Target detection method and device
By using the object detection model trained by recurrent neural network, the rotor modulation characteristics in the timing data of the electromagnetic scattering characteristics of the drone are detected, and the problem of difficult rotor information being detected in drone recognition is solved, and fast and accurate drone model identification and classification is achieved.
Patent Information
- Application Number
- CN202510058143.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-14
- Publication Date
- 2025-04-25
AI Technical Summary
During the drone identification process, the electromagnetic scattering characteristics of the rotor are weak and are easily overwhelmed by the strong scattering echo of the fuselage, making it difficult to judge whether there is rotor modulation information, which in turn affects the identification and classification of the drone model.
By obtaining the electromagnetic scattering characteristics timing data of the drone in the motion state, using a recurrent neural network, especially a long and short-term memory network, the object detection model is trained to detect whether the electromagnetic scattering characteristics timing data has modulation characteristics of the rotating components.
It realizes a rapid and accurate judgment of whether there are rotating components of the drone, provides an effective basis for identifying and classification of drone models, and solves the problem of classification and identification of aerial targets.
Smart Images
Figure CN119881819A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of electromagnetic scattering characteristic analysis, and in particular to a target detection method and device. Background Art
[0002] Radar target (such as drone) identification is of great significance for low-altitude defense. The electromagnetic scattering characteristic data of the target, as one of the important measurement values of radar detection, can provide important information such as the target's structural parameters and motion parameters.
[0003] In the process of drone identification, compared with the fuselage, the rotor is small in size, rotates fast, and has weak electromagnetic scattering characteristics, which makes it easy for the rotor modulation information received by the radar to be submerged in the strong scattering echo of the fuselage, making it difficult to determine whether there is rotor modulation information. This makes it difficult to further identify the drone model. Summary of the invention
[0004] The present invention provides a target detection method and device, which can quickly and accurately determine whether there are rotating parts in a drone, thereby providing an effective basis for model identification and classification of the drone.
[0005] In a first aspect, the present invention provides a target detection method, the method comprising: obtaining electromagnetic scattering characteristic time series data of a drone to be detected in a moving state; using a target detection model, processing the electromagnetic scattering characteristic time series data to obtain a detection result of whether the electromagnetic scattering characteristic time series data has modulation characteristics of a rotating component, wherein the target detection model is obtained by training a recurrent neural network.
[0006] Optionally, the recurrent neural network is a long short-term memory network.
[0007] Optionally, the method also includes: obtaining time series data samples of the electromagnetic scattering characteristics of the drone in motion; constructing a training set and a test set based on the electromagnetic scattering characteristics time series data samples; using the training set to perform supervised training on the long short-term memory network to obtain a trained long short-term memory network; when it is determined according to the test set that the performance of the trained long short-term memory network meets set conditions, using the trained long short-term memory network as the target detection model.
[0008] Optionally, obtaining the electromagnetic scattering characteristic time series data samples of the UAV in motion includes: constructing a simulation model of the UAV; and calculating the electromagnetic scattering characteristic time series sample data of the simulation model of the UAV in motion.
[0009] Optionally, the calculation of the electromagnetic scattering characteristic time series sample data of the simulation model of the UAV in a moving state includes: calculating the radar scattering cross-section time series data of the simulation model of the UAV at different radar frequencies, different radar line of sight observation angles, different motion trajectories, and different rotor speeds, and using the radar scattering cross-section data as the electromagnetic scattering characteristic time series sample data.
[0010] Optionally, the long short-term memory network includes an input layer, a first hidden layer, a second hidden layer and an output layer.
[0011] In a second aspect, the present invention provides a target detection device, characterized in that the device includes: an acquisition module for acquiring electromagnetic scattering characteristic time series data of a drone to be detected in a moving state; a detection module for processing the electromagnetic scattering characteristic time series data using a target detection model to obtain a detection result of whether the electromagnetic scattering characteristic time series data contains modulation characteristics of a rotating component, wherein the target detection model is obtained by training a recurrent neural network.
[0012] In a third aspect, a computer device is provided, which includes a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to implement the steps of the target detection method as described above.
