Multi-source information fusion low, slow and small target detection method and unmanned air defense system

Abstract

The application discloses a multi-source information fusion low, slow and small target detection method and an unmanned air defense system, and is used for identifying and tracking low-altitude air balloons, black flying unmanned planes, bird flocks and other foreign objects, and comprises the following steps: based on the power indexes of photoelectric, radar and radio detection equipment, an integrated target perception matrix is established, and photoelectric images and data information of low, slow and small targets are acquired; a multi-source information fusion and target classification model is established, feature level information is respectively processed by constructing a base classifier for photoelectric, radar and radio detection, and is comprehensively processed at an identity level, so that multi-source verification identification results are obtained; ADS-B data is used for discriminating cooperative plane targets, and a radio detection white list spectrum library is used for discriminating cooperative unmanned plane targets and black flying unmanned plane targets; for the black flying unmanned plane targets, an interference strategy is determined according to the region of the belonging perception matrix; a fusion rule is designed, target position information is fused, and a situation display is performed on airspace target position and identity information.

Description

Technical Field

[0001] This invention relates to a target detection method and an air defense system, particularly a multi-source information fusion method for detecting low, slow, and small targets and an unmanned air defense system. Background Technology

[0002] In recent years, cases of illegal use of drones have surged, causing flight delays at civil aviation airports, threatening aircraft take-off and landing safety, and disrupting airspace management order.

[0003] Using a single detection method for drone defense suffers from limitations such as a single intelligence source and insufficient identification capabilities, leading to difficulties in target identification and impacting decision-making. Currently available radar, electro-optical, and radio detection products all have certain shortcomings. Radar struggles to detect low-speed, hovering, and small targets; radio detection can detect low-speed, hovering, and small targets, but cannot detect targets not registered in the spectrum library; electro-optical can detect various targets, but its search range is small, detection distance is short, and search guidance is required.

[0004] How to effectively coordinate and integrate various detection methods to achieve automated equipment scheduling, target identification, and precise positioning remains an unresolved issue in the industry. Currently, some products employing multiple detection methods are simply a matter of overlaying multiple device functions without unified integration of identification and positioning information. This results in insufficient identification capabilities and no improvement in target detection accuracy compared to single-sensor systems. Furthermore, the scheduling of equipment (especially photoelectric detection equipment) lacks automated design, often requiring manual intervention, leading to complex operating procedures and low product usability. Summary of the Invention

[0005] Purpose of the invention: The technical problem to be solved by the present invention is to provide a multi-source information fusion method for detecting low, slow and small targets and an unmanned air defense system, which addresses the shortcomings of the existing technology.

[0006] To address the aforementioned technical problems, this invention discloses a multi-source information fusion method for detecting low-speed, small targets and an unmanned air defense system, comprising the following steps:

[0007] Step 1: Based on the power indicators of photoelectric, radar and radio detection equipment, establish an integrated target perception matrix to acquire photoelectric images and data information of the low, slow and small targets;

[0008] Step 2: Establish a multi-source information fusion and target classification model. Construct base classifiers for photoelectric, radar and radio detection to process feature-level information, and integrate them at the identity level to obtain multi-source verification and identification results, i.e., comprehensive identification results.

[0009] Step 3: Use ADS-B data to identify cooperative aircraft targets, and use a radio detection whitelist spectrum library to identify cooperative UAV targets and unauthorized UAV targets. For unauthorized UAV targets, determine the interference strategy based on the area of ​​the perception matrix to which they belong.

[0010] Step 4: Design fusion rules, fuse target location information, and summarize the airspace target location and identity information for situational display.

[0011] Beneficial effects:

[0012] 1. Based on the characteristics of a small number of empty targets and low ambiguity, the design adopts a perception matrix for target segmentation and association matching, which greatly saves computing resources.

[0013] 2. The entire multi-source information fusion detection, identification, and countermeasure process has an automated processing flow, which can automatically schedule the operation of photoelectric and countermeasure equipment without human intervention, and can achieve 24 / 7 unattended operation.

[0014] 3. Employ a combination of radar, photoelectric, and radio detection methods for comprehensive identification, and combine this with the ADS-B whitelist database for thorough target identification, so that the target's identity can be clearly identified.

[0015] 4. Based on the characteristics of the "1+1+N" detection system, fusion rules are designed to obtain target fusion tracks, which can achieve higher target detection accuracy. Attached Figure Description

[0016] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments, and the advantages of the present invention in the above and / or other aspects will become clearer.

[0017] Figure 1 This is a flowchart illustrating the method of the present invention.

[0018] Figure 2 This is a schematic diagram of the composition of the unmanned frame system in this invention.

[0019] Figure 3 This is a schematic diagram of the situation display of the UAV in an embodiment of the present invention.

[0020] Figure 4 This is a block diagram of the multi-source information comprehensive identification method in this invention.

[0021] Figure 5 This is a schematic diagram of the sensing and scheduling matrix.

[0022] Figure 6 This is a schematic diagram of the optoelectronic coordination process.

[0023] Figure 7 This is a schematic diagram of the identity recognition process using an SVM cascaded multidimensional hyperplane classifier.

[0024] Figure 8 This is a schematic diagram of the comprehensive identification process. Detailed Implementation

[0025] A multi-source information fusion method for detecting low-speed, small targets and an unmanned air defense system are proposed. This method flexibly deploys various types and quantities of UAV detection radars, optoelectronic detection devices, radio detection devices, and omnidirectional and directional countermeasure devices at single or multiple points to form an airspace detection and countermeasure network. Each detection and countermeasure device is connected to a control system through a network for centralized management and information fusion. Combined with cooperative information sources such as ADS-B (Automatic Dependent Surveillance-Broadcast), the system achieves comprehensive monitoring and threat warning of flights, legal UAVs, and unauthorized UAVs, while also providing UAV control and disposal methods.

[0026] The term "low, slow, and small targets" refers to targets that take off, such as drones, birds, and other small, slow-moving objects.

