Unmanned aerial vehicle detection method and device, storage medium and electronic equipment
By adding noise in the gradient range and performing image fusion and time-frequency conversion, the problem of poor recognition of graph signals in real environments is solved, and more accurate and fast drone category recognition is achieved.
Patent Information
- Application Number
- CN202510282531.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-11
- Publication Date
- 2025-08-01
AI Technical Summary
In the prior art, the level of detection and recognition based on the graph signal is relatively low, which is affected by the interference of real environmental noise, resulting in poor recognition effect.
By acquiring the drone's graph signal data, adding gradient noise within the target gradient range, forming a noise image and fusing it, then performing time-frequency graph conversion, and using the time-frequency image to identify the drone category.
It improves the accuracy and speed of drone detection and recognition, can better identify drone categories in real environments, and enhances the universality of recognition and signal characteristic representation.
Smart Images
Figure CN120408284A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of signal processing, and particularly relates to a method, device, storage medium and electronic device for detecting unmanned aerial vehicles (UAVs). Background Art
[0002] With the increasing awareness of the potential hazards of UAVs, UAV detection technology has naturally become a hot topic of concern, and current research is also developing in different directions. The main concept of UAV detection is to use certain characteristics of UAVs to separate them from other objects. The mainstream directions include detection means based on vision, sound, radar, radio frequency (RF), and WiFi.
[0003] The UAV detection technology based on radio signals is not restricted by visibility and urban environment. The RF-based UAV detection system is designed to detect the radio signals between UAVs and their control devices. Since UAVs usually communicate with their control devices through specific wireless channels to transmit information such as the operator's operation instructions and video streams, UAVs can be detected according to the characteristics of wireless signals, and the UAV video transmission signal is converted into a time-frequency diagram for UAV detection and identification. However, the real environment is different from the ideal environment, and there are often inevitable noise interferences, which will affect the quality of the video transmission signal. In this case, the information carried on the converted time-frequency diagram is not sufficient for specific identification, resulting in a reduction in the UAV detection and identification level. Summary of the Invention
[0004] The main purpose of the present application is to provide a method, device, storage medium and electronic device for detecting UAVs, aiming to solve the problem of the low level of UAV detection and identification based on video transmission signals in the prior art.
[0005] To achieve the above purpose, the technical solutions adopted in the embodiments of the present application are as follows:
[0006] In a first aspect, an embodiment of the present application provides a method for detecting UAVs, including the following steps:
[0007] Obtain the video transmission signal data of the UAV;
[0008] Add gradient noise to the video transmission signal data within the target gradient range to obtain a noise image at each gradient;
[0009] Fuse the noise images to obtain an image to be converted;
[0010] Perform time-frequency diagram conversion on the image to be converted to obtain a time-frequency image;
[0011] Detect and identify the UAV based on the time-frequency image to obtain the category of the UAV.
[0012] In a possible implementation of the first aspect, before adding gradient noise to the video transmission signal data within the target gradient range to obtain a noise image at each gradient, the method further includes:
[0013] Adding noise at multiple gradients to the sample video transmission signal to obtain a number of first sample video transmission signals;
[0014] Performing time-frequency map conversion based on the first sample video transmission signals to obtain a first time-frequency image;
[0015] Performing UAV detection and recognition based on the first time-frequency image to obtain a detection result;
[0016] Determining the target gradient range according to the accuracy rate of the detection result.
[0017] In a possible implementation of the first aspect, fusing the noise images to obtain an image to be converted includes:
[0018] Overlapping the noise images to obtain an overlapping image;
[0019] Using the sum of the pixel values at the overlapping positions on the overlapping image as the pixel value of the new position for fusion to obtain the image to be converted.
[0020] In a possible implementation of the first aspect, after overlapping the noise images to obtain an overlapping image, the method further includes:
[0021] Adjusting the overlapping image according to the periodic characteristics of the video transmission signal data to obtain a first overlapping image;
[0022] Using the sum of the pixel values at the overlapping positions on the overlapping image as the pixel value of the new position for fusion to obtain the image to be converted includes:
[0023] Using the sum of the pixel values at the overlapping positions on the first overlapping image as the pixel value of the new position for fusion to obtain the image to be converted.
[0024] In a possible implementation of the first aspect, performing time-frequency map conversion on the image to be converted to obtain a time-frequency image includes:
[0025] Performing time-frequency map conversion on the image to be converted to obtain a converted image;
[0026] Performing normalization processing on the converted image to obtain a normalized image;
[0027] Performing color mapping based on the normalized image to obtain a time-frequency image.