[0013] In a fourth aspect, a computer-readable storage medium is provided, characterized in that a computer program is stored in the storage medium, and when the computer program is executed by a processor, the steps of the target detection method as described above are implemented.
[0014] In a fifth aspect, a computer program product is provided, characterized in that it includes a computer program, and when the computer program is executed by a processor, the steps of the target detection method as described above are implemented.
[0015] The technical solution provided by the present invention can at least bring the following beneficial effects: by utilizing an end-to-end neural network model, it is possible to quickly and accurately detect whether there is information on the rotor modulation spectrum that is different from the overall target motion in the electromagnetic scattering characteristic time series data of the UAV, thereby facilitating the subsequent rapid and accurate identification of whether the UAV has rotating parts, thereby solving the problem of aerial target classification and identification. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0017] Figure 1 is a flow chart of a target detection method provided by an embodiment of the present invention;
[0018] Figure 2 is a flow chart of a target detection method provided by another embodiment of the present invention;
[0019] Figure 3 It is a structural schematic diagram of a simulation model of a drone provided by an embodiment of the present invention;
[0020] Figure 4 It is a schematic diagram of the structure of an LSTM network provided by an embodiment of the present invention;
[0021] Figure 5 It is a schematic diagram of the structure of an LSTM network provided by an embodiment of the present invention;
[0022] Figure 6 is a schematic diagram of the structure of a target detection device provided by an embodiment of the present invention;
[0023] Figure 7 It is a hardware architecture diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.
[0025] In order to solve the problems existing in the related art, the present invention provides a target detection method and device, which can quickly and accurately determine whether there are rotating parts in a drone, thereby providing an effective basis for the model identification and classification of the drone.
[0026] The following is a detailed description with reference to specific embodiments.
[0027] Please refer to Figure 1 , some embodiments of the present invention provide a target detection method including step S11 to step S12.
[0028] Step S11, obtaining time series data of electromagnetic scattering characteristics of the unmanned aerial vehicle to be detected in a moving state.
[0029] In some examples, the electromagnetic scattering characteristic time series data includes radar cross section (RCS) time series data. In specific implementations, radar can be used to collect RCS time series data of the unmanned aerial vehicle to be detected in a moving state. For example, radar is used to collect RCS time series data of the unmanned aerial vehicle to be detected in a moving state at different radar sight angles.
[0030] Step S12: Processing the electromagnetic scattering characteristic time series data using the target detection model to obtain a detection result of whether the electromagnetic scattering characteristic data has a modulation feature of a rotating component.
[0031] Among them, the target detection model is obtained by training a recurrent neural network. In some examples, the recurrent neural network is a long short-term memory network. The Long Short-Term Memory (LSTM) network is a special type of recurrent neural network (RNN). Unlike the traditional RNN model, the LSTM network uses the information of the previous moment and the input information of this moment to train together during training, so it is very suitable for processing and predicting time series data.
[0032] The detection result of step S12 is that the electromagnetic scattering characteristic data of the detected UAV has the modulation feature of the rotating component or the electromagnetic scattering characteristic data of the detected UAV does not have the modulation feature of the rotating component.
[0033] In an embodiment of the present invention, by utilizing an end-to-end target detection model to capture long-term dependencies in the electromagnetic scattering characteristic time series, it is possible to determine whether a weak rotor modulation spectrum appears during the movement of the UAV, thereby facilitating the subsequent rapid and accurate identification of whether the UAV has rotating parts.
[0034] Please refer to Figure 2 , some embodiments of the present invention provide a target detection method including steps S21 to S26.
[0035] Step S21, obtaining time series data samples of electromagnetic scattering characteristics of the UAV in motion.
[0036] In some examples, electromagnetic scattering characteristic time series data samples are obtained in the following manner: a simulation model of the drone is constructed; and time series sample data of electromagnetic scattering characteristics of the simulation model of the drone in motion is calculated.
[0037] For example, first construct Figure 3The simulation model of the quad-rotor UAV is shown, and then the simulation model of the quad-rotor UAV is imported into the electromagnetic calculation software to calculate the electromagnetic scattering characteristic timing data when the rotor is rotating during flight and the electromagnetic scattering characteristic timing data when the rotor is not rotating.