[0027] The photoelectric image target detection method described in this invention mainly includes establishing a "1+1+N" integrated target perception matrix (i.e., one radar, one radio detection device, and N photoelectric devices) based on the power indicators of photoelectric, radar, and radio detection equipment. This matrix acquires omnidirectional photoelectric images and point / track information of low-altitude, slow-moving, and small targets. A multi-source information fusion and target classification model is established, outputting comprehensive target identification information. ADS-B data and a radio detection whitelist spectrum library are used to distinguish cooperative targets. For unauthorized targets, interference and countermeasure plans are determined based on the identification results and the perception area. Furthermore, fusion rules are designed based on the characteristics of the "1+1+N" detection system to fuse target position information and summarize the airspace target position and identity information for situational awareness display. Figure 1 As shown, a multi-source information fusion method for detecting slow, small targets includes the following steps:

[0028] Step 1: Based on the power indicators of photoelectric, radar, and radio detection equipment, establish a "1+1+N" integrated target perception matrix to acquire photoelectric images and data information of low, slow, and small targets from all directions;

[0029] Step 1-1: Calculate and obtain the monitoring range of each sensor.

[0030] The real-time detection range of the photoelectric device is as follows:

[0031]

[0032] Among them, R p Let m be the distance from the target to the system. For signal extraction factors; The effective radiation area of ​​the target, in cm 2 ; The radiance of the target. ; The number of pixels the target occupies on the focal plane: Background radiance, ; Let m be the entrance pupil area of ​​the optical system. 2 ; Atmospheric transmittance, calculated using ART or LOWTRAN7 software; The transmittance of the optical system during operation; The area of ​​a single pixel of the detector, in cm 2 ; The integral time during work, in seconds; For effective detectability in the band, SNR is the lowest detectable signal-to-noise ratio output by the detector.

[0033] Obtain parameters such as radar system antenna gain and transmit power to calculate detection range and effective distance:

[0034]

[0035] in, For radar transmission power, For the gain of the omnidirectional receiving antenna, For the gain of the directional transmission antenna, For radar signal wavelength, For a typical target (such as a series of UAVs), the cross-sectional area is... Boltzmann's constant, For noise temperature, To receive the equivalent bandwidth of the system, For identification coefficients, It is system loss. This represents the noise figure.

[0036] Acquire parameters such as radio detection equipment, antenna gain, and sensible signal power to calculate the detection range and effective distance;

[0037]

[0038] in, Typical target transmission power, For the gain of the omnidirectional receiving antenna, For directional antenna gain, Typical detection signal wavelength, This is the minimum signal power that the device can sense.

[0039] For drones, considering that the targets are all above the ground and the actual detection distance is often much smaller than the line-of-sight distance, the influence of the line-of-sight distance on the detection distance is not considered.

[0040] Steps 1-2: Based on the detection field of view and effective range of photoelectric, radar, and radio detection equipment, establish an integrated sensing matrix consisting of 1 radar, 1 radio detection device, and N photoelectric devices;

[0041] An integrated target perception matrix is ​​constructed using a "1+1+N" layout. One radar searches for moving targets and generates tracks, one radio detector identifies targets, and multiple electro-optical sensors perform coordinated surveillance. Based on the detection range calculation results in step 1-1 and the sensor deployment locations, the surveillance area is divided into a rectangular grid area of ​​Q meters × Q meters, forming a perception scheduling matrix, as shown below. Figure 5 As shown. Each element in the matrix is ​​represented by... express,

[0042] ,in,

[0043]

[0044] Where R is the radar participation identifier for the rectangular grid area in the j-th row and i-th column, where 1 represents participation and 0 represents non-participation, and D is the radio detection participation identifier for the rectangular grid area in the j-th row and i-th column. The nth optoelectronic device participates in the sensing identification of the rectangular grid area in the j-th row and i-th column, where N represents the total number of optoelectronic devices. This refers to the detection identifier of the m-th countermeasure device participating in the rectangular grid area in the j-th row and i-th column, where M represents the number of countermeasure devices. Let be the distance from the center point of the rectangular grid area in the j-th row and i-th column to the radar detection device. Let be the distance from the center point of the rectangular grid area in row j and column i to the radio detection device. Let be the distance from the center point of the rectangular grid area in row j and column i to the nth optoelectronic device. The distance from the center point of the rectangular grid area in the j-th row and i-th column to the m-th countermeasure device can be calculated using the latitude and longitude of the center point of the rectangular grid area and the latitude and longitude of the detection device. This represents the maximum detection distance of the nth photoelectric detection device. The maximum effective distance of the m-th countermeasure device is a definite constant.

[0045] Steps 1-3: Based on the perception matrix and photoelectric automatic scheduling process, acquire photoelectric images and data information from 1 radar, 1 radio detection device, and N photoelectric devices.

[0046] The number of points / tracks of radar targets is obtained directly from the fully automatic radar acquisition and processing. The acquired data specifically includes azimuth, elevation, range, altitude, heading, speed, Doppler velocity, echo amplitude, signal-to-noise ratio, and number of points.

[0047] Radio detection equipment obtains information through radio information processing, including: type, model, and azimuth angle.

[0048] Unlike radar and radio detection equipment, optoelectronic devices cannot simultaneously cover a 360° detection area. This invention employs the following strategy for automatic scheduling of each optoelectronic device. For devices falling into the matrix... Radar track in the middle, ,determination The value of , for a non-zero value and the nth photoelectric sensor being in an idle state, such as Figure 6 As shown, optoelectronic coordination is scheduled in the following manner:

[0049] First, the radar target position is converted into azimuth and elevation values ​​relative to the electro-optical system (if the target is detected by radio detection, radio detection only provides the azimuth value, while the elevation can simulate a dynamic value that gradually changes from 0° to 15°). This value guides the electro-optical system to a designated direction, and a designated search field of view is set based on the distance parameters provided by the radar (relative to the electro-optical device). Then, electro-optical image detection processing is initiated. By detecting whether the target appears in the image, the highest probability target area is selected, the tracker is initialized, and the target is locked and continuously tracked. Simultaneously, PID control of the turntable pointing and field of view and focus adjustment are performed to maintain the target size and clarity. Tracking status is also evaluated, primarily determining whether the target is being tracked stably and whether the azimuth is consistent with the radar (radio detection) guidance data. The tracking process is repeated until manually stopped or the guidance signal ends.

[0050] Among them, the automatic photoelectric target detection and selection strategy utilizes the Canny segmentation operator (Canny, J. "A Computational Approach To Edge Detection" IEEE Trans. Pattern Analysis and Machine Intelligence. 1986, (8): 679–714.) to calculate the bounding rectangle of the potential target, and the target grayscale features are:

[0051]

[0052] in, Let be the grayscale feature coefficient of a certain detection target. The grayscale value of a pixel. Image coordinates , The starting coordinates of the target area. , Set the end coordinates of the target region. Select the largest. The bounding rectangle of the target value is used as the selection area.