[0028] In a possible implementation of the first aspect, before adding gradient noise to the video transmission signal data within the target gradient range to obtain a noise image at each gradient, the method further includes:
[0029] Perform data clipping on the video transmission signal data to obtain clipped data;
[0030] Perform time-domain windowing on the clipped data to obtain the target video transmission signal data.
[0031] In a possible implementation manner of the first aspect, before detecting and identifying an unmanned aerial vehicle (UAV) based on a time-frequency image and obtaining the category of the UAV, the method further includes:
[0032] Convert the video transmission signal data of different UAVs into time-frequency diagrams to obtain a plurality of time-frequency sample images;
[0033] Train based on the time-frequency sample images to obtain a detection and identification model;
[0034] Detecting and identifying a UAV based on a time-frequency image and obtaining the category of the UAV includes:
[0035] Input the time-frequency image into the detection and identification model to perform UAV detection and identification and obtain the category of the UAV.
[0036] In a second aspect, an embodiment of the present application provides a UAV detection device, including:
[0037] An acquisition module, which is used to acquire the video transmission signal data of the UAV;
[0038] A noise addition module, which is used to add gradient noise to the video transmission signal data within a target gradient range to obtain a noise image at each gradient;
[0039] A fusion module, which is used to fuse the noise images to obtain an image to be converted;
[0040] A time-frequency conversion module, which is used to perform time-frequency diagram conversion on the image to be converted to obtain a time-frequency image;
[0041] A detection and identification module, which is used to detect and identify a UAV based on the time-frequency image and obtain the category of the UAV.
[0042] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, storing a computer program, which, when loaded and executed by a processor, implements the UAV detection method provided in any one of the above first aspects.
[0043] In a fourth aspect, an embodiment of the present application provides an electronic device, including a processor and a memory, wherein,
[0044] The memory is used to store a computer program;
[0045] The processor is used to load and execute a computer program to enable an electronic device to execute the drone detection method provided in any one of the above first aspects.
[0046] Compared with the prior art, the beneficial effects of the present application are as follows:
[0047] The drone detection method, device, storage medium, and electronic device provided by the embodiments of the present application. The method includes: obtaining the video transmission signal data of the drone; adding gradient noise to the video transmission signal data within a target gradient range to obtain a noise image at each gradient; fusing the noise images to obtain an image to be converted; performing a time-frequency diagram conversion on the image to be converted to obtain a time-frequency image; and performing drone detection and recognition based on the time-frequency image to obtain the category of the drone. By converting the video transmission signal of the drone into a time-frequency diagram, the present application realizes drone detection based on image recognition. By adding noise gradiently within a certain range and extracting under multi-scale conditions, and fusing the images with added noise at multiple gradients, it can not only simulate the interference in the real environment to enhance the universality of recognition, but also highlight the signal characteristics of the drone after fusion, thereby improving the quality of the time-frequency image after time-frequency conversion. Finally, the category of the drone can be more accurately and quickly recognized using this time-frequency image to achieve detection, improving the level of drone detection and recognition based on the video transmission signal. Description of the Drawings
[0048] Figure 1 It is a schematic structural diagram of an electronic device for the hardware operating environment involved in the embodiments of the present application;
[0049] Figure 2 It is a schematic flowchart of the drone detection method provided by the embodiments of the present application;
[0050] Figure 3 It is several time-frequency images obtained by converting the video transmission signals of different categories of drones;
[0051] Figure 4 It is a schematic module diagram of the drone detection device provided by the embodiments of the present application;
[0052] Reference numerals in the figure: 101 - processor, 102 - communication bus, 103 - network interface, 104 - user interface, 105 - memory. Detailed Embodiments
[0053] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.
[0054] Refer to the appended Figure 1 appendix Figure 1Schematic diagram of the electronic device structure of the hardware operating environment involved in the solution of the embodiment of the present application. The electronic device may include: a processor 101, such as a central processing unit (CPU), a communication bus 102, a user interface 104, a network interface 103, and a memory 105. Among them, the communication bus 102 is used to realize the connection and communication between these components. The user interface 104 may include a display screen (Display), an input unit such as a keyboard (Keyboard). Optionally, the user interface 104 may further include a standard wired interface and a wireless interface. The network interface 103 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (WIreless-Fidelity, WI-FI) interface). The memory 105 may optionally be a storage device independent of the aforementioned processor 101. The memory 105 may be a high-speed random access memory (Random Access Memory, RAM) memory, or may be a stable non-volatile memory (Non-Volatile Memory, NVM), such as at least one disk memory; the processor 101 may be a general-purpose processor, including a central processor, a network processor, etc., or may also be a digital signal processor, an application-specific integrated circuit, a field programmable gate array, or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.
