Adaptive correlation method for situational information of non-homogeneous targets, electronic devices and media

By constructing a super-three-dimensional target feature vector space and automatically determining clustering parameters, the problem of adaptive association of non-homogeneous sensor data was solved, and efficient and accurate target situation information association was achieved.

CN118152991BActive Publication Date: 2026-11-17BEIJING INST OF TECH
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202410333804.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-03-22
Publication Date
2026-11-17
Estimated Expiration
2044-03-22

AI Technical Summary

Technical Problem

Existing technologies rely on manual parameter settings when processing data from non-homogeneous sensors, resulting in a large amount of computation and difficulty in adaptively associating and distinguishing targets, leading to high computational resource consumption and low accuracy.

Method used

By constructing a super three-dimensional target feature vector space, automatically determining the clustering parameter set, using the DBSCAN algorithm for clustering, and combining the cluster correlation judgment of adjacent time points, adaptive association of target situation information is achieved.

Benefits of technology

It achieves automated parameter adjustment without human intervention, reduces computational load, and improves the efficiency and accuracy of target situation information association.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118152991B_ABST
    Figure CN118152991B_ABST
Patent Text Reader

Abstract

This invention discloses an adaptive association method, electronic device, and medium for non-homogeneous target situational information. The invention includes acquiring ultra-3D target feature vector data, determining a clustering parameter set, clustering using the parameter set and selecting the optimal clustering result data, determining cluster clusters, judging the correlation of cluster clusters at adjacent sampling times, and adaptively associating target situational information. This invention fully considers target characteristics in cluster analysis, improving the efficiency and accuracy of multi-source, multi-target detection data screening and processing in situational analysis. This invention can automatically adjust parameter clustering, determining the optimal clustering result based on data characteristics without manual parameter adjustment, avoiding interference from subjective factors in the association results, and achieving automation of target situational information association. This invention reduces the computational load of target situational information association by first locking the optimal clustering result and then comparing and calculating, thus improving the efficiency of target situational information association.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of information processing technology. Specifically, it relates to an adaptive association method for situational information of non-same-source targets, an electronic device, and a computer-readable storage medium. Background Technology

[0002] With the rapid development of modern information technology and the increasing informatization of various fields, control systems are gradually becoming more digital, generating massive, rich, and diverse information. When detecting target situations, multiple dimensions such as land and air are involved, resulting in numerous targets, large and complex data volumes, and interference from electromagnetic and network factors. Long-range target detection is hampered by the distance and weak signal characteristics, making it difficult for a single detection device to independently monitor targets. Therefore, collaborative detection and identification of non-homogeneous sensors is an inevitable trend. However, in collaborative sensor detection, the scarcity and inconsistency of target information detected by each sensor make target correlation between sensors a primary and challenging issue.

[0003] Existing methods, when processing large amounts of non-similar data, rely heavily on human judgment for parameter settings, resulting in massive computational demands, requiring significant memory resources, and failing to adaptively associate the same target or differentiate between different targets. Summary of the Invention

[0004] In view of this, the purpose of this invention is to correctly integrate non-similar information and reasonably associate targets, thereby improving the accuracy of unified situational awareness.

[0005] To achieve the above objectives, in a first aspect, the present invention provides an adaptive association method for situational information of non-homogeneous targets, comprising:

[0006] The steps for acquiring ultra-three-dimensional target feature vector data include: acquiring target feature data detected by each sensor in the same physical space, constructing an ultra-three-dimensional target feature vector space, and calculating standardized ultra-three-dimensional target feature vector data.

[0007] The steps for determining the clustering parameter set include: for each sampling time, using the standardized hyper-3D target feature vector data corresponding to that sampling time to determine the clustering parameter set for that sampling time;

[0008] The steps include using a clustering parameter set to cluster and selecting the optimal clustering result data; in this step, for each sampling time, each clustering parameter in the clustering parameter set for that sampling time is used to cluster the standardized ultra-3D target feature vector data corresponding to that sampling time, to obtain the candidate clustering result dataset for that sampling time, and then selecting the optimal clustering result data from the candidate clustering result dataset;

[0009] The steps for determining clusters include: for each sampling time, determining the cluster at that sampling time based on the optimal clustering result data at that sampling time;

[0010] The steps for determining the correlation between clusters at adjacent sampling times: In this step, the correlation between clusters at adjacent sampling times is determined. If they are correlated, the correlated clusters at adjacent sampling times belong to the same target.