[0053] After selecting and identifying the target region, the standard KCF algorithm (Kernelized Correlation Filters, proposed in 2014 by Joao F. Henriques, Rui Caseiro, Pedro Martins, and Jorge Batista) is used for target locking and tracking, along with PID control and automatic field-of-view and focus adjustment. Consistency evaluation and stable tracking judgment are then performed based on the tracking results.

[0054] Tracking stability is determined by tracking feedback information using the KCF algorithm. If the tracking feedback is normal, the tracking stability judgment result is "Yes"; if the tracking fails, the tracking stability judgment result is "No".

[0055] The consistency evaluation mainly compares whether there is a difference of more than 3° between the radar guidance angle and the photoelectric monitoring angle. If the difference is greater than 3°, the result is "No"; if it is less than or equal to 3°, the result is "Yes".

[0056] This process can obtain point / track data of optoelectronic equipment targets, including information such as azimuth, pitch, size, grayscale, and image.

[0057] Step 2-1: Construct different signal-level (feature-level) base classifiers for radar, photoelectric, and radio detection respectively, and perform feature-level recognition, such as... Figure 6 As shown, obtain identity-level information;

[0058] Signal-level (characteristic-level) information, such as radar echo amplitude and signal-to-noise ratio, and identity-level information, such as identifying the target as a bird, drone, aircraft, and its model.

[0059] Different base classifiers are constructed for radar, photoelectric, and radio detection to identify and obtain identity-level information.

[0060] The radar data was analyzed using the ReliefF algorithm (Kononenko, I., Simec, E., & Robnik-Sikonja, M. (1997). Overcoming the myopia of inductive learning algorithms with RELIEFF. Retrieved from CiteSeerX). Target echo amplitude, signal-to-noise ratio, average echo spot width, target acceleration variance, and velocity variance were selected as target feature vectors. Figure 7 As shown, a multi-dimensional hyperplane classifier is constructed by cascading SVMs (Support Vector Machines), and the output target identity information is divided into four categories: "unknown", "bird", "airplane" and "drone".

[0061] Optical and photoelectric image target recognition is achieved by constructing a training sample library for "point targets," "birds," "drones," and "typical airborne objects." Based on historical data, aircraft landing gear areas are manually labeled with four categories: "point targets," "birds," "drones," and "typical airborne objects." The labels include the coordinates, width, and height of the bounding rectangle of the landing gear, as well as the category label. Multiple labeled samples are accumulated to form the sample library. A deep convolutional neural network detection model, YOLOv3, is used to train on the samples, with four categories: "point targets," "birds," "drones," and "typical airborne objects." Parameters such as the learning rate are adjusted during training based on the sample library, and the process stops after model convergence. Finally, the converged YOLOv3 model is obtained through training and used for real-time image prediction to obtain the optical and photoelectric target identity information: "point targets," "birds," "drones," and "typical airborne objects." "Point targets" are those targets that are too small to be identified by specific type but can be determined as aerial targets.

[0062] Radio detection employs a sample database comparison method, acquiring and storing typical target communication signals (with manually labeled target identification information). The identification capability is improved by increasing the number of samples in the database. If a newly discovered radio signal matches information stored in the database (using the k-nearest neighbor (KNN) criterion), the labeled identification information is output. Output categories include "unknown" and "drone," with the drone model also included, such as "drone model X."

[0063] Step 2-2: Combine data relationships to perform identity-level information fusion and reasoning to obtain a comprehensive recognition result.

[0064] After classifying and identifying the associated data using multiple methods such as photoelectric, radar, and radio detection to obtain identity-level information, the identity information is then inferred, categorized, and weighted by voting to obtain a comprehensive identification result. The processing flow is as follows: Figure 8 As shown:

[0065] The main identification information includes two parts: "target type" and "target model." Target type can be provided and identified by multiple sensors, so a weighted voting method is used. Radar, radio detection, electro-optical 1, ..., electro-optical N results are all multiplied by a weighted coefficient before voting to obtain a comprehensive target type result. Target types include: "unidentified," "birds," "drones," "airborne objects," and "aircraft." Target model is provided by radio detection, provided the target type is "drone."

[0066] The target type is determined using a multi-sensor weighted voting method. The goal is to ultimately identify the category corresponding to the category with the largest number of votes.

[0067] The voting values ​​for each category are calculated as follows:

[0068] in, , , , , The voting weights are respectively for radar, radio detection, and N optoelectronic devices. This is the type information detected by the radar. When the radar detects a related target and the target category is class, its value is 1; otherwise, it is 0. , , , This corresponds to the type information detected by radio detection and N optoelectronic devices.

[0069] Step 3 includes the following steps:

[0070] Step 3-1: Receive the secondary code message from ADS-B and determine whether the target is a cooperative aircraft target.

[0071] Receive ADS-B information, which is aircraft target information, including target address code, A / C mode, target position, and altitude information. Using the ADS-B target latitude and longitude information, assign the ADS-B target to the appropriate location. Target perception matrix unit, based on In the target detection process, radar, electro-optical, and radio detection targets are paired with ADS-B data. For paired targets, the target type is adjusted to "aircraft".

[0072] Step 3-2: Match and identify targets from the whitelist in the radio detection database to distinguish between unauthorized drone targets and cooperative drone targets.

[0073] After detecting a target type of "drone," the main spectral characteristics are saved through spectral data recording. Drones known to be in the whitelist are named in the whitelist. If a spectral characteristic matching the whitelist appears again, the target is considered a cooperative drone. The matching degree is defined as follows:

[0074]

[0075] in This is the nth whitelist feature vector, composed of five scalar values: instantaneous phase standard deviation, variance, mean, peak value, and instantaneous frequency. When the matching degree... Less than If the threshold is met, it is considered a target matching the whitelist; otherwise, it is considered a target flying illegally.

[0076] Step 3-3: Coordinated handling of unauthorized flying targets.

[0077] After determining whether the identified drone targets are flying illegally in step 3-2, the target's affiliation is then determined. middle The device number corresponding to the non-zero item. Select. The two smallest corresponding countermeasure devices perform interference operations, and during the process of interfering with and driving away the unauthorized flying target, they are based on their real-time... Value, real-time selection and switching of countermeasure devices on and pointing to The area is monitored until the target disappears or flies away from the jurisdiction.

