Target identification method and system based on multi-source information fusion

Abstract

The invention provides a target identification method and system based on multi-source information fusion, and relates to the technical field of low-altitude target detection. The method comprises the following steps: acquiring target radar track data, interception equipment track data and position area information data; based on target radar track data, extracting a first feature in an RCS form dimension, and extracting a second feature in a motion dimension; based on the target radar track data and the track data of the monitoring equipment, determining a frequency spectrum monitoring correlation factor and regional position information features; performing feature fusion on the first feature, the second feature, the spectrum interception correlation factor and the regional position information feature to obtain a target feature, and identifying a target type; and identifying a target threat level based on the target type and the target radar track data. The method and the device are used in a target identification process based on multi-source information fusion, and the technical problem that the target threat degree cannot be accurately identified in a complex environment in the prior art is solved.

Description

Technical Field

[0001] This application relates to the field of low-altitude target detection technology, and in particular to a target recognition method and system based on multi-source information fusion. Background Technology

[0002] With the increasing density of low-altitude activities, various low-speed, small targets and airspace targets are becoming increasingly complex in terms of quantity, type, and behavioral patterns, placing greater pressure on surveillance systems for identification and security management. In practical applications, target movement is highly variable, and radar observation characteristics are easily affected by environmental factors, terrain, and equipment performance, resulting in unstable and uncertain target states. Often, only basic attribute judgments can be made, making it difficult to accurately distinguish the degree of potential danger based on the target's actual behavioral characteristics and airspace situation. This leads to difficulties in timely and accurate assessment of whether a target poses a threat in complex scenarios. Therefore, how to accurately identify the threat level of targets in complex environments is a problem that urgently needs to be solved in related fields. Summary of the Invention

[0003] This application provides a target identification method and system based on multi-source information fusion, which solves the technical problem that existing technologies cannot accurately identify the threat level of targets in complex environments.

[0004] To achieve the above objectives, this application adopts the following technical solution: Firstly, a target identification method based on multi-source information fusion is provided, comprising: acquiring target radar track data, listening device track data, and site area information data; extracting a first feature in the RCS morphology dimension and a second feature in the motion dimension based on the target radar track data; determining spectrum listening correlation factors and site information features based on the target radar track data and listening device track data; fusing the first feature, second feature, spectrum listening correlation factors, and site information features to obtain target features and identify target type; and identifying target threat level based on target type and target radar track data.

[0005] In conjunction with the first aspect mentioned above, in one possible implementation, based on the target radar track data, a first feature is extracted in the RCS morphology dimension, and a second feature is extracted in the motion dimension. This includes: based on the target radar track data, extracting the altitude, velocity, amplitude, azimuth width, azimuth thickness, and the number of times the target passes through preset event points from N frames of the target radar track data in the RCS morphology dimension to construct a feature set; based on the feature set, performing feature dimensionality reduction through principal component analysis to obtain the first feature; and based on the target radar track data, extracting the target's average velocity, velocity standard deviation, heading deflection standard deviation, maneuvering factor, and oscillation frequency in the motion dimension to obtain the second feature.

[0006] In one possible implementation, the amplitude is obtained by dividing the original amplitude of each data point by the distance attenuation factor. Satisfy the following formula:

[0007] Where amp is the target original amplitude; Standardized distance; Distance to target; is the attenuation coefficient; m is the influence factor of short and long pulses, when When the long pulse coverage range is reached, m is 0. When the short pulse coverage area is... .

[0008] In one possible implementation, the average speed Satisfy the following formula:

[0009] in, This represents the target radar track data in the i-th frame. Let be the speed of the target radar track in the i-th frame; In conjunction with the first aspect mentioned above, in one possible implementation, the speed standard deviation... Satisfy the following formula:

[0010] in, This represents the target radar track data in the i-th frame. Let be the speed of the target radar track in the i-th frame. Average speed; In one possible implementation, the standard deviation of the heading deflection Satisfy the following formula:

[0011]

[0012] in, This represents the target radar track data in the i-th frame. The standard deviation threshold for heading deflection. Let be the target radar track heading deflection value in the i-th frame. The mean deviation of the target radar track heading. In one possible implementation, the maneuver factor satisfies the following formula :

[0013] in, For average speed, The standard deviation of heading deflection; In one possible implementation, the oscillation frequency Satisfy the following formula :

[0014] in, This indicates whether the radar trajectory of the target UAV in the i-th frame meets the oscillation mode condition, where N is the number of sampling points.

[0015] In conjunction with the first aspect mentioned above, in one possible implementation, the first feature, the second feature, the spectrum listening correlation factor, and the regional position information feature are fused to obtain the target feature, and the target type is identified. This includes: concatenating the first feature, the second feature, the spectrum listening correlation factor, and the regional position information feature to obtain the target feature; inputting the target feature into a trained machine learning model to identify the target type; the machine learning model is obtained through supervised training on a dataset with target type labels constructed from historical data; the target categories include: drones, birds, airplanes, vehicles, flying objects, and others.

