : Query whether there is preset category information identical to the category information in the feature database. If so, execute step 173 ; otherwise, execute step 172 .

[0070] Step 172: Update the category information and the time-frequency features of the target UAV to the feature database.

[0071] Step 173: Determine whether the time-frequency characteristics corresponding to the preset category information are the same as the time-frequency characteristics of the target UAV. If so, end; otherwise, execute step 174.

[0072] Step 174: Use the time-frequency features of the target UAV to replace the time-frequency features corresponding to the preset category information in the feature database.

[0073] The feature database includes preset category information and time-frequency features of multiple drones, and the preset category information of each drone corresponds to the time-frequency features one by one.

[0074] In order to improve the recognition capability of the target UAV 1 , after obtaining the category information of the target UAV 1 , it is possible to first compare whether the feature database stores preset category information that is the same as the category information of the target UAV 1 .

[0075] If the feature database does not store preset category information that is identical to the category information of the target UAV 1, the category information and time-frequency features of the target UAV 1 are stored in the feature database so that the radio detection device can identify the target UAV 1 through the feature database during the subsequent identification process; if the feature database stores preset category information that is identical to the category information of the target UAV 1, the time-frequency features corresponding to the preset category information in the feature database are compared with the time-frequency features of the target UAV 1.

[0076] If the time-frequency features corresponding to the preset category information are the same as the time-frequency features of the target UAV 1, the time-frequency features corresponding to the preset category information will not be processed; if the time-frequency features corresponding to the preset category information are different from the time-frequency features of the target UAV 1, it indicates that the time-frequency features corresponding to the preset category information stored in the feature database may be erroneous, resulting in the inability to identify the target UAV 1 based on the time-frequency features. Therefore, the time-frequency features corresponding to the preset category information are replaced with the time-frequency features of the target UAV 1, so that the radio detection equipment can identify the target UAV 1 in the subsequent identification process.

[0077] The time-frequency characteristics of the target UAV are updated to the feature database through category information. When the category information and time-frequency characteristics of the target UAV do not exist in the feature database, resulting in the radio detection equipment being unable to determine the category information of the target UAV based on the time-frequency characteristics, the category information and time-frequency information of the target UAV can be updated to the feature database in a timely manner, so that the radio detection equipment can subsequently determine the category information of the target UAV based on the time-frequency characteristics of the target UAV, thereby being able to combine it with the category information of the target UAV obtained by the UAV identification and positioning system to improve the recognition capability of the target UAV.

[0078] In order to reduce false alarms of target drones, the present application also provides an implementation method, optionally, Figure 6 A flow chart of a method for identifying and locating a drone provided in an embodiment of the present application is shown. Figure 6 As shown, the method for identifying and locating a drone includes the following steps:

[0079] Step 210: Acquire the time-frequency diagram, range Doppler frequency spectrum, and visual image of the target UAV as soon as possible.

[0080] In this embodiment of the present application, the target drone's time-frequency graph, range Doppler frequency spectrum, and visual image are acquired during different detection time periods to continuously identify and locate the target drone, reducing missed detections and enabling tracking of the drone. The different detection time periods can be of the same duration.

[0081] Step 220: extract features from the time-frequency graph, the range Doppler frequency spectrum, and the visual image to obtain time-frequency features, range Doppler frequency features, and image features.

[0082] Step 230: Fusing the time-frequency features, the range-Doppler frequency features, and the image features to obtain initial fused features.

[0083] Step 240: Perform feature fusion on the initial fusion features and the time-frequency features to obtain first feature data of the target UAV, where the first feature data includes spectrum information, category information, and a first existence confidence level.

[0084] Step 250: Perform feature fusion on the initial fusion feature and the range Doppler frequency feature to obtain second feature data of the target UAV, where the second feature data includes position information and speed information.

[0085] The specific implementation process and technical effects of steps 210 to 250 can be referred to Figure 2 The implementation process and technical effects of steps 110 to 150 shown in the embodiment will not be repeated here.

[0086] Step 260: Acquire a new visual image of the target UAV at a second time, where the new visual image is captured by the optoelectronic device after locating the target UAV based on the position information, wherein the second time and the first time belong to the same detection time period.