[0055] Those skilled in the art can understand that the structure shown in the appendix Figure 1 does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine some components, or have different component arrangements.
[0056] As shown in the appendix Figure 1 The memory 105, as a storage medium, may include an operating system, a network communication module, a user interface module, and a drone detection device.
[0057] In the electronic device shown in the appendix Figure 1 the network interface 103 is mainly used for data communication with a network server; the user interface 104 is mainly used for data interaction with a user; the processor 101 and the memory 105 in the present application may be set in the electronic device. The electronic device calls the drone detection device stored in the memory 105 through the processor 101 and executes the drone detection method provided by the embodiment of the present application.
[0058] As the awareness of the potential hazards of drones continues to grow, drone detection technology has naturally become a hot topic of concern, and current research is also developing in different directions. The main concept of drone detection is to use certain characteristics of drones to separate them from other objects. The mainstream directions include vision-based, sound-based, radar-based, radio frequency-based, and WiFi-based methods.
[0059] 1) Vision-based Drone Detection Technology
[0060] Vision-based drone detection systems use cameras to capture images containing drones and apply object recognition algorithms to identify drones from video frames. The success of vision detection largely depends on the spatial resolution of commercial cameras.
[0061] 2) Sound-based Drone Detection Technology
[0062] Sound-based drone detection systems use microphones to obtain acoustic information from the motors equipped on nearby drones and extract the characteristics of audio data in the time domain or frequency domain. By using sound sensors to capture the surrounding environmental sounds, the purpose of monitoring drones can be achieved, and the effective working range of most acoustic sensors is about 10 meters.
[0063] 3) Radar-based Drone Detection Technology
[0064] Radar-based drone detection systems utilize the electromagnetic echo characteristics and spectral correlation functions of Doppler radar and apply deep belief networks (DBNs) to identify drones from other targets. Radar is an active sensor and belongs to an active detection scheme, using the principle of reflected electromagnetism to achieve the detection of drones.
[0065] 4) Temperature-based Drone Detection Technology
[0066] Some fixed-wing drones use turbofans and turbine gases as propulsion systems and are equipped with engines that emit hot air. However, most consumer drones on the market are mainly quadcopters and hexacopters, and the heat they release is not sufficient to be detected by such methods.
[0067] Compared with the above several drone detection methods, radio signal-based drone detection technology is not restricted by visibility and urban environments. Radio frequency-based drone detection systems are designed to detect the radio signals between drones and their control devices. Since drones usually communicate with their control devices through specific wireless channels to transmit information such as the operator's operation instructions and video streams, drones can be detected based on the characteristics of wireless signals. The most commonly used frequency bands for drones are the two industrial, scientific, and medical (ISM) radio frequency bands around 2.4 GHz and 5.8 GHz, which can be used without permission.
[0068] Convert the video transmission signal of the drone into a time-frequency graph for the detection and identification of the drone. However, the real environment is different from the ideal environment, and there are often inevitable noise interferences. Such interferences will affect the quality of the video transmission signal. In this case, the information carried on the converted time-frequency graph is not sufficient for specific identification, resulting in a reduction in the detection and identification level of the drone. Therefore, referring to the attached Figure 2 , based on the hardware device of the foregoing embodiment, an embodiment of the present application provides a drone detection method, including the following steps:
[0069] S10: Obtain the video transmission signal data of the drone.
[0070] In the specific implementation process, the video transmission signal of the drone refers to the signal that the drone wirelessly transmits the aerial photographed image to the ground display device in real time through its video transmission system. A detection range can be set. When it is detected that a drone enters the detection range, the detection and identification are triggered. Since different types of drones have different video transmission signals, obtaining their video transmission signal data can specifically determine the specific type of the drone that enters the detection range.
[0071] S20: Add gradient noise to the video transmission signal data within the target gradient range to obtain a noise image at each gradient.