[0016] In conjunction with the first aspect mentioned above, in one possible implementation, the spectrum listening factor is calculated by associating the distance and azimuth in the time synchronization of listening data and radar data; In one possible implementation, for the protocol objective, the spectrum sensing correlation factor Satisfy the following formula:

[0017]

[0018] in, To obtain the target's location coordinates from the listening data. These are the target's position coordinates in the radar data. Preset distance threshold; In one possible implementation, for non-protocol targets, the spectrum sensing correlation factor Satisfy the following formula:

[0019]

[0020] in, The direction angle of the target in the radar data. To detect the direction and angle of the target in the listening data, Preset direction and angle thresholds.

[0021] In conjunction with the first aspect mentioned above, in one possible implementation, the target threat level is identified based on the target type and target radar track data, including: determining the target threat score based on the heading, speed, and distance in the target radar track data; and matching the target threat score with the preset threat level of the target type to determine the target threat level.

[0022] In one possible implementation, target threat score Satisfy the following formula:

[0023]

[0024]

[0025]

[0026] in, Weighting of the heading threat score. Weighting of speed threat score, The distance threat score is weighted, and the sum of the three weights is 1. Scoring for heading threat, To score for speed threat, This is the distance threat score, with a value between 0 and 100. For the standard heading value, For speed standard value, This is the standard distance value; For heading value, This is the speed value. This is the distance value.

[0027] Secondly, a target recognition system based on multi-source information fusion is provided, including: a data acquisition module and electronic equipment; The system includes a data acquisition module for acquiring target radar track data, listening device track data, and site area information data; and an electronic device for extracting a first feature in the RCS morphology dimension and a second feature in the motion dimension based on the target radar track data; determining spectrum listening correlation factors and site information features based on the target radar track data and listening device track data; fusing the first feature, second feature, spectrum listening correlation factors, and site information features to obtain target features and identify target type; and identifying target threat level based on target type and target radar track data.

[0028] In conjunction with the second aspect above, in one possible implementation, the electronic device is further configured to: based on the target radar track data, extract the altitude, velocity, amplitude, azimuth width, azimuth thickness, and the number of times the target passes through preset event points from N frames of the target radar track data in the RCS morphology dimension, and construct a feature set; based on the feature set, perform feature dimensionality reduction through principal component analysis to obtain the first feature; and based on the target radar track data, extract the target's average velocity, velocity standard deviation, heading deflection standard deviation, maneuvering factor, and oscillation frequency in the motion dimension to obtain the second feature.

[0029] This application provides a target recognition method and system based on multi-source information fusion. By fusing radar and reconnaissance data, utilizing the precise positioning of radar and the signal characteristics of reconnaissance, combined with the feature information of the system's coverage area, more accurate target recognition and classification are achieved. This enables the identification of UAV targets in complex low-altitude environments while improving the system's robustness. By extracting representative features and fusing them, data representations from different perspectives are obtained, increasing feature diversity and improving the model's generalization ability. PCA dimensionality reduction is used to extract the main features of the data, removing redundancy and noise, improving data processing efficiency and model performance, reducing computational costs, and avoiding multicollinearity problems. By fusing multi-source target information features, target trajectory complementarity is achieved, increasing the target detection update rate and robustness. To a certain extent, redundancy complementarity between radar and reconnaissance equipment is realized, enriching the target information of the situational awareness terminal. Furthermore, the threat level of blacklisted targets to core areas is calculated, providing decision support for jamming and countermeasure systems. This method has stronger practicality and solves the technical problem that existing technologies cannot accurately identify the threat level of targets in complex environments.

[0030] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0031] Figure 1 A system architecture diagram of a target recognition system based on multi-source information fusion is provided for embodiments of this application; Figure 2 A flowchart illustrating a target recognition method based on multi-source information fusion provided in this application embodiment; Figure 3 A flowchart illustrating another target recognition method based on multi-source information fusion provided in this application embodiment; Figure 4 A flowchart illustrating another target recognition method based on multi-source information fusion provided in this application embodiment; Figure 5 A schematic diagram illustrating the recognition result of another target recognition method based on multi-source information fusion provided in this application embodiment; Figure 6 A flowchart illustrating another target recognition method based on multi-source information fusion provided in this application embodiment; Figure 7 This is a schematic diagram illustrating the forensic investigation of a high-threat score target, as provided in an embodiment of this application. Detailed Implementation

[0032] In the description of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. The "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Furthermore, "at least one" means one or more, and "multiple" means two or more. The terms "first," "second," etc., do not limit the quantity or order of execution, and "first," "second," etc., do not necessarily imply differences.