[0011] The step of adaptive association of target situation information: In this step, for all sampling times in the ultra-3D target feature vector dataset, the clusters are labeled from the initial sampling time, and all ultra-3D target feature vector data are associated with the target to which they belong.

[0012] In a second aspect, the present invention provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the adaptive association method for non-homogeneous target situational information.

[0013] Thirdly, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the adaptive association method for non-homogeneous target situational information.

[0014] The present invention has the following beneficial effects:

[0015] This invention can automatically adjust clustering parameters and determine the optimal clustering result based on data characteristics, eliminating the need for manual parameter adjustment and avoiding interference from subjective factors in the association results, thus automating the association of target situation information. By first locking the optimal clustering result and then performing comparative calculations, this invention reduces the computational workload of target situation information association and improves its efficiency. Attached Figure Description

[0016] Figure 1 This is a flowchart of the adaptive association method for non-homogeneous target situation information of the present invention.

[0017] Figure 2 This is a flowchart of one embodiment of the method of the present invention.

[0018] Figure 3 This is a flowchart of a sub-step in one embodiment of the method of the present invention.

[0019] Figure 4 This is a schematic diagram of the electronic device in one embodiment of the present invention. Detailed Implementation

[0020] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0021] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0022] To accurately integrate non-homogeneous target information and reasonably associate targets, thereby improving the accuracy of unified situational awareness, this invention provides an adaptive association method for non-homogeneous target situational information. This method first collects non-homogeneous target data from different sensors at various sampling times, constructing a super-three-dimensional target feature vector. Based on the data dimensionality features, the original data is standardized to obtain data suitable for direct application of clustering algorithms. Then, the standardized non-homogeneous target data is adaptively clustered to obtain clusters. Finally, based on the correlation judgment of clusters with adjacent timestamps, adaptive target association is performed on the clusters starting from the initial sampling time. Figure 1 As shown, it specifically includes:

[0023] Step S100: Acquiring ultra-three-dimensional target feature vector data; In this step, target feature data detected by each sensor in the same physical space are acquired, an ultra-three-dimensional target feature vector space is constructed, and standardized ultra-three-dimensional target feature vector data is calculated.

[0024] Step S200: Determine the clustering parameter set. In this step, for each sampling time, the clustering parameter set for that sampling time is determined using the standardized ultra-three-dimensional target feature vector data corresponding to that sampling time.

[0025] Step S300: Clustering and selecting the optimal clustering result data using a clustering parameter set; In this step, for each sampling time, each clustering parameter in the clustering parameter set for that sampling time is used to cluster the standardized ultra-three-dimensional target feature vector data corresponding to that sampling time, to obtain the candidate clustering result dataset for that sampling time, and the optimal clustering result data is selected from the candidate clustering result dataset.

[0026] Step S400: Determine the clustering cluster. In this step, for each sampling time, determine the clustering cluster for that sampling time based on the optimal clustering result data for that sampling time.

[0027] Step S500: The cluster correlation judgment between adjacent sampling times is performed. If the clusters between adjacent sampling times are correlated, the correlated clusters between adjacent sampling times belong to the same target.

[0028] Step S600: Adaptive association of target situation information; In this step, for all sampling times in the ultra-three-dimensional target feature vector dataset, the clusters are labeled from the initial sampling time, and all ultra-three-dimensional target feature vector data are associated with the target to which they belong.

[0029] Example 1

[0030] This embodiment is used to describe in detail the adaptive association method for non-homogeneous target situational information of the present invention, such as... Figure 2 As shown, in one embodiment of the present invention, an adaptive association method for non-homogeneous target situational information includes:

[0031] In the step of acquiring ultra-three-dimensional target feature vector data, target feature data detected by each sensor in the same physical space are acquired, with a sensor sampling time interval of T. Ultra-three-dimensional target feature vector space is constructed and standardized ultra-three-dimensional target feature vector data is calculated.

[0032] The target feature data includes at least velocity, three-dimensional position, reflectivity, and corresponding timestamps. The three-dimensional position can be sensor coordinates, target orientation, and target tilt angle. Furthermore, any data that can be used for target similarity calculation can be considered target feature data.