[0078] Step 4 includes the following steps:

[0079] Step 4-1: Design fusion rules based on the characteristics of the "1+1+N" detection system, and perform target location information fusion.

[0080] In acquiring the target's integrated track, this system uses a "1+1+N" matching method. Both radio detection and electro-optical direction finding only perform lateral direction finding on the target, and the accuracy of radio detection is relatively low. The integrated track is generated based on the combination of radar track and electro-optical direction finding information. The specific target track point locations are generated as follows:

[0081] Using the radar's location coordinates as the origin of a two-dimensional rectangular coordinate system, with due north as the positive Y-axis and due east as the positive X-axis, assume that the nearest photoelectric sensor n to the radar is located at... The target drone has been locked, and its location is... , To detect the angle between the target and the due east direction (values) ), looking up and down , To detect the angle between the target and the horizontal direction (values) The radar detects the target at a projected range of R, with an azimuth of [missing information]. , To detect the angle between the target and the due east direction (values) After fusion, the target fusion point is:

[0082] It can be obtained from the following calculations:

[0083]

[0084] exist , Two solutions are obtained by calculating the angle between the corresponding points and the due east direction. , ,calculate , and The angle difference is used to determine the final fusion target position by selecting the solution with the smallest angle difference. The target height value can be obtained using the following formula.

[0085]

[0086] Step 4-2: Situation display of the summarized airspace target locations and identification information, including comprehensive flight paths, target types, models, and other information.

[0087] Target type and model information is generated through steps 2-2, 3-1, and 3-2. Combined with the fused flight track information obtained in step 4-1, this information is integrated with GIS to assess target situation and present the overall drone detection situation within the jurisdiction. The displayed content includes target flight track number (unique target identifier), latitude and longitude, altitude, speed, heading, type (including unidentified, drone, bird, aircraft, and airborne object), model (e.g., a specific drone model), and information on unauthorized flights.

[0088] Example:

[0089] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0090] like Figure 2As shown, the unmanned air defense system specifically comprises information acquisition equipment, countermeasure equipment, data processing equipment, and situation display equipment. The information acquisition equipment includes radar, electro-optical, and radio detection equipment; the countermeasure equipment includes electromagnetic interference suppression equipment; the data processing equipment includes image processing workstations and data processing workstations; and the situation display equipment includes a situation display workstation and a display screen. The electro-optical images and related data information in the system are acquired through SDI image acquisition channels and Ethernet communication channels, respectively. The system's detection, fusion, and countermeasure functions are uniformly processed by software. The software runs on image processing workstations and data processing workstations, while the display software runs on a situation display workstation. All processing flows in this invention are implemented through software.

[0091] This invention discloses a multi-source information fusion method for detecting low-speed, small targets and an unmanned air defense system, comprising the following steps:

[0092] (1) Based on the power indicators of photoelectric, radar and radio detection equipment, a “1+1+N” integrated target perception matrix is ​​established to obtain photoelectric images and data information of low, slow and small targets in all directions.

[0093] First, dynamically acquire the power indicators of photoelectric, radar, and radio detection equipment, and then determine the capability range of the equipment based on typical target parameters and equipment parameter states.

[0094] The real-time detection range of the photoelectric device is as follows:

[0095]

[0096] Among them, R p Let m be the distance from the target to the system. For signal extraction factors; The effective radiation area of ​​the target, in cm2; The radiance of the target. ; The number of pixels the target occupies on the focal plane: Background radiance, ; Let m be the entrance pupil area of ​​the optical system; Atmospheric transmittance, calculated using ART or LOWTRAN7 software; The transmittance of the optical system during operation; The area of ​​a single pixel of the detector is in cm². The integral time during work, in seconds; For effective detectability in the band, SNR is the lowest detectable signal-to-noise ratio output by the detector. The detection range of a typical UAV target is usually calculated as shown in Table 1. If a certain type of UAV is taken as a typical target, the detection range is 3.7km.

[0097] Table 1. Detection Range of Typical UAV Targets by Electro-optical Systems

[0098]

[0099] Obtain parameters such as radar system antenna gain and transmit power to calculate detection range and effective distance:

[0100]

[0101] in, For radar transmission power, For the gain of the omnidirectional receiving antenna, For the gain of the directional transmission antenna, For radar signal wavelength, This represents the typical target's reflective cross-section. Boltzmann's constant, For noise temperature, To receive the equivalent bandwidth of the system, For identification coefficients, It is system loss. This represents the noise figure. The detection range for a typical UAV (with a reflective cross-sectional area of ​​0.01 meters for a certain model) is approximately 8.8–9.3 km.

[0102] Acquire parameters such as radio detection equipment, antenna gain, and sensible signal power to calculate the detection range and effective distance;

[0103]

[0104] in, Typical target transmission power, For the gain of the omnidirectional receiving antenna, For directional antenna gain, Typical detection signal wavelength, This represents the minimum signal power that the device can detect. For a certain series of drones, the detection range is approximately 5.5 km.

[0105] For drones, considering that the targets are all above the ground and the actual detection distance is often much smaller than the line-of-sight distance, the influence of the line-of-sight distance on the detection distance is not considered.

[0106] Based on the detection field of view and effective range of optoelectronic, radar, and radio detection equipment, an integrated sensing matrix is ​​established consisting of one radar, one radio detection device, and N optoelectronic devices;

[0107] A grid-like matrix region is constructed within the area that can cover the maximum range of all detection devices in the region. This region is divided into rectangular grid areas of Q meters × Q meters (typical parameter 100 meters × 100 meters; setting too small is not recommended, otherwise there may be matching failures during ADS-B and radar matching). This forms the sensing and scheduling matrix. Figure 5 As shown. Each element in the matrix is ​​represented by... express,

[0108] ,in,

[0109]

[0110] Where R is the radar participation identifier for the rectangular grid area in the j-th row and i-th column, where 1 represents participation and 0 represents non-participation, and D is the radio detection participation identifier for the rectangular grid area in the j-th row and i-th column. The nth optoelectronic device participates in the sensing identification of the rectangular grid area in the j-th row and i-th column, where N represents the total number of optoelectronic devices. This refers to the detection identifier of the m-th countermeasure device participating in the rectangular grid area in the j-th row and i-th column, where M represents the number of countermeasure devices. Let be the distance from the center point of the rectangular grid area in the j-th row and i-th column to the radar detection device. Let be the distance from the center point of the rectangular grid area in row j and column i to the radio detection device. Let be the distance from the center point of the rectangular grid area in row j and column i to the nth optoelectronic device. The distance from the center point of the rectangular grid area in the j-th row and i-th column to the m-th countermeasure device can be calculated using the latitude and longitude of the center point of the rectangular grid area and the latitude and longitude of the detection device. This represents the maximum detection distance of the nth photoelectric detection device. The maximum effective distance of the m-th countermeasure device is a definite constant.