[0033] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0034] The target recognition method based on multi-source information fusion provided in this application embodiment can be applied to, for example... Figure 1 In the target recognition system based on multi-source information fusion shown, such as Figure 1 As shown, the system includes: a data acquisition module 101 and an electronic device 102; Among them, the data acquisition module 101 is used to acquire target radar track data, listening equipment track data and field area information data; Electronic device 102 is used to extract a first feature in the RCS morphology dimension and a second feature in the motion dimension based on target radar track data; determine spectrum listening correlation factors and regional site information features based on target radar track data and listening device track data; fuse the first feature, the second feature, the spectrum listening correlation factors, and the regional site information features to obtain target features and identify target type; and identify target threat level based on target type and target radar track data.

[0035] To address the technical problem that existing technologies cannot accurately identify the threat level of targets in complex environments, embodiments of this application provide a target identification method based on multi-source information fusion. This method includes: acquiring target radar track data, listening device track data, and site area information data; extracting a first feature in the RCS morphology dimension and a second feature in the motion dimension based on the target radar track data; determining a spectrum listening correlation factor and regional site information features based on the target radar track data and listening device track data; fusing the first feature, second feature, spectrum listening correlation factor, and regional site information features to obtain target features and identify the target type; and identifying the target threat level based on the target type and target radar track data.

[0036] Figure 2 This is a flowchart illustrating the target recognition method based on multi-source information fusion provided in the embodiments of this application, as shown below. Figure 2 As shown, the method includes: S201: Acquire target radar track data, listening equipment track data, and site area information data.

[0037] Among them, target radar track data refers to the set of measurement information returned by the radar to the same target at continuous sampling times, which usually includes raw observations such as timestamp, distance, azimuth, altitude, radial velocity, and echo amplitude; listening equipment track data refers to the observation records of target signals recorded by spectrum listening or wireless direction finding devices, which includes signal arrival direction, signal strength, protocol or signaling characteristics, and time information; site area information data refers to the static or semi-static information of the deployment environment, such as site coordinates, terrain undulations, obstacle distribution, common obstruction areas, antenna location information, and metadata such as deployment parameters of radar and listening equipment.

[0038] In one possible implementation, target trajectory data output by radar and radio spectrum listening equipment is acquired; the radar trajectory data includes information such as the target's GPS, azimuth, distance, altitude, speed, heading, amplitude, and the thickness and width of the target's markings; the listening equipment trajectory data is divided into two forms: for standard UAVs, it includes information such as the target's frequency, GPS, azimuth, distance, altitude, and speed, while for non-standard UAVs, it only includes information such as the target's frequency and azimuth.

[0039] It should be noted that spatial and temporal synchronization is performed on the received target location information. This application only considers spatial synchronization between the radar and the spectrum monitoring equipment in a two-dimensional planar coordinate system, specifically between polar coordinates, with the monitoring equipment's coordinate system as the reference. Since the data sampling rates of the radar and the monitoring equipment are mismatched, temporal synchronization and target association are achieved through interpolation or truncation. The aligned radar and spectrum monitoring data are combined to form comprehensive target information, and the data is normalized and scaled to the same range for subsequent fusion and analysis.

[0040] As an example, in this embodiment, the radar covers an azimuth range of 0° to 360°, has a scanning period of 4 seconds / frame, a range of 15km, a range length of 7.5m, a range detection accuracy of ≤10m, and an azimuth detection accuracy of ≤0.6°. The radio spectrum listening device has a detection direction of 0° to 360°, a detection range of ≥5km, a detection frequency range of 300MHz to 6000MHz, and a direction finding accuracy (under line-of-sight conditions) of ≤3°. By default, both devices use true north as 0° and the listening coordinate system as the reference. Spatial synchronization, point P in the radar equipment polar coordinates Transformed into polar coordinates with the listening center Satisfy the following formula:

[0041]

[0042]

[0043]

[0044] in, Let P be the coordinates of point P in the listening coordinate system. Let P be the coordinates of point P in the radar coordinate system.

[0045] Time synchronization employs soft synchronization, determining the time deviation by calculating the timestamp difference between two sensor frames. This time deviation is then combined with the predicted position of the target's motion trajectory and the time deviation to calculate the position corresponding to the interpolated time, thus achieving the association and synchronization of multi-source data. This assumes the radar has an existing track list. Among them, the flight path The latest dot update timestamp is The coordinates are speed is The heading is When receiving the timestamp from the listening device Target data At that time, the flight path was calculated. exist The coordinates of the time are ( Specifically: Based on flight path speed ,course Estimated time difference The path polar coordinate vector is:

[0046] Then corresponding Time, Track The coordinates of the synchronization point are:

[0047] List of sequential flight paths Coordinates of all synchronized points in the system:

[0048] This step provides high-quality, structured, and traceable multi-source observation data for subsequent feature extraction and fusion; standardized data acquisition and preprocessing can reduce subsequent feature bias and improve the reliability of multi-source data association, thereby providing a solid data foundation for target identification and hazard level assessment.