[0038] Among them, the electromagnetic scattering characteristic time series data when the rotor is rotating and the electromagnetic scattering characteristic time series data when the rotor is not rotating can further include: radar scattering cross-section time series data of the UAV simulation model under different radar frequencies, different radar line of sight observation angles, different motion trajectories, and different rotor speeds.
[0039] Step S22, constructing a training set and a test set according to the electromagnetic scattering characteristic time series data samples.
[0040] In this step, the electromagnetic scattering characteristic time series data samples can be divided into a training set and a test set according to a preset ratio (for example, 7:3). Afterwards, the training set is labeled. In specific implementation, the training set can be labeled based on a manual labeling method or an automatic labeling method, or a combination of the two labeling methods. The labeled label is used to indicate whether the data sample is a modulation feature of a rotating component.
[0041] Step S23, using the training set, supervised training is performed on the long short-term memory network to obtain a trained long short-term memory network.
[0042] In some examples, the long short-term memory network includes an input layer, a first hidden layer, a second hidden layer, and an output layer. For example, the long short-term memory network uses Figure 4 and 5 The network structure shown in Figure 5 In the figure, A represents the nodes of the first hidden layer, A' represents the nodes of the second hidden layer, and A" represents the nodes of the output layer.
[0043] In the disclosed embodiment, by superimposing a multi-layer network structure on the LSTM model, deeper feature information can be extracted, which facilitates better extraction of patterns in time series data and better mining of the modulation spectrum features of weak rotating components in the RCS time series data samples of the drone, thereby helping to improve the performance of the trained model while accelerating the convergence of the model during the training process.
[0044] Step S24, when it is determined according to the test set that the performance of the trained long short-term memory network meets the set conditions, the trained long short-term memory network is used as the target detection model.
[0045] In some examples, the performance of the trained LSTM network is represented by one or more performance indicators, and the setting condition is that the performance indicator meets a preset value range. For example, the performance indicator includes accuracy, and the setting condition is that the accuracy is greater than or equal to an accuracy threshold.
[0046] For example, in step S24, the test set is input into the trained LSTM model to obtain an output result; the accuracy of the trained LSTM model is determined according to the output result. When it is determined that the accuracy of the trained LSTM model is greater than or equal to the accuracy threshold, the trained LSTM model is used as the target detection model; when the accuracy of the trained LSTM model is less than the accuracy threshold, step S23 is performed again. Step S25, the electromagnetic scattering characteristic time series data of the unmanned aerial vehicle to be detected in a moving state is obtained.
[0047] For example, in step S25, radar is used to collect RCS time series data of the unmanned aerial vehicle to be detected in motion at different radar sight angles.
[0048] Step S26, using the target detection model, processing the electromagnetic scattering characteristic time series data to obtain a detection result of whether the drone to be detected has a rotating component.
[0049] In the disclosed embodiment, the above process can be used to construct a target detection model for identifying the modulation characteristics of rotating parts in the RCS timing data of the drone, and based on the model, it is possible to quickly and accurately identify whether there is modulation spectrum information of rotating parts in the RCS timing data of the drone to be detected, which helps to subsequently determine whether the drone has rotating parts, thereby solving the problem of drone classification and identification.
[0050] Please refer to Figure 6 Some embodiments of the present invention provide a target detection device for executing the target detection method as described above. The device specifically includes an acquisition module 61 and a detection module 62.
[0051] The acquisition module 61 is used to acquire the electromagnetic scattering characteristic time series data of the unmanned aerial vehicle to be detected in a moving state.
[0052] In some examples, the recurrent neural network is a long short-term memory network, wherein the long short-term memory network may include an input layer, a first hidden layer, a second hidden layer, and an output layer.
[0053] The detection module 62 is used to process the electromagnetic scattering characteristic time series data using the target detection model to obtain a detection result of whether the electromagnetic scattering characteristic time series data of the unmanned aerial vehicle to be detected has the modulation characteristics of the rotating parts. The target detection model is obtained by training the recurrent neural network.
[0054] In some examples, the target detection device further includes a training module. The training module is used to obtain electromagnetic scattering characteristic time series data samples of the UAV in motion; construct a training set and a test set based on the electromagnetic scattering characteristic time series data samples; use the training set to perform supervised training on the long short-term memory network to obtain a trained long short-term memory network; when it is determined based on the test set that the performance of the trained long short-term memory network meets the set conditions, the trained long short-term memory network is used as a target detection model.