[0111] Based on the perception matrix and photoelectric automatic scheduling process, photoelectric images and data information from one radar, one radio detection device, and N photoelectric devices are acquired.

[0112] The number of points / tracks of radar targets is obtained directly from the fully automatic radar acquisition and processing. The acquired data specifically includes azimuth, elevation, range, altitude, heading, speed, Doppler velocity, echo amplitude, signal-to-noise ratio, and number of points.

[0113] Radio detection equipment obtains information through radio information processing, including: type, model, and azimuth angle.

[0114] Unlike radar and radio detection equipment, optoelectronic devices cannot simultaneously cover a 360° detection area. This invention employs the following strategy for automatic scheduling of each optoelectronic device. For devices falling into the matrix... Radar track in the middle, ,determination The value of , for a non-zero value and the nth photoelectric sensor being in an idle state, such as Figure 6 As shown, optoelectronic coordination is scheduled in the following manner:

[0115] First, the radar target position is converted into azimuth and elevation values ​​relative to the electro-optical system (if the target is detected by radio detection, radio detection only provides the azimuth value, while the elevation can simulate a dynamic value that gradually changes from 0° to 15°). This value guides the electro-optical system to a designated direction, and a designated search field of view is set based on the distance parameters provided by the radar (relative to the electro-optical device). Then, electro-optical image detection processing is initiated. By detecting whether the target appears in the image, the highest probability target area is selected, the tracker is initialized, and the target is locked and continuously tracked. Simultaneously, PID control of the turntable pointing and field of view and focus adjustment are performed to maintain the target size and clarity. Tracking status is also evaluated, primarily determining whether the target is being tracked stably and whether the azimuth is consistent with the radar (radio detection) guidance data. The tracking process is repeated until manually stopped or the guidance signal ends.

[0116] The automatic photoelectric target detection and selection strategy utilizes the Canny segmentation operator to calculate the bounding rectangle of potential targets, and the target grayscale features are as follows:

[0117]

[0118] in, Let be the grayscale feature coefficient of a certain detection target. The grayscale value of a pixel. Image coordinates , The starting coordinates of the target area. , Set the end coordinates of the target region. Select the largest. The bounding rectangle of the target value is used as the selection area.

[0119] After selecting and identifying the target area, the standard KCF algorithm is used for target locking and tracking, along with PID control and automatic field-of-view and focus adjustment. Consistency evaluation and stability assessment are then performed based on the tracking results.

[0120] Tracking stability is determined by tracking feedback information using the KCF algorithm. If the tracking feedback is normal, the tracking stability judgment result is "Yes"; if the tracking fails, the tracking stability judgment result is "No".

[0121] The consistency evaluation mainly compares whether there is a difference of more than 3° between the radar guidance angle and the photoelectric monitoring angle. If the difference is greater than 3°, the result is "No"; if it is less than or equal to 3°, the result is "Yes".

[0122] This process can obtain point / track data of optoelectronic equipment targets, including information such as azimuth, pitch, size, grayscale, and image.

[0123] (2) Establish a multi-source information fusion and target classification model, construct base classifiers for photoelectric, radar, and radio detection to process feature-level information, and perform comprehensive analysis at the identity level to obtain multi-source verification and identification, such as... Figure 4 As shown.

[0124] Different signal-level (feature-level) base classifiers are constructed for radar, photoelectric, and radio detection respectively to perform feature-level recognition and obtain identity-level information;

[0125] Signal-level (characteristic-level) information, such as radar echo amplitude and signal-to-noise ratio, and identity-level information, such as identifying the target as a bird, drone, aircraft, and its model.

[0126] Different base classifiers are constructed for radar, photoelectric, and radio detection to identify and obtain identity-level information.

[0127] The radar data uses the ReliefF algorithm to select features such as target echo amplitude, signal-to-noise ratio, average width of echo points, target acceleration variance, and velocity variance as target feature vectors. Figure 7 As shown, a multi-dimensional hyperplane classifier is constructed using SVM cascade, and the output target identity information is divided into four categories: "unknown", "bird", "airplane" and "drone".

[0128] Optical and photoelectric image target recognition is achieved by constructing a training sample library for "point targets," "birds," "drones," and "typical airborne objects." Based on historical data, aircraft landing gear areas are manually labeled with four categories: "point targets," "birds," "drones," and "typical airborne objects." The labels include the coordinates, width, and height of the bounding rectangle of the landing gear, as well as the category label. Multiple labeled samples are accumulated to form the sample library. A deep convolutional neural network detection model, YOLOv3, is used to train on the samples, with four categories: "point targets," "birds," "drones," and "typical airborne objects." Parameters such as the learning rate are adjusted during training based on the sample library, and the process stops after model convergence. Finally, the converged YOLOv3 model is obtained through training and used for real-time image prediction to obtain the optical and photoelectric target identity information: "point targets," "birds," "drones," and "typical airborne objects." "Point targets" are those targets that are too small to be identified by specific type but can be determined as aerial targets.

[0129] Radio detection employs a sample database comparison method, acquiring and storing typical target communication signals (with manually labeled target identification information). The identification capability is improved by increasing the number of samples in the database. If a newly discovered radio signal matches information stored in the database (using the KNN criterion), the labeled identification information is output. Output categories include "unknown" and "drone," with the drone model also included, such as "drone model X."

[0130] A comprehensive identification result is obtained by combining data correlations and performing identity-level information fusion and reasoning. For the correlated data, multiple methods such as photoelectric, radar, and radio detection are used for classification and identification to obtain identity-level information. Then, identity information is reasoned, categorized, and weighted voting is performed to obtain the comprehensive identification result. The processing flow is as follows: Figure 8 As shown:

[0131] The main identification information includes two parts: "target type" and "target model." Target type can be provided and identified by multiple sensors, so a weighted voting method is used. Radar, radio detection, electro-optical 1, ..., electro-optical N results are all multiplied by a weighted coefficient before voting to obtain a comprehensive target type result. Target types include: "unidentified," "birds," "drones," "airborne objects," and "aircraft." Target model is provided by radio detection, provided the target type is "drone."