[0049] S202. Based on the target radar track data, extract the first feature in the RCS morphology dimension and the second feature in the motion dimension.

[0050] Among them, the first feature of the RCS morphology dimension refers to the statistical or morphological description reflecting the target echo shape and amplitude distribution, while the second feature of the motion dimension refers to the quantitative indicators reflecting the temporal motion characteristics of the target, such as speed, acceleration, heading changes and maneuverability.

[0051] In one possible implementation, based on target radar track data, in the RCS morphology dimension, the altitude, velocity, amplitude, azimuth width, azimuth thickness, and the number of times the target passes through preset event points of N frames of target radar track data are extracted to construct a feature set. Based on the feature set, feature dimensionality reduction is performed through principal component analysis to obtain the first feature. Based on the target radar track data, in the motion dimension, the target's average velocity, velocity standard deviation, heading deflection standard deviation, maneuvering factor, and oscillation frequency are extracted to obtain the second feature.

[0052] It should be noted that the extraction of morphological and motion features should take into account the effects of radar observation geometry, ranging accuracy, and sampling rate. When encountering short frame sequences or occluded frames, interpolation or confidence labeling should be used to avoid misleading subsequent judgments. The units, dimensions, and normalization methods of features must be consistent to facilitate subsequent stitching and use by machine learning models.

[0053] Based on the above steps, this step converts the original radar observation data into structured features that can reflect the differences in the target's shape and behavior. The first and second features, after normalization and dimensionality reduction, help reduce the influence of noise in the original data and improve feature discrimination, thereby providing effective input for subsequent multi-source fusion and target classification.

[0054] S203. Based on target radar track data and listening equipment track data, determine spectrum listening correlation factors and regional site information characteristics.

[0055] Among them, the spectrum listening correlation factor refers to the index used to quantify the consistency or correlation between the listening equipment observation and the radar track in terms of time, distance or direction; the regional site information characteristics refer to the environmental quantitative description derived from the site environment that affects observation and identification.

[0056] In one possible implementation, a spectrum listening correlation factor is calculated by correlating the distance and azimuth during the time synchronization of the listening data and radar data in the data preprocessing. This spectrum listening correlation factor reflects the attribute that the target is an unmanned aerial vehicle (UAV). The radar coverage area is then defined according to a map, and regional information features include roads and others. Regional information features are obtained by calculating the road features of the target area sequence; these regional information features reflect the attribute that the target is a ground target.

[0057] As an example, in an embodiment of this application, for the protocol objective, the spectrum sensing correlation factor... Satisfy the following formula:

[0058]

[0059] in, To obtain the target's location coordinates from the listening data. These are the target's position coordinates in the radar data. Preset distance threshold; As an example, in an embodiment of this application, for non-protocol targets, the spectrum sensing correlation factor Satisfy the following formula:

[0060]

[0061] in, The direction angle of the target in the radar data. To detect the direction and angle of the target in the listening data, Preset direction and angle thresholds.

[0062] As an example, in an embodiment of this application, the regional information feature Satisfy the following formula:

[0063]

[0064] in, This represents the target radar track data in the i-th frame, where M is the sampling point.

[0065] Based on the above steps, this step constructs a bridging quantitative index that connects reconnaissance observation, radar observation, and environmental constraints. The correlation factor and regional information features can enhance the comparability and interpretability between different source data, provide contextual constraints for multi-source feature fusion, thereby reducing identification errors caused by erroneous correlations and improving overall identification robustness.

[0066] S204. The first feature, the second feature, the spectrum listening correlation factor, and the regional position information feature are fused to obtain the target feature and identify the target type.

[0067] In one possible implementation, the first feature, the second feature, the spectrum listening correlation factor, and the regional site information feature are concatenated to obtain the target feature. The target feature is then input into a trained machine learning model to identify the target type.

[0068] It should be noted that this application considers the correlation and redundancy between features during fusion, and uses feature selection and dimensionality reduction methods to reduce dimensionality before splicing; for training the input model, historical multi-source datasets with class labels are used and imbalanced samples are processed to prevent a certain target from being weakened due to the scarcity of samples.

[0069] Based on the above steps, this step realizes the transformation from multi-source heterogeneous raw data to unified, high-dimensional discriminative features, and identifies target types through the trained discriminative model. This fusion method improves the reliability and stability of category discrimination by integrating various complementary information, reduces misjudgments caused by single feature dominance, and provides more reliable category input for subsequent hazard level assessment.

[0070] S205: Identify target threat levels based on target type and target radar track data.

[0071] Among them, the target threat level refers to the classification result of the potential danger of a target under a given discrimination framework.

[0072] In one possible implementation, a target threat score is determined based on the target's heading, speed, and distance from radar track data. This target threat score is then matched against a preset threat level for the target type to determine the target threat level.

