A target detection fusion method based on double radar networking

Abstract

The application belongs to the technical field of distributed target detection, and discloses a target detection fusion method based on double radar networking, comprising the following steps: S1: data acquisition: acquiring track data from at least two radar devices, recording the timestamp, latitude and longitude, azimuth angle, pitch angle, speed and other information of each radar, and then executing S2; S2: data preprocessing: correcting the radar data in real time, including angle mapping, north deviation angle calculation and distance error correction, and then executing S3; S3: track point association determination: using the azimuth angle error and the speed difference value as the judgment condition to match the track points of different radars, and then executing S4. The application adopts advanced multi-thread data processing technology, can realize high-precision target detection on the premise of ensuring fast response, improves the overall efficiency of the system by distributing the data processing task to multiple threads for parallel execution, and shortens the time delay of target identification.

Description

Technical Field

[0001] This invention belongs to the field of power electronics technology, specifically a target detection fusion method based on dual-radar networking, and in particular a data fusion and target detection method based on the collaborative operation of dual radar devices. Background Technology

[0002] Currently, the detection capability of a single radar is limited by physical performance: in terms of detection range, it has a weak ability to acquire small targets at long distances and is prone to missed detection; the azimuth resolution is affected by the beam width, making it difficult to distinguish dense targets (such as drone swarms), which can easily lead to point confusion; the number of target points is limited by hardware computing and storage, making them easy to lose in high-density scenarios; there are many blind spots in complex environments, resulting in low detection reliability.

[0003] To overcome the performance bottleneck of single radars, dual-radar collaborative networking has become an effective technical means. However, the heterogeneity of data, different data rates, and asynchronous timestamps brought about by dual-radar collaborative operation, along with complex correlation calculations, make the real-time performance and accuracy of data fusion a key challenge. Existing multi-radar fusion methods often suffer from poor real-time performance due to insufficient processing efficiency when meeting high-precision requirements. While simplifying algorithms can improve efficiency, they sacrifice accuracy and cannot adapt to actual needs. Therefore, how to solve data heterogeneity and achieve high-precision fusion while ensuring real-time performance is a key issue in this field.

[0004] Therefore, a target detection fusion method based on dual radar networking is proposed. This method can improve the target detection rate and accuracy in complex environments and is suitable for application scenarios such as UAV swarm detection, border control, and traffic monitoring. Summary of the Invention

[0005] To address the problems mentioned in the background section, this invention provides a target detection fusion method based on dual-radar networking, comprising the following steps; S1: Data Acquisition: Acquire track data from at least two radar devices, record the timestamp, latitude and longitude, azimuth, elevation angle, speed and other information of each radar, and then execute S2; S2: Data preprocessing: Real-time correction of radar data, including angle mapping, north angle calculation, and range error correction, and then execute S3; S3: Track point association determination: Using azimuth error and velocity difference as judgment conditions, track points of different radars are matched, and then S4 is executed; S4: Fusion Calculation: The associated waypoints are fused using a weighted average method, and the fused target waypoints are calculated and output. Then, S5 is executed. S5: Fusion Result Output: The fused data is stored in a database and displayed in real time on a map, recording the number of detection points for each radar. and And the number of points N after fusion, Where K is the total number of observations within the data collection period.

[0006] Preferably, the timestamp of the radar data is accurate to the second level, ignoring millisecond-level errors. When performing track point association determination, if the timestamp difference Δt between two radar track points is ≤5s, the time matching is considered successful, and the next association determination is performed; otherwise, the current track point is regarded as an unrelated target and stored in an independent track list.

[0007] Preferably, the conditions for determining the correlation of the waypoints include that the azimuth error is within three times the accuracy range and the speed difference is less than three meters per second.

[0008] Preferably, the fusion calculation prioritizes the selection of associated points at the same time. If there is no data at the same time, the track points at the previous time are queried for matching.

[0009] Preferably, the data preprocessing uses the GeographicLib library for geographic coordinate transformation and distance calculation.

[0010] Preferably, the map loading uses locally stored tile data and is visualized based on the Leaflet framework.

[0011] Preferably, the radar data acquired is transmitted to the data processing module via UDP and stored in an SQLite database.

[0012] Preferably, the adjustment of radar equipment parameters to match the data rate involves using an extrapolation algorithm to supplement missing data points in sparse radar detection data, thereby increasing data continuity.

[0013] Preferably, the real-time track information displayed on the map includes single radar detection points, extrapolated track points, and fused track points, which are dynamically updated and displayed using different colors or markers.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention employs advanced multi-threaded data processing technology, which enables high-precision target detection while ensuring rapid response. By distributing data processing tasks to multiple threads for parallel execution, the overall efficiency of the system is improved and the latency of target recognition is shortened.