[0132] The target type is determined using a multi-sensor weighted voting method. The goal is to ultimately identify the category corresponding to the category with the largest number of votes.

[0133] The voting values ​​for each category are calculated as follows:

[0134]

[0135] in, , , , , The voting weights are respectively for radar, radio detection, and N optoelectronic devices. This is the type information detected by the radar. When the radar detects a related target and the target category is class, its value is 1; otherwise, it is 0. , , , Corresponding to the type information detected by radio detection and N optoelectronic devices

[0136] (3) Use ADS-B data to identify cooperative aircraft targets, use a whitelist spectrum library for radio detection to identify cooperative UAVs and unauthorized UAVs, and determine the interference strategy based on the perception matrix area to which the target belongs for unauthorized targets.

[0137] Receive ADS-B information, which is aircraft target information, including target address code, A / C mode, target position, and altitude information. Using the ADS-B target latitude and longitude information, assign the ADS-B target to the appropriate location. Target perception matrix unit, based on In the target detection process, radar, electro-optical, and radio detection targets are paired with ADS-B data. For paired targets, the target type is adjusted to "aircraft," and the attribute is "aircraft," which means it is a cooperative aircraft target and is no longer identified as a drone.

[0138] The system uses a whitelist matching function within a radio detection database to distinguish between unauthorized drone targets and cooperative drone targets.

[0139] After detecting a target type of "drone," the main spectral characteristics are saved through spectral data recording. Drones known to be in the whitelist are named in the whitelist. If a spectral characteristic matching the whitelist appears again, the target is considered a cooperative drone. The matching degree is defined as follows:

[0140]

[0141] in This is the nth whitelist feature vector, composed of five scalar values: instantaneous phase standard deviation, variance, mean, peak value, and instantaneous frequency. When the matching degree... Less than If the threshold is met, it is considered a target matching the whitelist; otherwise, it is considered a target flying illegally.

[0142] For the drone targets identified through comprehensive analysis, after determining whether they are flying illegally, the target's affiliation is determined. middle The device number corresponding to the non-zero item. Select. The two smallest corresponding countermeasure devices perform interference operations, and during the process of interfering with and driving away the unauthorized flying target, they are based on their real-time... Value, real-time selection and switching of countermeasure devices on and pointing to The area is monitored until the target disappears or flies away from the jurisdiction.

[0143] (4) Based on the characteristics of the “1+1+N” detection system, design fusion rules, fuse target location information, and summarize the airspace target location and identity information for situation display.

[0144] In acquiring the target's integrated track, this system uses a "1+1+N" matching method. Both radio detection and electro-optical direction finding only perform lateral direction finding on the target, and the accuracy of radio detection is relatively low. The integrated track is generated based on the combination of radar track and electro-optical direction finding information. The specific target track point locations are generated as follows:

[0145] Using the radar's location coordinates as the origin of a two-dimensional rectangular coordinate system, with due north as the positive Y-axis and due east as the positive X-axis, assume that the nearest photoelectric sensor n to the radar is located at... The target drone has been locked, and its location is... , To detect the angle between the target and the due east direction (values) ), looking up and down , To detect the angle between the target and the horizontal direction (values) The radar detects the target at a projected range of R, with an azimuth of [missing information]. , To detect the angle between the target and the due east direction (values) After fusion, the target fusion point is:

[0146] It can be obtained from the following calculations:

[0147]

[0148] exist , Two solutions are obtained by calculating the angle between the corresponding points and the due east direction. , ,calculate , and The angle difference is used to determine the final fusion target position by selecting the solution with the smallest angle difference. The target height value can be obtained using the following formula.

[0149]

[0150] After obtaining information such as flight paths, the relevant information is combined with GIS to analyze the target situation and present the overall situation of UAV detection within the jurisdiction, such as... Figure 3 As shown. The displayed content includes target track number (unique identifier of the target), latitude and longitude, altitude, speed, heading, type (including unidentified, drone, bird, aircraft, airborne object), model (e.g., "a certain model of drone"), and information such as illegal flight identification.

[0151] In its specific implementation, this application provides a computer storage medium and a corresponding data processing unit. The computer storage medium is capable of storing a computer program, which, when executed by the data processing unit, can run the invention's content regarding a multi-source information fusion method for detecting low-speed, slow-moving, and small targets and an unmanned air defense system, as well as some or all of the steps in various embodiments. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0152] Those skilled in the art will clearly understand that the technical solutions in the embodiments of the present invention can be implemented using computer programs and their corresponding general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of computer programs, i.e., software products. These computer program software products can be stored in a storage medium and include several instructions to cause a device containing a data processing unit (which may be a personal computer, server, microcontroller, MUU, or network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments of the present invention.

[0153] This invention provides a method for detecting low-altitude, slow-moving, and small targets using multi-source information fusion, as well as the concept and approach of an unmanned air defense system. Many methods and approaches exist for implementing this technical solution; the above description is merely a preferred embodiment of the invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications should also be considered within the scope of protection of this invention. All components not explicitly stated in this embodiment can be implemented using existing technologies.

Claims

1. A method for detecting low-speed, small targets using multi-source information fusion, characterized in that, Includes the following steps: Step 1: Based on the power indicators of photoelectric, radar and radio detection equipment, establish an integrated target perception matrix to acquire photoelectric images and data information of the low, slow and small targets; In this context, each element in the integrated sensing matrix is ​​used... It is expressed as follows: ； Where R is the radar's identification identifier for the rectangular grid area in the j-th row and i-th column, and D is the radio detection identification identifier for the rectangular grid area in the j-th row and i-th column. The nth optoelectronic device participates in the sensing identification of the rectangular grid area in the j-th row and i-th column, where N represents the total number of optoelectronic devices. This is the identification marker for the m-th countermeasure device participating in the rectangular grid area in the j-th row and i-th column, where M represents the number of countermeasure devices; Step 2: Establish a multi-source information fusion and target classification model. Construct base classifiers for photoelectric, radar and radio detection to process feature-level information, and integrate them at the identity level to obtain multi-source verification and identification results, i.e., comprehensive identification results. Step 3: Use ADS-B data to identify cooperative aircraft targets, and use a radio detection whitelist spectrum library to distinguish between cooperative UAV targets and unauthorized UAV targets. For unauthorized UAV targets, determine the jamming strategy based on their respective sensing matrix regions, and coordinate the handling of unauthorized UAV targets. The specific methods are as follows: For targets of unauthorized drone flights, determine their affiliation. middle Select the countermeasure device number corresponding to the non-zero item. The two smallest corresponding countermeasures devices perform interference operations; during the process of interfering with and driving away the black-flying drone target, the real-time data is used to... Value, real-time selection and switching of countermeasure devices on and pointing to The area is monitored until the target disappears or flies out of the jurisdiction; among which, This represents the distance from the center point of the rectangular grid area in row j and column i to the countermeasure device m. Step 4: Design fusion rules, fuse target location information, and summarize the airspace target location and identity information for situational display.