[0087] The new visual image can be a visible light image or an infrared image of the target drone. The optoelectronic device used to capture the new visual image can be the same optoelectronic device used to capture the visual image in step 210. That is, the visual image in step 210 and the new visual image in step 260 are captured sequentially using the same optoelectronic device. Of course, the optoelectronic device used to capture the new visual image can also be another optoelectronic device.

[0088] For example, during the first time of detection period 1, the target drone's time-frequency graph, range-Doppler frequency spectrum, and visual image are acquired, and the target drone is identified and located for the first time based on the time-frequency graph, range-Doppler frequency spectrum, and visual image. After the target drone's location information is determined for the first time, the optoelectronic device can be controlled to rotate based on this location information so that the optoelectronic device captures the target drone during the second time of detection period 1, obtaining a new visual image of the target drone. After capturing the new visual image, the optoelectronic device transmits the new visual image to the electronic device.

[0089] Step 270: Update the first existence confidence according to the new visual image to obtain an updated first existence confidence.

[0090] After receiving and processing the new visual image, the electronic device can directly identify the target drone based on the new visual image to obtain a second existence confidence level, that is, the probability of the target drone's existence. The first existence confidence level can then be updated based on the second existence confidence level to obtain a third existence confidence level, that is, the updated first existence confidence level. Specifically, the first existence confidence level and the second existence confidence level can be fused. For example, the first existence confidence level and the second existence confidence level can be input into a trained Bayesian model, and the Bayesian model outputs the third existence confidence level. Alternatively, the first existence confidence level and the second existence confidence level can be fused using the Dempster-Shafer Theory (DST) to obtain the third existence confidence level.

[0091] Step 280: Identify the target UAV based on the spectrum information, category information, and the updated first existence confidence level.

[0092] After each identification and positioning of the target drone, the target drone is located using the target drone's position information, and a new visual image of the target drone is re-photographed. The first existence confidence level obtained by each identification and positioning of the target drone can be updated based on the new visual image, thereby improving the accuracy of the first existence confidence level and reducing false alarms of the target drone.

[0093] In order to more accurately locate the target drone, this application also provides an implementation method. Figure 6 As shown in the figure, after step 280, the following steps are also included:

[0094] Step 290: Acquire a first new visual image and a second new visual image of the target UAV at a third time, where the first new visual image and the second new visual image are respectively taken by different optoelectronic devices at different positions after locating the target UAV based on position information, wherein the third time belongs to the same detection time period as the first time and the second time.

[0095] Among them, one of the first photoelectric device for capturing the first new visual image and the second photoelectric device for capturing the second new visual image can be the photoelectric device for capturing the new visual image in step 260, or the photoelectric device for capturing the visual image in step 210, or other photoelectric devices.

[0096] For example, after identifying and locating the target drone for the first time during detection time period 1, i.e., determining the target drone's location information for the first time, the first optoelectronic device and the second optoelectronic device can be controlled to rotate based on the location information, so that the first optoelectronic device and the second optoelectronic device respectively capture the target drone at a third time during detection time period 1, thereby obtaining a first new visual image and a second new visual image of the target drone. After capturing and obtaining the first and second new visual images, the first and second optoelectronic devices respectively transmit the first and second new visual images to the electronic device.

[0097] Step 300: Determine new position information of the target UAV based on the first new visual image and the second new visual image.

[0098] Based on the first new visual image and the second new visual image, new position information of the target UAV can be determined using three-dimensional positioning technology.

[0099] Step 310: Update the location information according to the new location information to obtain updated location information.

[0100] Step 320: Position the target UAV based on the updated position information and speed information.

[0101] After each identification and positioning of the target UAV, the target UAV is positioned using the position information of the target UAV, and a first new visual image and a second new visual image of the target UAV are re-photographed using two different optoelectronic devices. The new position information of the target UAV can be determined based on the first new visual image and the second new visual image. The position information obtained by each identification and positioning of the target UAV is updated using the new position information, which can improve the accuracy of the position information, thereby enabling the target UAV to be positioned more accurately.

[0102] Figure 7 The following is a schematic diagram showing the structure of the identification and positioning device of the drone provided in the embodiment of the present application. Figure 7 As shown, the UAV identification and positioning device 400 includes: an acquisition module 410, a feature extraction module 420, a first feature fusion module 430, a second feature fusion module 440, a third feature fusion module 450 and an identification and positioning module 460.