[0015] Furthermore, in order to improve the accuracy of target detection, the radar data rate and time synchronization mechanism are adjusted to adapt to the performance of different radars. A weighted average fusion algorithm is used to integrate data from different radars. This algorithm assigns different weights to the output of each radar according to the performance and measurement accuracy of each radar, so as to better reflect the characteristics of each radar in the fusion process.

[0016] Furthermore, by using a geospatial database to unify the coordinate system of radar data, data errors caused by different radar observation characteristics are effectively reduced. By converting all radar data into a unified coordinate system, the problem of inaccurate target positioning caused by inconsistencies in coordinate systems is avoided, thus enhancing the accuracy of data fusion. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the implementation steps of the present invention; Figure 2 This is a schematic diagram of the test scenario for the present invention; Figure 3 This is the algorithm logic design diagram of the present invention; Figure 4 This is a flowchart of the loop structure of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To make the technical solution of the present invention clearer, the following description is based on specific embodiments. Before deployment, two radar devices are deployed, namely radar 1 and radar 2, with the distance between the two radars represented by R0. At the same time, the flight path of the UAV swarm is set so that it passes through a specific area within the detection range of the two radars to ensure that the target data can be effectively collected by the two radars. The scanning frequency of radar 2 is adjusted to be consistent with that of radar 1, such as setting the data acquisition interval to 6 seconds. The detection data of radar 2 is extrapolated to supplement data points to improve the detection rate.

[0020] like Figures 1 to 4 As shown, the present invention provides a target detection fusion method based on dual radar networking, including the following steps; S1: Data Acquisition: Acquire track data from at least two radar devices, record the timestamp, latitude and longitude, azimuth, elevation angle, speed and other information of each radar, and then execute S2; S2: Data preprocessing: Real-time correction of radar data, including angle mapping, north angle calculation, and range error correction, and then execute S3; S3: Track point association determination: Using azimuth error and velocity difference as judgment conditions, track points of different radars are matched, and then S4 is executed; S4: Fusion Calculation: The associated waypoints are fused using a weighted average method, and the fused target waypoints are calculated and output. Then, S5 is executed. S5: Fusion Result Output: The fused data is stored in a database and displayed in real time on a map, recording the number of detection points for each radar. and And the number of points N after fusion, Where K is the total number of observations within the data collection period.

[0021] Specifically, after completing the deployment of dual radars, setting the flight path of the UAV swarm, and preparing data recording, target detection fusion is performed according to the following steps: Step 1: Simultaneously acquire target trajectory data from at least two radar devices. Key information to be recorded includes the timestamp of each radar, target latitude and longitude, azimuth, elevation angle, and velocity. The raw data collected is transmitted to the data processing module via UDP protocol and stored in the SQLite database to provide the raw data source for subsequent processing. After the data acquisition is completed, perform step 2 for data preprocessing. The radar data collected in step 1 is corrected in real time, and the angle mapping and north offset are calculated to unify the angle measurement standard. Based on the distance error curve of the radar factory calibration, the measured values ​​of different distance segments are corrected in segments. Then, the GeographicLib library is called to convert the radar local polar coordinates (azimuth, distance, etc.) into WGS84 latitude and longitude. The relative offset of the target is calculated through the latitude and longitude reference point of the radar station and superimposed to achieve coordinate system unification. If the data is sparse (e.g., a certain radar point is missing), the extrapolation algorithm is used to supplement the missing points to enhance the data continuity. After the preprocessing is completed, step 3, the track point association determination, is executed. Matching track points from different radars requires determining the correlation between track points based on time consistency, azimuth error, and velocity difference. Time consistency requires prioritizing the matching of track points at the same time; if none are found, the data from the previous time point is queried, and the difference Δt between the two radar timestamps must be ≤5s (to ensure that the target's movement range is within the radar positioning error). Azimuth error must be within three times the radar azimuth accuracy (based on the normal distribution 3σ principle) and increase proportionally with the target distance, balancing the correlation accuracy with the problem of missing correlation of distant targets; Speed ​​difference: The speed difference between associated points must be ≤3m / s. Track points that meet the above conditions are determined to be associated points and step 4 fusion calculation is performed; otherwise, the track points are regarded as unassociated targets and stored in an independent list. The associated waypoints identified in step 3 are fused using a weighted average method, and the calculation formula is as follows: ,in, and The weights of the two radars are set according to their measurement accuracy. Priority is given to selecting the associated points at the same time moment for fusion; if none are found, the associated points at the previous time moment are used. After calculating the fused target track points (including timestamp, latitude and longitude, speed, etc.), step 5 is executed to output the fusion result. The fused trackpoint data (including individual radar detection points, extrapolated trackpoints, and fused points) is stored in a fusion table with fields including timestamp, latitude and longitude, speed, and altitude. Then, a local tile map is loaded based on the Leaflet framework, dynamically displaying various trackpoints with different colors / markers. This achieves real-time visualization and interaction between the front-end and back-end, and records the number of radar 1 detection points within the data collection period. Radar 2 And the number of points N after fusion, through the formula Where K is the total number of observations, the detection rate improvement ratio is calculated, and the fusion effect is quantified. Through the above steps, efficient fusion of dual radar data is achieved, improving the detection rate and accuracy of group targets, which is suitable for distributed scenarios such as UAV swarm detection.