[0033] The basis for constructing the hyper-3D target feature vector space is the target feature data, and the hyper-3D target feature vector data X in the hyper-3D target feature vector space. it X is composed of the i-th set of target feature data acquired by one of multiple sensors at sampling time t. it =[v it ,x it ,y it ,z it ,rcs it ,…,t],v it Let x represent the velocity component in the i-th group of target feature data. it ,y it ,z it Let rcs represent the three coordinate components of the three-dimensional position in the i-th set of target feature data, such as longitude, latitude, and altitude. it The reflectivity component in the i-th group of target feature data, where t is the timestamp, representing the sampling time t.

[0034] Different features often have different dimensions and units, which can affect the results of data clustering analysis. To eliminate the influence of dimensions between indicators, it is necessary to standardize the feature vectors at each time point. The ultra-three-dimensional target feature vector data X at sampling time t is shown below. it There are n sets of ultra-3D target feature vector data X. it All components except the timestamp are standardized to obtain standardized data of the ultra-3D target feature vector. in,

[0035]

[0036]

[0037]

[0038]

[0039]

[0040] ………

[0041] In the step of determining the clustering parameter set, for each sampling time, the clustering parameter set for that sampling time is determined using the standardized hyper-3D target feature vector data corresponding to that sampling time. The existing DBSCAN requires manual parameter setting, and the selection of parameters greatly affects the final clustering result. To eliminate human intervention in DBSCAN clustering, it is necessary to automatically determine the clustering parameters for sampling time t based on the dataset: the neighborhood radius Eps of a point, and the minimum number of points in a cluster MinPts. Considering that Eps and MinPts exist in a fit, once a suitable MinPts for the dataset is determined, a better clustering result can be obtained by combining Eps that fits the MinPts. Therefore, each suitable MinPts for the dataset can correspond to a better clustering result. This invention is based on observations of the average neighborhood curve obtained from hyper-3D target feature vector data. When the k-Dist distance is less than the suitable neighborhood radius Eps, the average number of neighborhood points increases gradually with the increase of the k-Dist distance. However, as the k-Dist distance gradually approaches the neighborhood radius Eps, the average number of neighborhood points experiences a significant spike. As the k-Dist distance continues to increase, data points from each target are assigned to reasonable clusters, and the dataset gradually converges under these parameters, resulting in a period of flattening for the average neighborhood point curve. When the k-Dist distance continues to increase beyond a certain range of the neighborhood radius Eps, data points from different targets are assigned to the same cluster, causing a second spike in the average neighborhood point count. Therefore, the average number of neighborhood points corresponding to the neighborhood radius Eps that best matches the current MinPts lies within the region of these two spikes. By optimizing the clustering results within this region, the optimal clustering result and the suitable neighborhood radius Eps can be determined.

[0042] like Figure 3 As shown, this step specifically includes:

[0043] Step S201: Set the minimum number of cluster points MinPts = 2Dim according to the dimension of the standardized ultra-3D target feature vector data participating in clustering. Here, timestamp data does not participate in clustering, and Dim represents the number of dimensions of the remaining components of the standardized ultra-3D target feature vector data excluding timestamps.

[0044] Step S202: For each timestamp t, calculate the distance from all data points at that sampling time to other data points at the same time. In this embodiment, specifically, two data points... and Distance between Using Euclidean distance:

[0045]

[0046] Step S203: Let k = MinPts-1, standardize the ultra-3D target feature vector data. is a data point in the hyper-3D target feature vector space, and each data point the distance to its k-th nearest neighbor is taken as the k-Dist distance D of the data point it , and the k-Dist distances of all data points constitute a k-Dist distance matrix D matix =[D 1t ,D 2t ,…D nt , traverse the matrix D matix , and calculate the number of data points k in the neighborhood of each k-Dist distance D of each data point at the sampling time t tt ijt , to form a neighborhood point count matrix K t ,

[0047]

[0048] the average number of neighborhood data points corresponding to the k-Dist distance of the j-th data point is denoted as d jt , and the mean value d of the number of neighborhood points corresponding to each k-Dist distance D is calculated jt jt ,

[0049] in step S204, all the mean values d of the number of neighborhood points corresponding to the k-Dist distances D jt jt are sorted in ascending order to obtain a sorted mean sequence, the least squares polynomial fitting method is used to perform curve fitting on the sorted mean sequence to obtain a fitted curve f(x), 1≤x≤n, and obtain the inflection points (x1,f(x1)) and (x2,f(x2)) at two sharply rising portions of the fitted curve f(x), 1≤x1<x2≤n, all the mean values d of the number of neighborhood points that satisfy f(x1)≤d jt ≤f(x2) jt the corresponding k-Dist distances D jt are taken as candidate neighborhood radii to form a candidate neighborhood radius set E.