[0072] In the specific implementation process, after obtaining the video transmission signal, it is converted into a time-frequency graph for discrimination. One step in the conversion process is to add noise, and the addition of fixed noise cannot adapt to all types of drones. The noise interference in the complex real environment is not fixed either. Therefore, a multi-scale addition method is adopted, and noise is added according to the gradient, that is, the noise is gradually increased or decreased, so as to obtain multiple images with added noise.
[0073] In one embodiment, before adding gradient noise to the video transmission signal data within the target gradient range to obtain a noise image at each gradient, the method further includes:
[0074] Add multi-gradient noise to the sample video transmission signal to obtain a number of first sample video transmission signals;
[0075] Perform time-frequency graph conversion based on the first sample video transmission signal to obtain a first time-frequency image;
[0076] Perform drone detection and identification based on the first time-frequency image to obtain a detection result;
[0077] Determine the target gradient range according to the accuracy rate of the detection result.
[0078] In the specific implementation process, since the addition of noise cannot simulate the real environment without limit, adding noise beyond a certain range will cause the time-frequency diagram to be unrecognizable. Therefore, it is necessary to carry out within the target gradient range. Add noise to the sample image transmission signal without setting a range. After adding, perform time-frequency conversion processing on it, and then use the converted time-frequency image for detection and recognition, and determine the range based on the accuracy rate of the detection result. For example, set a threshold of 95%. The accuracy rate of adding noise and recognizing at a certain gradient can reach 95%. That is, within the target gradient range, gradually expand the noise range, increase and decrease the noise range until the accuracy rate is lower than 95%. At this time, the noise is the two end values of the target gradient range, and thus the target gradient range for adding noise is determined.
[0079] In one embodiment, before adding gradient noise to the image transmission signal data within the target gradient range to obtain noise images at each gradient, the method further includes:
[0080] Perform data clipping on the image transmission signal data to obtain clipped data;
[0081] Perform time-domain windowing processing on the clipped data to obtain target image transmission signal data.
[0082] In the specific implementation process, the original signal is continuously generated. Therefore, it is only necessary to perform data clipping on it to intercept a part for identification and detection. Time-domain windowing processing involves applying a window function with a finite length to the time-domain signal to reduce spectral leakage or Gibbs effect generated when the signal is truncated. Its basic principle is that when truncating (framing) a continuous signal, instead of simply using a rectangular window for truncation, a window function with a smoother shape (such as a Hanning window, a Hamming window, etc.) is used to gradually reduce the amplitude of the signal at the truncation edge, thereby reducing the distortion of the spectrum.
[0083] S30: Fuse the noise images to obtain an image to be converted.
[0084] In the specific implementation process, the same image transmission signal data forms multiple noise images after adding gradient noise. By fusing multiple noise images into one image to represent the presence of the UAV and being able to improve the quality of the carried feature information, thereby improving the quality of the subsequent time-frequency image. Specifically, fusing the noise images to obtain an image to be converted includes:
[0085] Overlay the noise images to obtain an overlaid image;
[0086] Use the sum of the pixel values of the overlapping points on the overlaid image as the pixel value of the new point for fusion to obtain the image to be converted.
[0087] The fusion of images can be carried out in a superposition manner. Since the final detected and recognized time-frequency map confirms the corresponding UAV category based on the distribution characteristics of the bar code-like features on it, directly adding the image pixels will not affect the recognition. Instead, it will make the feature part more prominent. Further, after overlapping the noise images to obtain the overlapping image, the method further includes:
[0088] According to the periodic characteristics of the video transmission signal data, the overlapping image is adjusted to obtain the first overlapping image.
[0089] In the specific implementation process, since the UAV video transmission signal has periodic characteristics, in order to make the quality of the fused image better and make the alignment of the effective parts more accurate during the fusion process, the periodic characteristics are used to adjust the overlapping image. Taking its period as the adjustment node, after aligning the time axis of the data, the features on the noise image within a single cycle time are further aligned to obtain an overlapping image with better quality.
[0090] Based on the foregoing steps, the sum of the pixel values of the overlapping points on the overlapping image is used as the pixel value of the new point for fusion to obtain the image to be converted, including:
[0091] The sum of the pixel values of the overlapping points on the first overlapping image is used as the pixel value of the new point for fusion to obtain the image to be converted.
[0092] S40: Perform time-frequency map conversion on the image to be converted to obtain a time-frequency image.