[0022] like Figures 1 to 4 As shown, the timestamps of radar data are accurate to the second level, ignoring millisecond-level errors. When determining the association of track points, if the difference in timestamps between two radar track points Δt ≤ 5s, the time matching is considered successful, and the next step of association determination is initiated; otherwise, the current track point is regarded as an unrelated target and stored in an independent track list.

[0023] Specifically, by standardizing the timestamp accuracy to the second level, a directly comparable time scale is provided for the track data of Radar 1 and Radar 2, avoiding meaningless data fragmentation caused by millisecond-level differences, reducing interference from the time dimension, and providing standardized input for subsequent association determination. The timestamp difference between track points in Radar 1 and Radar 2 is Δt. If Δt≤5s, combined with a 6-second data acquisition interval, the target can move a maximum of 100m within 5 seconds. The threshold setting of Δt≤5s, combined with the 6-second data acquisition interval and target motion characteristics, ensures that time-matched track points have spatial association possibilities, while filtering out invalid data with Δt>5s, reducing redundant calculations, and improving the efficiency of association determination. If the value is within the radar positioning error range and is less than the acquisition interval, it can cover the adjacent sampling period of asynchronous data, balancing the association success rate and timeliness. In this case, the time match is considered successful, and the next step of association determination is carried out. Otherwise, the current track point is regarded as an unrelated target and stored in an independent track list.

[0024] like Figures 1 to 4 As shown, the conditions for determining the correlation between track points include an azimuth error within three times the accuracy range and a speed difference of less than three meters per second.

[0025] Specifically, the allowable error in the distance between the target and the radar is three times the azimuth accuracy. Based on the normal distribution 3σ principle, the radar azimuth measurement error has a 99.7% probability of falling within three times the accuracy. At the same time, the influence of target distance is considered. The allowable error range increases proportionally with the distance. This ensures the accuracy of the correlation and avoids missing correlation due to large spatial deviations of distant targets. The velocity difference of the correlated track points is less than or equal to 3 m / s.

[0026] like Figures 1 to 4 As shown, the fusion calculation prioritizes related points at the same time. If there is no data at the same time, it queries the track points at the previous time for matching.

[0027] Specifically, assuming the two points are related, a weighted average method is used for data fusion processing of already related waypoints. ,in, and The weights of the two radars can be set according to actual needs. The fused track point data includes fields such as timestamp, latitude, longitude, altitude, speed, and azimuth and elevation angles, and is stored in the fusion table for subsequent display and analysis. The fused track points are loaded onto the offline map through the front-end page for display. If the track points of the two radars at the same time are found to meet the conditions that the azimuth error is within three times the accuracy and the velocity difference is ≤3m / s, then these points are directly used as the correlation points for fusion calculation to ensure that the fusion result is based on the most real-time synchronized data and to minimize the error caused by the time difference. If no related points are found at the same time, and the timestamps are out of sync due to differences in radar data rates or transmission delays, the system will automatically backtrack to the previous time point, extract the track points of the two radars at that time, and re-verify the azimuth error and velocity difference conditions. If the correlation rules are met, the track points of the previous time point will be used as the matching object for fusion. The continuity of historical data will make up for the defects of time asynchrony and avoid fusion interruption caused by data gaps.

[0028] like Figures 1 to 4 As shown, data preprocessing uses the GeographicLib library for geographic coordinate transformation and distance calculation.