[0073] It should be noted that threat level determination should take into account the confidence level of target identification and the influence of the field environment, and the weights and thresholds can be adjusted according to the actual application scenario to adapt to different task requirements and risk preferences.

[0074] Based on the above steps, this step combines the identification results with the real-time situation to form an actionable threat level output. This output helps the command and control system allocate resources according to priority, trigger corresponding protection or response measures, and reduce erroneous responses caused by the separation of target classification and situation information, thereby improving overall protection efficiency and security decision-making quality.

[0075] Based on the above steps, this step combines the identification results with the real-time situation to form a threat level. This helps the command and control system allocate resources according to priority, trigger corresponding protection or response measures, and reduce erroneous responses caused by the separation of target classification and situational information, thereby improving overall protection efficiency and the quality of security decisions.

[0076] This application, through the technical solutions described in S201-S205, enables target information to be synergistically supplemented in spatial, motion, signal, and regional environmental dimensions. This effectively overcomes problems such as incomplete information from a single sensor, insufficient feature representation capabilities, and unstable recognition results due to scene changes, allowing target type recognition to maintain higher accuracy and robustness in complex airspace environments. Furthermore, based on target type recognition, it further combines target trajectory behavior and real-time situational awareness to determine threat levels, further clarifying whether the target is dangerous. This improves the monitoring system's ability to detect potential threats early, its risk classification capabilities, and its response efficiency, solving the technical problem that existing technologies cannot accurately identify the threat level of targets in complex environments.

[0077] In one possible implementation of the embodiments of this application, combined with Figure 2 ,like Figure 3 As shown, the above S202 can be specifically implemented through the following S301, S302 and S303, which are explained in detail below: S301. Based on the target radar track data, extract the altitude, speed, amplitude, azimuth width, azimuth thickness, and the number of times the target passes through preset event points from N frames of the target radar track data in the RCS morphology dimension, and construct a feature set.

[0078] In one possible implementation, observations are read frame by frame from N consecutive radar tracks. The altitude sequence, velocity sequence, and amplitude sequence are calculated according to the frame sequence. The azimuth width and azimuth thickness are determined based on echo lateral and longitudinal edge detection or threshold segmentation. At the same time, the number of intersections between the track and preset event points is counted to generate a crossing count. Preprocessing is performed on the raw amplitude, including range attenuation correction and pulse type normalization. Noise or abnormal observations are filtered by in-window median filtering or threshold removal to ensure feature robustness.

[0079] As an example, in an embodiment of this application, the amplitude is obtained by dividing the original amplitude of each data point by the distance attenuation coefficient. Satisfy the following formula:

[0080] Where amp is the target original amplitude; Standardized distance; Distance to target; is the attenuation coefficient; m is the influence factor of short and long pulses, when When the long pulse coverage range is reached, m is 0. When the short pulse coverage area is... .

[0081] Based on the above steps, this step transforms the original temporal observations into a structured candidate feature set that reflects the morphology and observational distribution characteristics of the target, providing a more descriptive input for subsequent classification, thereby improving morphological discrimination ability and reducing the impact of noise interference on the recognition results.

[0082] S302. Based on the feature set, perform feature dimensionality reduction through principal component analysis to obtain the first feature.

[0083] Principal component analysis refers to mapping the original multidimensional features into several mutually orthogonal new features through linear transformation. These new features are sorted in descending order of explained variance to achieve dimensionality reduction and redundancy removal.

[0084] In one possible implementation, the constructed candidate feature set is first subjected to mean centering and variance normalization, then the covariance matrix is ​​calculated to obtain the eigenvectors and eigenvalues, and several principal components whose cumulative explained variance reaches a predetermined proportion are selected as the first eigenvector output.

[0085] It should be noted that PCA is a linear dimensionality reduction method, suitable for situations where features are linearly correlated or approximately linearly correlated. For features that are highly nonlinear or have a strong skewed distribution, a nonlinear transformation can be applied before PCA or alternative dimensionality reduction methods can be considered.

[0086] As an example, in this embodiment of the application, during the training phase, a covariance matrix is ​​constructed based on labeled historical track samples and principal components are calculated. The first four principal components are selected so that the cumulative explained variance is ≥85% as the first feature representation. During online operation, the candidate features in the window are projected onto the principal component basis and the standardized principal component scores are output. At the same time, the projection residuals are recorded to assess the risk of feature extrapolation.

[0087] Based on the above steps, this step compresses high-dimensional, redundant, or noise-sensitive morphological features into first features with strong expressive power and low dimensionality through dimensionality reduction. This not only reduces the computational complexity and overfitting risk of subsequent models, but also improves the stability and interpretability of the features, making it easier to achieve more robust class discrimination under limited training sample conditions.

[0088] S303. Based on the target radar track data, extract the target's average speed, speed standard deviation, heading deflection standard deviation, maneuvering factor, and oscillation frequency in the motion dimension to obtain the second feature.