[0055] In the present invention, the above device can detect whether there is rotor modulation spectrum information different from the overall target movement in the electromagnetic scattering characteristic time series data of the drone by utilizing an end-to-end neural network model, which helps to quickly and accurately identify whether the drone has rotating parts, thereby solving the problem of aerial target classification and identification.
[0056] The embodiment of the present application also provides a computer device, please refer to Figure 7 The computer device includes a processor and a memory, in which at least one instruction, at least one program, a code set or an instruction set is stored, and the at least one instruction, at least one program, a code set or an instruction set is loaded and executed by the processor to implement the target detection method provided by the above-mentioned method embodiments.
[0057] An embodiment of the present application also provides a computer-readable storage medium, on which is stored at least one instruction, at least one program, code set or instruction set, and the at least one instruction, at least one program, code set or instruction set is loaded and executed by a processor to implement the target detection method provided by the above-mentioned method embodiments.
[0058] An embodiment of the present application also provides a computer program product, which includes a computer program. A processor of a computer device reads the computer program from a computer-readable storage medium, and the processor executes the computer program, so that the computer device executes any target detection method described in the above embodiments.
[0059] For the convenience of description, the above system or device is described by dividing it into various modules or units according to its functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.
[0060] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present application can be essentially or partly contributed to the prior art in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present application or certain parts of the embodiments.
[0061] Finally, it should be noted that, in this article, relational terms such as first, second, third and fourth are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.
[0062] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.
Claims
1. A target detection method, characterized in that: The method comprises: Obtain the time series data of the electromagnetic scattering characteristics of the drone to be detected in motion; The electromagnetic scattering characteristic time series data is processed using a target detection model to obtain a detection result of whether the electromagnetic scattering characteristic time series data has a modulation feature of a rotating component, wherein the target detection model is obtained by training a recurrent neural network.
2. The target detection method according to claim 1, characterized in that: The recurrent neural network is a long short-term memory network.
3. The target detection method according to claim 2, characterized in that: The method further comprises: Obtain time series data samples of electromagnetic scattering characteristics of the UAV in motion; Constructing a training set and a test set according to the electromagnetic scattering characteristic time series data samples; Using the training set, performing supervised training on the long short-term memory network to obtain a trained long short-term memory network; When it is determined according to the test set that the performance of the trained long short-term memory network meets the set conditions, the trained long short-term memory network is used as the target detection model.
4. The target detection method according to claim 2, characterized in that: The method of obtaining the electromagnetic scattering characteristic time series data samples of the UAV in motion includes: Build a simulation model of the drone; Calculate the time series sample data of the electromagnetic scattering characteristics of the simulation model of the UAV in motion.
5. The target detection method according to claim 4, characterized in that: The calculation of the electromagnetic scattering characteristics time series sample data of the simulation model of the drone in motion includes: The radar cross-section time series data of the simulation model of the UAV under different radar frequencies, different radar line-of-sight observation angles, different motion trajectories, and different rotor speeds are calculated, and the radar cross-section data are used as the electromagnetic scattering characteristic time series sample data.
6. The target detection method according to any one of claims 2 to 5, characterized in that: The long short-term memory network includes an input layer, a first hidden layer, a second hidden layer and an output layer.
7. A target detection device, characterized in that: The device comprises: An acquisition module is used to acquire the electromagnetic scattering characteristic time series data of the UAV to be detected in a moving state; The detection module is used to process the electromagnetic scattering characteristic time series data using a target detection model to obtain a detection result of whether the electromagnetic scattering characteristic time series data has a modulation feature of a rotating component, wherein the target detection model is obtained by training a recurrent neural network.
8. A computer device, characterized in that: The computer device includes a memory and a processor, the memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory to implement the steps of the target detection method as described in any one of claims 1-6.
9. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the target detection method according to any one of claims 1 to 6 are implemented.
10. A computer program product, characterized in that It comprises a computer program, which, when executed by a processor, implements the steps of the target detection method according to any one of claims 1 to 6.