[0103] The acquisition module 410 is used to acquire the time-frequency diagram, range Doppler frequency spectrum and visual image of the target UAV; the feature extraction module 420 is used to perform feature extraction on the time-frequency diagram, range Doppler frequency spectrum and visual image respectively to obtain time-frequency features, range Doppler frequency features and image features; the first feature fusion module 430 is used to perform feature fusion on the time-frequency features, range Doppler frequency features and image features to obtain initial fusion features; the second feature fusion module 440 is used to perform feature fusion on the initial fusion features and the time-frequency features to obtain first feature data of the target UAV; the third feature fusion module 450 is used to perform feature fusion on the initial fusion features and the range Doppler frequency features to obtain second feature data of the target UAV; the identification and positioning module 460 is used to identify and locate the target UAV based on the first feature data and the second feature data.

[0104] The drone identification and positioning device 400 provided in this embodiment is used to execute the technical solution of the drone identification and positioning method in the aforementioned method embodiment. Its implementation principle and technical effects are similar and will not be repeated here.

[0105] Figure 8 A schematic structural diagram of an electronic device provided in an embodiment of the present application is shown. The specific embodiment of the present application does not limit the specific implementation of the electronic device.

[0106] like Figure 8 As shown, the electronic device 14 may include a processor 142 and a memory 144 .

[0107] The memory 144 is used to store a computer program 146. The memory 144 may include a high-speed RAM memory, or may also include a non-volatile memory, such as at least one disk memory. The computer program 146 may include computer-executable instructions.

[0108] The processor 142 is configured to execute the computer program 146 to implement the above-mentioned drone identification and positioning method embodiment.

[0109] Processor 142 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in an electronic device may be processors of the same type, such as one or more CPUs, or may be processors of different types, such as one or more CPUs and one or more ASICs.

[0110] An embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the embodiment of the above-mentioned drone identification and positioning method is implemented.

[0111] An embodiment of the present application provides a computer program that can be executed by a processor to implement the above-mentioned drone identification and positioning method embodiment.

[0112] An embodiment of the present application provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the above-mentioned drone identification and positioning method embodiment.

[0113] In the several embodiments provided in this application, if any function is implemented in the form of a software function module / unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, part or all of the technical solution of this application 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 enabling a computer device (which can be a personal computer, server or other electronic device) to execute all or part of the steps of the method described in each embodiment of this application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), magnetic disk or optical disk, and other media that can store computer program code.

[0114] The algorithm or demonstration provided here are not inherently relevant to any particular computer, virtual system or other equipment. Various general purpose systems can also be used together with the teachings based on this. According to the above description, it is obvious that the structure required for constructing this type of system. In addition, the present application embodiment is not directed to any specific programming language yet. It should be understood that various programming languages can be utilized to realize the content of the present application described here, and the above description of specific languages is for the purpose of disclosing the best mode of implementation of the present application.

[0115] It should be noted that the above embodiments illustrate rather than limit the present application, and that a person skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between brackets should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present application may be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In claims that list several means, several units or modules of these means may be embodied by the same item of hardware. The use of the words first, second, and third etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments should not be understood as limiting the order of execution unless otherwise specified.

[0116] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for identifying and locating a drone, characterized in that: The method comprises: Obtain time-frequency diagram, range Doppler frequency spectrum and visual image of target UAV; Performing feature extraction on the time-frequency graph, the range Doppler frequency spectrum, and the visual image, respectively, to obtain time-frequency features, range Doppler frequency features, and image features; Performing feature fusion on the time-frequency feature, the range Doppler frequency feature, and the image feature to obtain an initial fusion feature; Performing feature fusion on the initial fusion feature and the time-frequency feature to obtain first feature data of the target UAV; Performing feature fusion on the initial fusion feature and the range Doppler frequency feature to obtain second feature data of the target UAV; The target UAV is identified and located according to the first feature data and the second feature data.

2. The method according to claim 1, characterized in that The first feature data includes spectrum information, category information and a first existence confidence level, and the second feature data includes position information and speed information; The step of performing feature fusion on the initial fusion feature and the time-frequency feature to obtain first feature data of the target UAV, and performing feature fusion on the initial fusion feature and the range Doppler frequency feature to obtain second feature data of the target UAV further includes: Performing feature fusion on the initial fusion feature and the time-frequency feature to obtain the spectrum information, the category information, and the first existence confidence; Performing feature fusion on the initial fusion feature and the range Doppler frequency feature to obtain the position information and the velocity information; The identifying and locating the target UAV according to the first feature data and the second feature data further includes: Identifying the target UAV according to the spectrum information, the category information, and the first existence confidence level; The target UAV is positioned according to the position information and the speed information.