[0093] In the specific implementation process, time-frequency transformation can correlate the time-domain and frequency-domain characteristics of the signal. The signal can be represented by a two-dimensional function containing time and frequency, intuitively presenting the relationship between the time domain and the frequency domain. Therefore, time-frequency analysis is regarded as a very valuable signal analysis and processing tool in the recognition and classification of communication signals. Usually, neural network models are used for detection and recognition. Therefore, before detection and recognition, it will be converted into a time-frequency map as the input of the neural network. Specifically, performing time-frequency map conversion on the image to be converted to obtain a time-frequency image includes:
[0094] Perform time-frequency map conversion on the image to be converted to obtain a converted image;
[0095] Perform normalization processing on the converted image to obtain a normalized image;
[0096] Perform color mapping based on the normalized image to obtain a time-frequency image.
[0097] In the specific implementation process, time-frequency map conversion is an important technology in the field of signal processing. It can convert a one-dimensional time-domain signal into a two-dimensional time-frequency domain representation, thereby revealing the characteristics of the signal in the two dimensions of time and frequency, such as Figure 3Shown are several time-frequency images obtained by converting the video transmission signals of different categories of drones. This conversion is particularly important for analyzing non-stationary signals because the frequency components of these signals change over time. The Short-Time Fourier Transform (STFT) is a commonly used signal processing method in time-frequency analysis, which combines time-domain and frequency-domain information to show the characteristics of a signal in both the time and frequency dimensions. STFT is particularly effective for analyzing non-stationary signals (such as time-varying frequency signals). The short-time Fourier transform is widely used in time-frequency analysis. This method utilizes the local stationary characteristics of the signal, slides a window function of a certain length to intercept the signal, and then performs a Fourier transform on the intercepted signal.
[0098] The purpose of normalizing the time-frequency map is to limit the pixel values or amplitudes of the time-frequency map within a specific range, so that the time-frequency maps of different signals are numerically comparable while maintaining the relative relationships between the signals. Color mapping is a method of data visualization by associating the numerical range of the data with a predefined color table. In the time-frequency map transformation, color mapping maps the time-frequency characteristics of the signal (such as amplitude, phase, or energy) into the color space, and different colors represent different numerical values or attributes.
[0099] S50: Conduct drone detection and identification based on the time-frequency image to obtain the category of the drone.
[0100] In the specific implementation process, as shown in the Figure 3 time-frequency image, different categories of drones have their corresponding time-frequency map characteristics and the distribution of time-frequency map characteristics. By detecting and identifying these characteristics and their distributions, the category of the drone can be corresponding. To achieve fast detection and identification, a neural network model can be utilized, that is: before conducting drone detection and identification based on the time-frequency image to obtain the category of the drone, the method further includes:
[0101] Perform time-frequency map conversion on the video transmission signal data of different drones to obtain several time-frequency sample images;
[0102] Train based on the time-frequency sample images to obtain a detection and identification model;
[0103] Conduct drone detection and identification based on the time-frequency image to obtain the category of the drone, including:
[0104] Input the time-frequency image into the detection and identification model to conduct drone detection and identification and obtain the category of the drone.
[0105] A large number of time-frequency images are collected through testing, and the training set, validation set, and test set are divided proportionally. Let the time-frequency sample images in the training set, that is, the time-frequency images converted from the original video transmission signal data of different drones, be used as the input of the detection and recognition model, and the drone category be used as the output for supervised training, so that the model can learn the characteristics of the time-frequency images and be able to output the corresponding drone category. The neural network architecture of the model can adopt EfficientNet, which is a convolutional neural network (CNN) architecture for image classification tasks. The design concept is to find a balance point through a compound scaling method, weigh between different dimensions to achieve higher performance and efficiency. Use the EfficientNet model as a pre-trained model and train it as a recognition network for drone categories to obtain a detection and recognition model with better quality.
[0106] In this embodiment, the video transmission signal data of the drone is obtained; within the target gradient range, gradient noise is added to the video transmission signal data to obtain a noise image under each gradient; the noise images are fused to obtain an image to be converted; the image to be converted is subjected to time-frequency diagram conversion to obtain a time-frequency image; based on the time-frequency image, drone detection and recognition are performed to obtain the category of the drone. In this application, by converting the video transmission signal of the drone into a time-frequency diagram, drone detection based on image recognition is realized. By adding noise gradiently within a certain range and extracting under multi-scale conditions, and fusing the images with added noise under multiple gradients, not only can the interference of the real environment be simulated to enhance the universality of recognition, but also the signal characteristics of the drone can be highlighted after fusion, thereby improving the quality of the time-frequency image after time-frequency conversion. Finally, the category of the drone can be more accurately and quickly recognized using this time-frequency image to achieve detection, improving the level of drone detection and recognition based on video transmission signals.