[0089] In one possible implementation, the system performs denoising and frame interpolation on the continuous frame velocity and heading sequences, then calculates the average velocity and velocity standard deviation within the window, calculates the deflection standard deviation for the heading difference sequence, and calculates the motion factor according to a predetermined formula or weight combination. At the same time, it performs frequency domain transformation or autocorrelation analysis on the velocity or heading sequences to estimate the oscillation frequency, and outputs the above-mentioned quantitative indicators as a second feature vector to describe the motion behavior characteristics of the target.

[0090] As an example, in an embodiment of this application, the average speed Satisfy the following formula:

[0091] in, This represents the target radar track data in the i-th frame. Let be the speed of the target radar track in the i-th frame; As an example, in an embodiment of this application, the speed standard deviation Satisfy the following formula:

[0092] in, This represents the target radar track data in the i-th frame. Let be the speed of the target radar track in the i-th frame. Average speed; As an example, in an embodiment of this application, the standard deviation of heading deflection... Satisfy the following formula:

[0093]

[0094] in, This represents the target radar track data in the i-th frame. The standard deviation threshold for heading deflection. Let be the target radar track heading deflection value in the i-th frame. The mean deviation of the target radar track heading. As an example, in an embodiment of this application, the maneuver factor satisfies the following formula :

[0095] in, For average speed, The standard deviation of heading deflection; As an example, in an embodiment of this application, the oscillation frequency... Satisfy the following formula :

[0096] in, This indicates whether the radar trajectory of the target UAV in the i-th frame meets the oscillation mode condition, where N is the number of sampling points.

[0097] It should be noted that whether the radar track of the target UAV in the i-th frame meets the oscillation mode condition is determined in the following way: First, by calculating the relative difference in heading between each adjacent sampling point in the track. Based on radar measurement errors, the track sampling points are calculated. Heading difference symbol:

[0098] in, The heading deflection error threshold can be set according to the actual radar angle measurement error. As an example, this application... Set to 2; Based on the set of heading angle sign values The sign transformation relationship defines two heading oscillation modes: and

[0099] and

[0100] When either of the two heading oscillation modes described above is met, the system is determined to be in an oscillation mode, and the number of oscillations is counted, denoted by . :

[0101] Once the number of heading oscillations is determined, the oscillation frequency can be calculated.

[0102] Based on the above steps, this step transforms the target's temporal trajectory into quantitative indicators that can characterize maneuverability, stability, and periodic behavior, which helps to distinguish targets with different flight strategies and behavioral patterns, thereby enhancing the ability to identify target behavior categories.

[0103] This application embodiment constructs a candidate feature set reflecting the target's shape distribution and trajectory spatial behavior by extracting morphological features such as height, speed, amplitude, azimuth width, azimuth thickness, and the number of times it passes through preset event points. Principal component analysis is then used to reduce the dimensionality of this feature set, compressing redundant information and noise to form a stable, expressive, and appropriately dimensional first feature. Furthermore, by extracting motion features such as average speed, speed standard deviation, heading deflection standard deviation, maneuvering factor, and oscillation frequency, the target's behavior patterns and maneuverability differences are further characterized. This provides a reliable basic input for multi-source feature fusion, target type identification, and threat level assessment, improving the accuracy, stability, and interpretability of target identification while reducing system computational complexity and noise sensitivity.

[0104] In one possible implementation of the embodiments of this application, combined with Figure 2 ,like Figure 4 As shown, the above S204 can be implemented through the following S401 and S402, which are explained in detail below: S401. The first feature, the second feature, the spectrum listening correlation factor, and the regional position information feature are concatenated to obtain the target feature.

[0105] In one possible implementation, the extracted first feature, second feature, spectrum listening correlation factor, and regional position information feature are concatenated to obtain the target feature.

[0106] As an example, in an embodiment of this application, the target feature Satisfy the following formula:

[0107] in, Let T represent the first feature and T represent the second feature. As a correlation factor, This refers to the characteristics of regional positioning information.

[0108] S402. Input the target features into the trained machine learning model to identify the target type.

[0109] In one possible implementation, target features are input into a trained machine learning model to identify target types. The machine learning model is trained in a supervised manner on a dataset with target type labels constructed from historical data, and includes, but is not limited to, SVM, RandomForest, and AdaBoost models. Identifiable target categories include: drones, birds, airplanes, vehicles, flying objects, and others.

[0110] As an example, in the embodiments of this application, the target type Satisfy the following formula:

[0111] in, 1 represents drone, 2 represents bird, 3 represents airplane, 4 represents flying object, and 5 represents other.

[0112] As an example, in an embodiment of this application, Figure 5 This is a schematic diagram of the recognition results provided in the embodiments of this application, such as... Figure 5 As shown, the target type was identified as a drone based on radar data and listening data.