3. The method according to claim 2, characterized in that The identifying the target UAV according to the spectrum information, the category information, and the first existence confidence level further includes: Obtaining environmental target information of the dynamic environment in which the target UAV is located; Extracting features from the environmental target information to obtain environmental target features; Inputting the environmental target feature into a false alarm confidence model to obtain a second existence confidence of the target UAV, wherein the false alarm confidence model is trained using sample environmental target information in a dynamic environment; Determining a final existence confidence level of the target UAV according to the first existence confidence level and the second existence confidence level; The target UAV is identified according to the spectrum information, the category information and the final existence confidence.

4. The method according to claim 3, characterized in that The environmental target information includes a noise spectrum and radar false alarm target information, and the false alarm confidence model is obtained by training sample noise spectrum and sample radar false alarm target information in a dynamic environment; The feature extraction of the environmental target information to obtain environmental target features, and inputting the environmental target features into a false alarm confidence model to obtain a second existence confidence of the target UAV further includes: Extracting features from the noise spectrum and the radar false alarm target information to obtain noise spectrum features and radar false alarm target information features; The noise spectrum characteristics and the radar false alarm target information characteristics are input into the false alarm confidence model to obtain the second existence confidence.

5. The method according to claim 2, characterized in that The method further comprises: Querying a feature database for preset category information identical to the category information, wherein the feature database includes preset category information and time-frequency features of a plurality of drones, and the category information of each drone corresponds to the time-frequency features one-to-one; If there is no preset category information identical to the category information in the feature database, updating the category information and the time-frequency features of the target UAV to the feature database; If there is preset category information identical to the category information in the feature database, determining whether the time-frequency feature corresponding to the preset category information is identical to the time-frequency feature of the target UAV; If the time-frequency feature corresponding to the preset category information is different from the time-frequency feature of the target UAV, the time-frequency feature corresponding to the preset category information in the feature database is replaced by the time-frequency feature of the target UAV.

6. The method according to claim 2, characterized in that The obtaining of the time-frequency graph, range Doppler frequency spectrum and visual image of the target UAV further includes: Acquire a time-frequency diagram, a range Doppler frequency spectrum, and a visual image of the target UAV as soon as possible; The identifying the target UAV according to the spectrum information, the category information, and the first existence confidence level further includes: Acquire a new visual image of the target UAV at a second time, where the new visual image is captured by the optoelectronic device after locating the target UAV based on the position information, wherein the second time and the first time belong to the same detection time period; updating the first existence confidence level according to the new visual image to obtain an updated first existence confidence level; The target UAV is identified according to the spectrum information, the category information and the updated first existence confidence.

7. The method according to claim 6, characterized in that Positioning the target UAV according to the position information and the speed information further includes: Acquire a first new visual image and a second new visual image of the target UAV at a third time, where the first new visual image and the second new visual image are respectively captured by different optoelectronic devices at different locations after locating the target UAV based on the position information, wherein the third time, the first time, and the second time belong to the same detection time period; Determine new position information of the target UAV based on the first new visual image and the second new visual image; updating the location information according to the new location information to obtain updated location information; The target UAV is positioned according to the updated position information and the speed information.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory, wherein: The processor executes the computer program to implement the drone identification and positioning method according to any one of claims 1 to 7.

9. A drone identification and positioning system, characterized in that: The drone identification and positioning system comprises a radio detection device, a radar, an optoelectronic device, and the electronic device according to claim 8, wherein the electronic device is communicatively connected to the radio detection device, the radar, and the optoelectronic device respectively; The radio detection device, the radar, and the optoelectronic device are respectively used to obtain a time-frequency diagram, a range Doppler frequency spectrum, and a visual image of the target UAV, and respectively send the time-frequency diagram, the range Doppler frequency spectrum, and the visual image to the electronic device; The electronic device is used to receive the time-frequency diagram, the range Doppler frequency spectrum and the visual image, and identify and locate the target UAV based on the time-frequency diagram, the range Doppler frequency spectrum and the visual image.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for identifying and locating a drone according to any one of claims 1 to 7 is implemented.

Images