[0107] Refer to the appendix Figure 4 , based on the same inventive concept as in the foregoing embodiment, the embodiment of the present application further provides a drone detection device, including:
[0108] An acquisition module, which is used to acquire the video transmission signal data of the drone;
[0109] A noise addition module, which is used to add gradient noise to the video transmission signal data within the target gradient range to obtain a noise image under each gradient;
[0110] A fusion module, which is used to fuse the noise images to obtain an image to be converted;
[0111] A time-frequency conversion module, which is used to perform time-frequency diagram conversion on the image to be converted to obtain a time-frequency image;
[0112] A detection and recognition module, which is used to detect and recognize drones based on time-frequency images and obtain the categories of drones.
[0113] Those skilled in the art should understand that the division of each module in the embodiment is only a logical function division. In actual application, it can be fully or partially integrated into one or more actual carriers, and these modules can all be implemented in the form of software called by a processing unit, or all be implemented in the form of hardware, or be implemented in the form of a combination of software and hardware. It should be noted that each module in the drone detection device in this embodiment corresponds one by one to each step in the drone detection method in the foregoing embodiment. Therefore, the specific implementation manner of this embodiment can refer to the implementation manner of the foregoing drone detection method, and will not be elaborated here.
[0114] Based on the same inventive concept as in the foregoing embodiment, an embodiment of the present application also provides a computer-readable storage medium storing a computer program, which, when loaded and executed by a processor, implements the drone detection method provided by the embodiment of the present application.
[0115] Based on the same inventive concept as in the foregoing embodiment, an embodiment of the present application also provides an electronic device, including a processor and a memory, wherein,
[0116] The memory is used to store a computer program;
[0117] The processor is used to load and execute the computer program so that the electronic device executes the drone detection method provided by the embodiment of the present application.
[0118] In some embodiments, the computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, flash memory, magnetic surface memory, optical disc, or CD-ROM; or it may be various devices including one or any combination of the above memories. The computer may be various computing devices including smart terminals and servers.
[0119] In some embodiments, the executable instructions may be in the form of a program, software, software module, script, or code, and may be written in any form of programming language (including compiled or interpreted languages, or declarative or procedural languages), and may be deployed in any form, including being deployed as an independent program or being deployed as a module, component, subroutine, or other unit suitable for use in a computing environment.
[0120] As an example, executable instructions may or may not correspond to files in a file system, and may be stored as part of a file that holds other programs or data. For example, they may be stored in one or more scripts in a Hyper Text Markup Language (HTML) document, in a single file dedicated to the program being discussed, or in multiple cooperating files (such as files that store one or more modules, subroutines, or code portions).
[0121] As an example, executable instructions may be deployed to execute on one computing device, or on multiple computing devices located at one site, or on multiple computing devices distributed across multiple sites and interconnected by a communication network.
[0122] It should be noted that in this document, the terms "including", "comprising", or any other variant thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or system that includes a series of elements includes not only those elements but also other elements not expressly listed, or elements that are inherent to such process, method, article, or system. Without further limitation, an element qualified by the phrase "including a..." does not exclude the presence of additional identical elements in the process, method, article, or system that includes such element.
[0123] The serial numbers of the embodiments of the present application above are merely for description and do not represent the superiority or inferiority of the embodiments.
[0124] Through the description of the above embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present application, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as a read-only memory / random access memory, magnetic disk, optical disk) and includes several instructions to enable a multimedia terminal device (which may be a mobile phone, computer, television receiver, or network device, etc.) to execute the methods described in the various embodiments of the present application.