[0113] This application integrates morphological features, motion features, listening correlation factors, and regional site information features within a unified framework. This allows target features to simultaneously reflect the target's physical characteristics, motion behavior, signal correlation, and environmental impact, thereby forming a high-dimensional, structured, and information-complete target representation. After inputting these target features into a trained machine learning model, the system can accurately identify target types, fully utilize the supervisory information from historical labeled data to improve the stability and reliability of identification, and distinguish multiple types of low-altitude targets in complex airspace environments.

[0114] In one possible implementation of the embodiments of this application, combined with Figure 2 ,like Figure 6 As shown, the above S205 can be implemented through the following S601 and S602, which are explained in detail below: S601. Determine the target threat score based on the heading, speed, and distance in the target radar track data.

[0115] In one possible implementation, the system calculates a heading threat score, a speed threat score, and a range threat score based on each frame of data from the target's radar track. Then, it weights and sums these three scores according to preset weights to obtain a comprehensive threat score, and forensic evidence is collected from targets with high threat scores. Figure 7 This is a schematic diagram illustrating the forensic investigation of targets with high threat scores, provided as an embodiment of this application.

[0116] As an example, in an embodiment of this application, the target threat score... Satisfy the following formula:

[0117]

[0118]

[0119]

[0120] in, Weighting of the heading threat score. Weighting of speed threat score, The distance threat score is weighted, and the sum of the three weights is 1. Scoring for heading threat, To score for speed threat, This is the distance threat score, with a value between 0 and 100. For the standard heading value, For speed standard value, This is the standard distance value; For heading value, This is the speed value. This is the distance value.

[0121] S602. Match the target threat score with the preset threat level of the target type to determine the target threat level.

[0122] In one possible implementation, the overall threat score is compared with a preset threat threshold table corresponding to the target type. If the threat score falls within a certain threshold range, the target is marked as the corresponding level. This application sets different thresholds for different types of targets to accommodate the varying potential security impacts of different targets such as drones, birds, aircraft, and vehicles in the same area.

[0123] As an example, in this embodiment of the application, for drone targets, a comprehensive threat score of 0-30 points corresponds to low threat, 31-60 points to medium threat, 61-85 points to high threat, and 86-100 points to emergency threat; the system calculates the score and matches the level for each target in real time so that the command system can allocate resources or issue warnings according to priority.

[0124] This invention achieves a complete conversion from quantitative threat scores to discrete threat levels by combining the target's motion status with the identified target type. Key indicators in target trajectory data are transformed into measurable threat scores, making the potential danger of targets quantifiable and comparable, while also providing robust handling of anomalies and missing data. The threat scores are matched with preset threat thresholds for target types to form clear threat levels, enabling the system to classify and manage different targets according to their degree of danger. This achieves quantifiable and hierarchical management of target threats, allowing the surveillance system to prioritize monitoring of key targets and respond to high-risk targets, improving situational awareness and security efficiency, while reducing the risk of misjudgment due to target type or motion anomalies, providing a reliable and operable basis for decision-making and protective measures.

[0125] The above primarily describes the solutions of the embodiments of this application from the perspective of device implementation. It is understood that each device, such as a target recognition system based on multi-source information fusion, includes at least one of the hardware structures and software modules corresponding to each function in order to achieve the above-mentioned functions. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is implemented by hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0126] This application embodiment can divide the target recognition system based on multi-source information fusion into functional units according to the above method example. For example, each function can be divided into separate functional units, or two or more functions can be integrated into one processing unit. The integrated unit can be implemented in hardware or as a software functional unit. It should be noted that the unit division in this application embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0127] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented using software programs, implementation can be, in whole or in part, in the form of a computer program product. This computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device containing one or more servers, data centers, etc., that can be integrated with the medium. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state disks (SSDs)).

[0128] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple instances. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce good results.

[0129] Although this application has been described in conjunction with specific features and embodiments, it is obvious that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and drawings are merely exemplary illustrations of this application as defined by the appended claims, and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from the spirit and scope of this application. Thus, if such modifications and modifications of this application fall within the scope of the claims of this application and their equivalents, this application is also intended to include such modifications and modifications.

Claims

1. A target recognition method based on multi-source information fusion, characterized in that, include: Acquire target radar track data, listening equipment track data, and information data about the site area; Based on the target radar track data, a first feature is extracted in the RCS morphology dimension and a second feature is extracted in the motion dimension. Based on the target radar track data and the listening device track data, determine the spectrum listening correlation factor and regional site information characteristics; The first feature, the second feature, the spectrum listening correlation factor, and the regional site information feature are fused to obtain the target feature and identify the target type. Based on the target type and the target radar track data, the target threat level is identified.