[0050] wherein, the objective function of the least squares method is as follows:

[0051]

[0052] the function that best matches the original data distribution is found by minimizing the sum of squared errors, and the least squares polynomial curve fitting is as follows:

[0053] f(x)=θ0+θ1x+θ2x 2 +…+θ k x k ,​​​

[0054] Where k is the order of the polynomial, θ i (i = 1, 2, 3, ..., k) are the coefficients of the polynomial.

[0055] The inflection point is obtained by solving for the curvature of the fitted curve. The formula for calculating curvature k is as follows:

[0056]

[0057] In the step of clustering using a clustering parameter set and selecting the optimal clustering result data, for each sampling time, each clustering parameter in the clustering parameter set for that sampling time is used to cluster the standardized ultra-3D target feature vector data corresponding to that sampling time, obtaining a candidate clustering result dataset for that sampling time. The optimal clustering result data is then selected from the candidate clustering result dataset. For example... Figure 3 As shown, this step includes:

[0058] Step S301: For sampling time t, use the minimum number of cluster points MinPts and the radius values ​​of each candidate neighborhood in the candidate neighborhood radius set E as clustering parameters to cluster the standardized ultra-three-dimensional target feature vector data at that sampling time, and obtain the candidate clustering result set at that sampling time.

[0059] Step S302: For sampling time t, the optimal clustering result is selected by evaluating the candidate clustering result set at that sampling time using the silhouette coefficient. This optimal clustering result is taken as the final clustering result at that sampling time, and the k-Dist distance D corresponding to the optimal clustering result is calculated. jy Let Eps be the neighborhood radius at sampling time t. t .

[0060] Specifically, for the standardized data of the hyper-3D target feature vector at sampling time t... First calculate its profile coefficient S i Then calculate all profile coefficients S at sampling time t. i The average is used to obtain the total silhouette coefficient of the clustering results at sampling time t. The optimal clustering result is obtained by comparing the total silhouette coefficients of the candidate clustering results.

[0061] The formula for the profile coefficient is as follows:

[0062]

[0063] Where a(i) is the average distance between sample point i and other sample points within the same cluster, used to quantify the cohesion within the cluster, and b(i) is the average distance between sample point i and its nearest other sample point not belonging to the same cluster, used to quantify the separation between clusters. The silhouette coefficient ranges from [-1, 1], with values ​​closer to 1 indicating better clustering and values ​​closer to -1 indicating worse clustering. If a sample's silhouette coefficient is near 0, it may mean that the sample is located on the boundary between two clusters.

[0064] In the step of determining the cluster, for each sampling time, the cluster at that sampling time is determined based on the optimal clustering result data at that sampling time.

[0065] Specifically, for sampling time t, the mean center of each cluster in the optimal clustering result data is used as the cluster center of the corresponding cluster. Using the cluster center to represent each cluster, the set of clusters W at that sampling time is obtained. t ={w 1t ,w 2t ,…,w gt Each cluster represents a target.

[0066] In the step of determining the correlation between clusters at adjacent sampling times, the correlation between clusters at adjacent sampling times is determined. If they are correlated, the correlated clusters at adjacent sampling times belong to the same target.

[0067] Specifically, for two adjacent sampling times t and t+T, where T is the sampling interval, the neighborhood radii of these two sampling times are Eps, respectively. t and Eps t+T Then the average neighborhood radius of adjacent sampling times t and t+T is Eps. (t,t+T) =(Eps t +Eps t+T The distance between cluster centers at two sampling times is calculated as ) / 2. If the distance between cluster centers at two different sampling times is less than Eps, then... (t,t+T) If the two cluster centers are related, then the clusters represented by the two cluster centers are classified as belonging to the same target.

[0068] In the step of adaptive association of target situation information, for all sampling times in the ultra-3D target feature vector dataset, the clusters are labeled from the initial sampling time, and all ultra-3D target feature vector data are associated with the target to which they belong.