[0125] In summary, the drone detection method, device, storage medium, and electronic device provided by this application include: obtaining the video transmission signal data of the drone; adding gradient noise to the video transmission signal data within a target gradient range to obtain a noise image at each gradient; fusing the noise images to obtain an image to be converted; performing a time-frequency diagram conversion on the image to be converted to obtain a time-frequency image; and performing drone detection and recognition based on the time-frequency image to obtain the category of the drone. By converting the video transmission signal of the drone into a time-frequency diagram, this application realizes drone detection based on image recognition. By adding noise gradiently within a certain range and extracting under multi-scale conditions, and fusing the images with added noise at multiple gradients, it can not only simulate the interference in the real environment to enhance the universality of recognition, but also highlight the signal characteristics of the drone after fusion, thereby improving the quality of the time-frequency image after time-frequency conversion. Finally, the category of the drone can be more accurately and quickly recognized using this time-frequency image to achieve detection, improving the level of drone detection and recognition based on video transmission signals.
[0126] The foregoing are only the preferred embodiments of this application and are not intended to limit this application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of this application shall be included within the protection scope of this application.
Claims
1. A method for detecting an unmanned aerial vehicle, characterized in that, Including the following steps: Obtain the video transmission signal data of the drone; Within the target gradient range, add gradient noise to the video transmission signal data to obtain a noise image at each gradient; Fuse the noise images to obtain an image to be converted; Perform time-frequency diagram conversion on the image to be converted to obtain a time-frequency image; Based on the time-frequency image, perform drone detection and recognition to obtain the category of the drone.
2. The drone detection method according to claim 1, characterized in that, Before adding gradient noise to the video transmission signal data within the target gradient range to obtain a noise image at each gradient, the method further includes: Add noise with multiple gradients to the sample video transmission signal to obtain a number of first sample video transmission signals; Based on the first sample video transmission signals, perform time-frequency diagram conversion to obtain a first time-frequency image; Based on the first time-frequency image, perform drone detection and recognition to obtain a detection result; Determine the target gradient range according to the accuracy rate of the detection result.
3. The drone detection method according to claim 1, wherein The fusing the noise images to obtain an image to be converted includes: Overlap the noise images to obtain an overlapping image; Using the sum of the pixel values of the overlapping points on the overlapping image as the pixel value of the new point for fusion to obtain an image to be converted.
4. The drone detection method according to claim 3, wherein, After overlapping the noise images to obtain an overlapping image, the method further includes: According to the periodic characteristics of the video transmission signal data, adjust the overlapping image to obtain a first overlapping image; The using the sum of the pixel values of the overlapping points on the overlapping image as the pixel value of the new point for fusion to obtain an image to be converted includes: Using the sum of the pixel values of the overlapping points on the first overlapping image as the pixel value of the new point for fusion to obtain an image to be converted.
5. The drone detection method according to claim 1, characterized in that, The performing time-frequency diagram conversion on the image to be converted to obtain a time-frequency image includes: Perform time-frequency diagram conversion on the image to be converted to obtain a converted image; Perform normalization processing on the converted image to obtain a normalized image; Based on the normalized image, perform color mapping to obtain a time-frequency image.
6. The drone detection method according to claim 1, wherein, Before adding gradient noise to the video transmission signal data within the target gradient range to obtain a noise image at each gradient, the method further includes: Perform data cropping on the video transmission signal data to obtain cropped data; Perform time-domain windowing processing on the cropped data to obtain target video transmission signal data.
7. The drone detection method according to claim 1, characterized in that, Before performing drone detection and recognition based on the time-frequency image to obtain the category of the drone, the method further includes: Perform time-frequency diagram conversion on the video transmission signal data of different drones to obtain a number of time-frequency sample images; Based on the time-frequency sample images, perform training to obtain a detection and recognition model; The performing drone detection and recognition based on the time-frequency image to obtain the category of the drone includes: Input the time-frequency image into the detection and recognition model to perform drone detection and recognition to obtain the category of the drone.
8. An unmanned aerial vehicle detection device, characterized in that, Including: An acquisition module, which is used to acquire the video transmission signal data of the drone; A noise addition module, which is used to add gradient noise to the video transmission signal data within the target gradient range to obtain a noise image at each gradient; A fusion module, which is used to fuse the noise images to obtain an image to be converted; A time-frequency conversion module, which is used to perform time-frequency map conversion on the image to be converted to obtain a time-frequency image; A detection and recognition module, which is used to perform UAV detection and recognition based on the time-frequency image to obtain the category of the UAV.
9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is loaded and executed by a processor, it implements the UAV detection method according to any one of claims 1-7.
10. An electronic device, characterized in that, It includes a processor and a memory, wherein, The memory is used to store a computer program; The processor is used to load and execute the computer program so that the electronic device executes the UAV detection method according to any one of claims 1-7.