2. The method for detecting low-speed, small targets by multi-source information fusion according to claim 1, characterized in that, Step 1, which involves acquiring the photoelectric image and data information of the low-speed, small target, specifically includes the following methods: Step 1-1: Obtain the thermal sensitivity and focal length parameters of the photoelectric detection device, and calculate the detection field of view and effective range of the photoelectric detection device; obtain the antenna gain and receiver sensitivity parameters of the radar detection device, and calculate the detection field of view and effective range of the radar detection device; obtain the receiving antenna gain and detection wavelength parameters of the radio detection device, and calculate the detection field of view and effective range of the radio detection device; the specific method is as follows: Step 1-1-1: Calculate the detection field of view and effective distance of the photoelectric detection device, as follows: ； Among them, R p The effective distance from the target to the photoelectric detection equipment; For signal extraction factors; The effective radiation area of ​​the target; The radiance of the target; The number of pixels the target occupies on the focal plane; Background radiance; The entrance pupil area of ​​the optical system; Atmospheric transmittance; The transmittance of the optical system during operation; The area of ​​a single pixel in the detector; This refers to the time spent working. The effective detectivity of the band; SNR is the lowest signal-to-noise ratio; Step 1-1-2: Calculate the detection field of view and effective range of the radar detection equipment, using the following method: ； in, The effective range of the target from the radar detection equipment; For radar transmission power, For the gain of the omnidirectional receiving antenna, For the gain of the directional transmission antenna, For radar signal wavelength, This represents the typical target's reflective cross-section. Boltzmann's constant, For noise temperature, To receive the equivalent bandwidth of the system, For identification coefficients, It is system loss. Noise figure; Step 1-1-3: Calculate the detection field of view and effective range of the radio detection equipment, using the following method. ； in, The effective range from the target to the radio detection equipment; Typical target transmission power, For the gain of the omnidirectional receiving antenna, For directional antenna gain, For radar signal wavelength, The minimum signal power that the device can sense; Steps 1-2: Based on the detection field of view and effective range of photoelectric, radar and radio detection equipment, establish an integrated perception matrix consisting of 1 radar detection device, 1 radio detection device and N photoelectric detection devices; Each element in the integrated sensing matrix is ​​used middle: ； Where R is the radar participation identifier for the rectangular grid area in the j-th row and i-th column, where 1 represents participation and 0 represents non-participation; and D is the radio detection participation identifier for the rectangular grid area in the j-th row and i-th column. The nth optoelectronic device participates in the sensing identification of the rectangular grid area in the j-th row and i-th column, where N represents the total number of optoelectronic devices. This refers to the detection identifier of the m-th countermeasure device participating in the rectangular grid area in the j-th row and i-th column, where M represents the number of countermeasure devices. Let be the distance from the center point of the rectangular grid area in row j and column i to the radar detection device. Let be the distance from the center point of the rectangular grid area in row j and column i to the radio detection device. The distance is the distance from the center point of the rectangular grid area in the j-th row and i-th column to the n-th photoelectric detection device. The distance from the center point of the rectangular grid area in the j-th row and i-th column to the m-th countermeasure device; This represents the maximum detection range of the nth photoelectric detection device. The maximum effective range of the m-th countermeasure device is a definite constant. Steps 1-3: Based on the integrated sensing matrix and photoelectric automatic scheduling process, acquire the photoelectric images and data information of the target detected by one radar detection device, one radio detection device, and N photoelectric detection devices; The data information of the target detected by the radar detection equipment includes: the target's azimuth, elevation, range, altitude, heading, speed, Doppler velocity, echo amplitude, signal-to-noise ratio, and number of points; The data information of the target detected by the radio detection equipment includes: the type, model, and azimuth angle of the target; According to the photoelectric automatic scheduling process, the target data information detected by the photoelectric detection equipment includes: the target's azimuth, elevation, size, grayscale, and photoelectric image.

3. The method for detecting low-speed, small targets by multi-source information fusion according to claim 2, characterized in that, The photoelectric automatic scheduling process described in steps 1-3 specifically includes: For falling into the matrix The trajectory of the target detected by the radar detection equipment in the middle. ,determination For a value that is not 0 and the nth photoelectric detection device is in an idle state, the photoelectric coordination is scheduled in the following manner: First, the target position detected by radar detection equipment is converted into azimuth and elevation values ​​relative to the electro-optical detection equipment. If the target is detected by radio detection equipment, the radio detection equipment only provides the azimuth value. The electro-optical detection equipment is guided to point in a specified direction using the azimuth and elevation values, and a specified search field of view is set based on the distance parameters provided by the radar detection equipment. Then, according to the electro-optical target automatic detection and selection strategy, electro-optical image detection processing is performed. Depending on whether a target appears during the processing, the target area is selected, the tracker of the electro-optical detection equipment is initialized, and the target is locked and continuously tracked. At the same time, PID control of the turntable pointing of the electro-optical detection equipment and field of view and focus adjustment are performed to maintain the target size and clarity. The tracking status is evaluated to determine whether the target is being tracked stably and whether the azimuth is consistent with the radar or radio detection guidance data. The above tracking process is repeated until manual stopping or the guidance signal ends. Among them, the automatic photoelectric target detection and selection strategy uses the Canny segmentation operator to calculate the bounding rectangle of potential targets, and the target grayscale features are: ； in, The grayscale feature coefficients of the target to be detected. The grayscale value of a pixel. Image coordinates , The starting coordinates of the target area. , Set the end coordinates of the target region; select the largest. The bounding rectangle of the target value is used as the selection area; After selecting the target area that matches the target, the standard KCF algorithm is used to lock and track the target, and consistency evaluation and stable tracking judgment are performed based on the tracking situation.