[0069] Specifically, starting from the initial sampling time, cluster centers are labeled. The points in the cluster represented by the cluster center belong to the same target. If a cluster is determined to be the same target as the cluster at the previous time, the two clusters are labeled the same and no new label is assigned to the cluster. Otherwise, a new label is assigned. Finally, all data points in the space are associated with the target to which they belong.

[0070] This embodiment uses heterogeneous data from multiple sensors to construct ultra-three-dimensional target feature data, and uses this as the basis for global situational target clustering and target association. It fully considers target features to perform cluster analysis, thereby improving the efficiency and accuracy of multi-source and multi-target detection data screening and processing in situational analysis.

[0071] This embodiment determines the optimal DBSCAN clustering result based on data characteristics. Specifically, it obtains the inflection point of the fitted curve by calculating the curvature of the fitted curve, and then obtains the candidate neighborhood radius set E. This realizes the automation of DBSCAN clustering parameter adjustment, eliminating the need for manual parameter adjustment, avoiding the interference of subjective factors on the association results, and improving the efficiency and accuracy of target situation information association.

[0072] This embodiment reduces the computational load for associating target situation information by first locking the optimal clustering result and then performing comparison calculations, thereby improving the efficiency of target situation information association.

[0073] Example 2

[0074] This invention is used to describe in detail the electronic device according to one embodiment of the invention, such as... Figure 4 As shown, the electronic device includes at least one processor and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to execute the aforementioned adaptive association method for non-homogeneous target situational information.

[0075] The memory and processor are connected via a bus, which can include any number of interconnecting buses and bridges, connecting various circuits of one or more processors and memories. The bus can also connect various other circuits, such as peripherals, voltage regulators, and power management circuits, via interfaces, as is well known in the art. Interfaces provide a connection between the bus and the transceiver, such as communication interfaces or user interfaces. The transceiver can be a single component or multiple components, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over the wireless medium via an antenna, which further receives data and transmits it to the processor.

[0076] The processor manages the bus and general processing, and also provides various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory is used to store data used by the processor during operation.

[0077] Example 3

[0078] This invention is used to describe in detail, in one embodiment of the invention, a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described embodiment of the adaptive association method for situational information of non-same-source targets.

[0079] Those skilled in the art will understand from the foregoing description that all or part of the steps in the methods of the above embodiments can be implemented by a program instructing related hardware. This program is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes, but is not limited to, various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic storage devices, and optical storage devices.

[0080] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software 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.

[0081] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0082] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in this application should still be covered by the claims of this application.

Claims

1. An adaptive association method for situational information of non-homogeneous targets, comprising: Target feature data from different sensors probing the same physical space at various sampling times are acquired. This target feature data is non-homogeneous and includes velocity, 3D position, reflectivity, and the corresponding sampling time. A hyper-3D target feature vector space is constructed based on this data, and each data point within this hyper-3D target feature vector space is calculated. data points It is standardized ultra-three-dimensional target feature vector data; The steps for determining the clustering parameter set include: for each sampling time, using the standardized hyper-3D target feature vector data corresponding to that sampling time to determine the clustering parameter set for that sampling time; The steps include using a clustering parameter set to cluster and selecting the optimal clustering result data; in this step, for each sampling time, each clustering parameter in the clustering parameter set for that sampling time is used to cluster the standardized ultra-3D target feature vector data corresponding to that sampling time, to obtain a candidate clustering result dataset for that sampling time, and then selecting the optimal clustering result data from the candidate clustering result dataset; For each sampling time, the cluster at that sampling time is determined based on the optimal clustering result data at that sampling time; The correlation between clusters at adjacent sampling times is determined. If they are correlated, the correlated clusters at adjacent sampling times belong to the same target. The step of adaptive association of target situation information; in this step, for all sampling times, the clusters are labeled from the initial sampling time, and all ultra-three-dimensional target feature vector data are associated with the target to which they belong; The steps to determine the clustering parameter set include: Step S201: Set the minimum number of cluster points according to the dimension of the standardized ultra-3D target feature vector data participating in clustering. = ,in, This represents the number of dimensions of the remaining components of the standardized ultra-3D target feature vector data excluding the sampling time. Step S202: For each sampling time t, calculate the distance from all data points at that sampling time to other data points at the same time. Step S203: Let Each data point To its first The distance between the nearest neighbors is used as the k-Dist distance of the data point. Calculate the sampling time Each data point At each k-Dist distance Number of data points in the neighborhood And obtain the distance of each k-Dist. The mean number of corresponding neighborhood points , ; Step S204, all k-Dist distances The mean of the number of corresponding neighborhood points Sort the sequence in ascending order to obtain the sorted mean sequence. Then, use the least squares polynomial fitting method to perform curve fitting on the sorted mean sequence to obtain the fitted curve. , And obtain the fitted curve The inflection points of the two sharp rises and , Select the one that satisfies Mean of the number of all neighboring points Corresponding k-Dist distance These serve as candidate neighborhood radii and form the candidate neighborhood radius set E.