2. The target recognition method based on multi-source information fusion according to claim 1, characterized in that, The extraction of a first feature in the RCS morphology dimension and a second feature in the motion dimension based on the target radar track data includes: Based on the target radar track data, in the RCS morphology dimension, the altitude, velocity, amplitude, azimuth width, azimuth thickness, and the number of times the target passes through preset event points of N frames of the target radar track data are extracted to construct a feature set; Based on the feature set, feature dimensionality reduction is performed through principal component analysis to obtain the first feature; Based on the target radar track data, the target's average speed, speed standard deviation, heading deflection standard deviation, maneuvering factor, and oscillation frequency are extracted in the motion dimension to obtain the second feature.

3. The target recognition method based on multi-source information fusion according to claim 2, characterized in that, The amplitude is obtained by dividing the original amplitude of each data point by the distance attenuation coefficient. Satisfy the following formula: Where amp is the target original amplitude; Standardized distance; The target distance; is the attenuation coefficient; m is the influence factor of short and long pulses, when When the long pulse coverage range is reached, m is 0. When the short pulse coverage area is... .

4. The target recognition method based on multi-source information fusion according to claim 2, characterized in that, The average speed Satisfy the following formula: in, This represents the target radar track data in the i-th frame. Let be the speed of the target radar track in the i-th frame; The speed standard deviation Satisfy the following formula: in, This represents the target radar track data in the i-th frame. Let be the speed of the target radar track in the i-th frame. Average speed; The standard deviation of heading deflection Satisfy the following formula: in, This represents the target radar track data in the i-th frame. The standard deviation threshold for heading deflection. Let be the target radar track heading deflection value in the i-th frame. The mean deviation of the target radar track heading. The mobility factor satisfies the following formula : in, For average speed, The standard deviation of heading deflection; The oscillation frequency Satisfy the following formula : in, This indicates whether the radar trajectory of the target UAV in the i-th frame meets the oscillation mode condition, where N is the number of sampling points.

5. The target recognition method based on multi-source information fusion according to claim 1, characterized in that, The step of fusing the first feature, the second feature, the spectrum listening correlation factor, and the regional site information feature to obtain target features and identify target types includes: The first feature, the second feature, the spectrum listening correlation factor, and the regional site information feature are concatenated to obtain the target feature; The target features are input into a trained machine learning model to identify the target type; the machine learning model is obtained through supervised training on a dataset with target type labels constructed from historical data; the target categories include: drones, birds, airplanes, vehicles, flying objects, and others.

6. The target recognition method based on multi-source information fusion according to claim 1, characterized in that, The spectrum listening factor is calculated by associating distance and azimuth in the time synchronization of listening data and radar data. For the protocol objective, the spectrum sensing correlation factor Satisfy the following formula: in, To obtain the target's location coordinates from the listening data. These are the target's position coordinates in the radar data. Preset distance threshold; For non-protocol targets, the spectrum sensing correlation factor Satisfy the following formula: in, The direction angle of the target in the radar data. To detect the direction and angle of the target in the listening data, Preset direction and angle thresholds.

7. The target recognition method based on multi-source information fusion according to claim 1, characterized in that, The step of identifying the target threat level based on the target type and the target radar track data includes: Based on the heading, speed, and distance in the target radar track data, the target threat score is determined; The target threat score is matched with the preset threat level of the target type to determine the target threat level.

8. The target recognition method based on multi-source information fusion according to claim 7, characterized in that, The target threat score Satisfy the following formula: in, Weighting of the heading threat score. Weighting of speed threat score, The distance threat score is weighted, and the sum of the three weights is 1. Scoring for heading threat, To score for speed threat, This is the distance threat score, with a value between 0 and 100. For the standard heading value, For speed standard value, This is the standard distance value; For heading value, This is the speed value. This is the distance value.

9. A target recognition system based on multi-source information fusion, used to implement the method described in any one of claims 1-8, characterized in that, The target recognition system based on multi-source information fusion includes: a data acquisition module and electronic equipment; The data acquisition module is used to acquire target radar track data, listening device track data, and site area information data; The electronic device is configured to extract a first feature in the RCS morphology dimension and a second feature in the motion dimension based on the target radar track data; determine a spectrum listening correlation factor and regional site information features based on the target radar track data and the listening device track data; fuse the first feature, the second feature, the spectrum listening correlation factor, and the regional site information features to obtain target features and identify the target type; and identify the target threat level based on the target type and the target radar track data.

10. The target recognition system based on multi-source information fusion according to claim 9, characterized in that, The electronic device is also used for: Based on the target radar track data, in the RCS morphology dimension, the altitude, velocity, amplitude, azimuth width, azimuth thickness, and the number of times the target passes through preset event points of N frames of the target radar track data are extracted to construct a feature set; Based on the feature set, feature dimensionality reduction is performed through principal component analysis to obtain the first feature; Based on the target radar track data, the target's average speed, speed standard deviation, heading deflection standard deviation, maneuvering factor, and oscillation frequency are extracted in the motion dimension to obtain the second feature.

Images