4. The method for detecting low-speed, small targets by multi-source information fusion according to claim 3, characterized in that, Step 2 includes the following steps: Step 2-1: For the photoelectric images and data information detected by radar, photoelectric, and radio detection equipment, respectively, construct three different signal-level (i.e., feature-level) base classifiers to perform feature-level recognition and obtain the target's identity-level information; the specific method is as follows: Step 2-1-1: Based on the data information detected by the radar detection equipment, the ReliefF algorithm is used to select the target's echo amplitude, signal-to-noise ratio, average width of echo points, target acceleration variance, and velocity variance as the target's feature vector. A multi-dimensional hyperplane classifier is constructed using SVM cascade, and the target's identity information is output as four categories: unidentified, birds, aircraft, and drones. Step 2-1-2 involves performing photoelectric image target recognition based on the photoelectric images and data information detected by the photoelectric detection device, as detailed below: A training sample library for point targets, birds, drones, and typical airborne objects is constructed. The aircraft landing gear area is manually labeled based on historical data. The labeling categories are point targets, birds, drones, and typical airborne objects. The labeling includes the coordinates, width, height, and category label of the bounding rectangle of the landing gear. The labeled samples are accumulated to form a sample library. The YOLOv3 deep convolutional neural network detection model was adopted, with four categories: point targets, birds, drones, and typical airborne objects. The learning rate parameter was adjusted, and the model was trained based on the sample library. The training stopped after the model converged. By training a converged YOLOV3 model, the photoelectric image is predicted in real time to obtain the target's identity information: point target, bird, drone, or typical airborne object. Step 2-1-3: For the data information detected by the radio detection equipment, a sample library comparison method is used to establish a radio detection library as a sample library. Typical target communication signals are acquired and stored by manually annotating target identification information. If a newly discovered radio signal matches the information stored in the radio detection database, the identified information, i.e. the target's identity level information, is output, including: unidentified or drone, and the drone model information is also output. Step 2-2: Combine data association relationships to perform identity-level information fusion and reasoning to obtain a comprehensive recognition result; the information fusion and reasoning methods include reasoning and classification of the identity-level information and weighted voting, and the specific methods are as follows: The comprehensive identification result, i.e. the final identification information, includes two elements: target type and target model. Target types include: unidentified, birds, drones, airborne objects, and aircraft; when the target type is a drone, the target model is obtained by radio detection equipment. The target type is determined by weighted voting. The goal is to ultimately identify the category that corresponds to the category with the highest vote value; the vote values ​​for each category are calculated as follows: ； in, , , and The voting weights are assigned to radar, radio, and N photoelectric detection devices, respectively. This is the type information detected by the radar detection equipment. When the radar detects an associated target and the target category is class, its value is 1; otherwise, it is 0. , , and This corresponds to the type information detected by radio waves and N photoelectric detection devices.

5. The method for detecting low-speed, small targets by multi-source information fusion according to claim 4, characterized in that, Step 3 includes the following steps: Step 3-1: Receive ADS-B messages and use ADS-B data to identify cooperative aircraft targets; Step 3-2: Match and identify targets from the whitelist in the radio detection database to distinguish between unauthorized drones and cooperative drones; Step 3-3: Collaborative handling of unauthorized drone targets.

6. The method for detecting low-speed, small targets by multi-source information fusion according to claim 5, characterized in that, The specific method for identifying cooperative aircraft targets using ADS-B data as described in step 3-1 is as follows: Receive ADS-B information, and using the target latitude and longitude information in the ADS-B information, assign the target that sent the ADS-B information to the [missing information - likely a specific location or location]. Target perception matrix unit, according to The target detection process associates and pairs radar, electro-optical, and radio detection targets with ADS-B information. For successfully paired targets, the target type is adjusted to aircraft, i.e., it is identified as a cooperative aircraft target.

7. The method for detecting low-speed, small targets by multi-source information fusion according to claim 6, characterized in that, The specific method for distinguishing between unauthorized drone targets and cooperative drone targets, as described in step 3-2, is as follows: For targets of type UAV, spectral feature information is saved through spectral data recording. UAVs in the whitelist are named accordingly. Based on the matching degree, if a spectral feature matching the whitelist reappears, the target is considered a cooperative UAV. The matching degree is defined as follows: ； in, Let k be the whitelist feature vector. The vector consists of five scalar values: instantaneous phase standard deviation, variance, mean, peak value, and instantaneous frequency. When the matching degree... Less than the threshold If a target is identified as a whitelisted target (i.e., a cooperative drone), it is considered a target of unauthorized drone flight.

8. The method for detecting low-speed, small targets by multi-source information fusion according to claim 7, characterized in that, Step 4 includes the following steps: Step 4-1: Design fusion rules to fuse target location information. Specific methods include: Using the coordinates of the radar detection equipment's location as the origin of a two-dimensional rectangular coordinate system, with due north as the positive Y-axis and due east as the positive X-axis, assume that the nth photoelectric detection equipment closest to the radar is located at... The target drone has been locked, and its location is... , The angle between the detected target and the due east direction is set to a value. , looking up and down , The angle between the detected target and the horizontal direction is set to a value. The projected distance between the radar detection equipment and the target is R, and the azimuth is... , The angle between the detected target and the due east direction is set to a value. The target fusion point is: It is calculated using the following method: ； exist , Two solutions are obtained by calculating the angle between the corresponding points and the due east direction. and ,calculate , and The angle difference is used to select the solution with the smallest angle difference as the final fusion target position. The target height value is obtained by the following formula: ； Step 4-2: Situation display of the summarized airspace target locations and identification information, including comprehensive flight paths and target type, i.e., model information.

9. An unmanned air defense system, employing the multi-source information fusion method for detecting low-speed, small targets as described in any one of claims 1-8, characterized in that, include: It consists of four parts: information acquisition equipment, countermeasure equipment, data processing equipment, and situation display equipment; The information acquisition equipment consists of photoelectric, radar, and radio detection, signal conversion, and transmission equipment, which acquires target points, tracks, and image data; the countermeasure equipment uses electromagnetic interference to drive away drones; the data processing equipment consists of an image processing workstation, a data processing workstation, and signal conversion and transmission equipment, which performs target detection, target tracking, information fusion, and target classification; and the situation display equipment consists of a situation display workstation and signal conversion and transmission equipment, which performs human-machine interaction, situation display, and anomaly alarms.

Images