2. The adaptive association method for non-homogeneous target situational information as described in claim 1, wherein, The specific steps for adaptive correlation of target situation information are as follows: Starting from the initial sampling time, cluster centers are labeled. Points in the clusters represented by the cluster centers belong to the same target. If a cluster is determined to be the same target as the cluster at the previous time, the two clusters are labeled the same and no new label is assigned to the cluster. Otherwise, a new label is assigned. Finally, all data points in the super 3D target feature vector space are associated with the target to which they belong.

3. The adaptive association method for non-homogeneous target situational information as described in claim 1, in, Super 3D target feature vector data in super 3D target feature vector space By one of multiple sensors at the sampling time The obtained The group of target feature data consists of, , Indicates the first Velocity component in group target feature data, They represent the first The three coordinate components of the three-dimensional position in the group target feature data. No. The reflectance component in the target feature data. This indicates the sampling time. .

4. The adaptive association method for non-homogeneous target situational information as described in claim 3, wherein, Specifically, the standardized ultra-3D target feature vector data obtained through calculation is as follows: Sampling time Ultra-3D target feature vector data There are n sets of ultra-3D target feature vector data. All components except those at the sampling time are standardized to obtain standardized ultra-3D target feature vector data. , ,in, , , , , , 。 5. The adaptive association method for non-homogeneous target situational information as described in claim 1, wherein, The inflection point is obtained by solving the fitted curve. The curvature is obtained, curvature The solution is as follows: 。 6. The adaptive association method for non-homogeneous target situational information as described in claim 1, wherein, The steps for clustering and selecting the optimal clustering result data include: Step S301: Sampling time Using the minimum number of cluster points MinPts at the sampling time and the radius values ​​of each candidate neighborhood in the candidate neighborhood radius set E as clustering parameters, the standardized ultra-three-dimensional target feature vector data at the sampling time are clustered to obtain the candidate clustering result set at the sampling time. Step S302: Sampling time The optimal clustering result is selected by evaluating the candidate clustering result set at the sampling time using the silhouette coefficient. This optimal clustering result is then used as the final clustering result at that sampling time. The k-Dist distance corresponding to the optimal clustering result is then calculated. Let the neighborhood radius be t. .

7. The adaptive association method for non-homogeneous target situational information as described in claim 6, wherein, The steps to determine cluster types include: For sampling time t, the mean center of each cluster in the optimal clustering result data is used as the cluster center of the corresponding cluster. Using the cluster center to represent each cluster, the set of clusters at that sampling time is obtained. Each cluster represents a target.

8. The adaptive association method for non-homogeneous target situational information as described in claim 7, wherein, The steps for determining the cluster association between adjacent sampling times include: For two adjacent sampling times and , The sampling interval is the neighborhood radius at these two sampling times. and Then the sampling times of adjacent times and The average neighborhood radius is =( + / 2, calculate the distance between cluster centers at two sampling times. If the distance between cluster centers at two different sampling times is less than 2, then... If the two cluster centers are related, then the clusters represented by the two cluster centers are classified as belonging to the same target.

9. An electronic device, comprising: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the adaptive association method for non-homogeneous target situational information as described in any one of claims 1 to 8.

10. A computer-readable storage medium storing a computer program, wherein, When the computer program is executed by the processor, it implements the adaptive association method for non-homogeneous target situational information as described in any one of claims 1 to 8.

Citation Information

Patent Citations

  • Multi-source heterogeneous information association method based on information fusion period

    CN113989602A