[0029] Specifically, to ensure that the data from the two radars are processed in a unified coordinate system, the GeographicLib library is used to correct the latitude, longitude, and azimuth of radar 1. When using the GeographicLib library for correction, the local polar coordinates (azimuth, range, etc.) of the radar are first converted to WGS84 latitude and longitude. The latitude and longitude of the radar station are used as the reference point, and the eastward and northward offset of the target relative to the radar is calculated in combination with the azimuth. Then, the coordinates are superimposed on the reference point. At the same time, the range error curve of the radar's factory calibration is introduced to correct the measured values ​​of different range segments in segments. For cases where the data is sparse, an extrapolation algorithm is used to supplement missing points and increase data continuity.

[0030] like Figures 1 to 4 As shown, the map loads using locally stored tile data and is visualized based on the Leaflet framework.

[0031] Specifically, the map is loaded using locally stored tile data, eliminating the dependence on the network environment and avoiding problems such as map loading failure and lag caused by network latency, interruption or bandwidth limitations. This ensures that the target track can still be stably displayed in environments without network or with weak network (such as field radar deployment scenarios).

[0032] like Figures 1 to 4 As shown, the radar data acquired is transmitted to the data processing module via UDP and stored in the SQLite database.

[0033] Specifically, the UDP protocol eliminates the need for connection establishment and has low data transmission overhead, enabling it to quickly push radar-acquired track data (timestamps, latitude and longitude, speed, etc.) to the data processing module in real time, reducing transmission latency. For detection scenarios involving high-speed moving targets such as UAV swarms, this low-latency transmission ensures that subsequent data preprocessing and track association steps are performed based on the latest data, avoiding track matching deviations caused by transmission lag. The choice of SQLite database to store the data from the two radars separately offers advantages such as lightweight, file-based operation, eliminating the need for complex server configurations. This makes it suitable for distributed deployment of radar equipment in non-data center environments such as the field or border areas. Furthermore, the separate storage design clearly distinguishes the raw data from Radar 1 and Radar 2, facilitating individual calls during subsequent data preprocessing (such as using the GeographicLib library to correct coordinates). It also provides a structured data foundation for comparing cross-radar data in track association determination.

[0034] like Figures 1 to 4 As shown, the radar equipment parameters are adjusted to match the data rate, and the missing points in the sparse radar detection data are supplemented by an extrapolation algorithm to increase data continuity.

[0035] Specifically, when two radars work together, there may be differences in data rates (such as different scanning frequencies). By adjusting the radar equipment parameters (such as scanning frequency) to match their data rates, the time dimension mismatch caused by different data generation speeds can be eliminated. This provides a common data foundation for subsequent trackpoint association determination. For data sparsity caused by differences in radar performance or limitations in target detection conditions, extrapolation algorithms can be used to supplement missing tracks, increasing data continuity. This ensures that both radars have enough tracks to participate in the matching during trackpoint association, avoiding missed associations due to insufficient data from one radar, thereby improving the group target detection rate. In the target detection fusion scenario of dual-radar networking, the extrapolation algorithm can be implemented in two ways: linear extrapolation and Kalman filter extrapolation. Linear extrapolation: Assuming the target moves at a constant speed in a straight line for a short period of time, the average speed and direction are calculated based on the most recent n historical points to estimate the location of the missing point; Kalman filter extrapolation: Establish the target motion state equation (such as a uniform acceleration model), and estimate the optimal state by predicting and updating iteratively, fusing historical observations and system noise. The optimal state should be selected based on the scheme requirements and data characteristics.

[0036] like Figures 1 to 4 As shown, the real-time track information displayed on the map includes individual radar detection points, extrapolated track points, and fused track points, which are dynamically updated using different colors or markers.

[0037] Specifically, different colors are used to display the detection points of a single radar (raw data), extrapolated points (supplementary data), and fused track points (final results). This clearly distinguishes the source and processing status of the data. For example, blue marks the original point of radar 1, green marks the original point of radar 2, yellow marks the extrapolated point, and red marks the fused point. This allows operators to quickly identify the distribution and correlation of various types of data, understand the optimization process of the fusion algorithm on the original data, and solve the visualization confusion caused by the mixing of multiple data sources.

[0038] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0039] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A target detection fusion method based on dual-radar networking, characterized in that, Includes the following steps; S1: Data Acquisition: Acquire track data from at least two radar devices, record the timestamp, latitude and longitude, azimuth, elevation angle, speed and other information of each radar, and then execute S2; S2: Data preprocessing: Real-time correction of radar data, including angle mapping, north angle calculation, and range error correction, and then execute S3; S3: Track point association determination: Using azimuth error and velocity difference as judgment conditions, track points of different radars are matched, and then S4 is executed; S4: Fusion Calculation: The associated waypoints are fused using a weighted average method, and the fused target waypoints are calculated and output. Then, S5 is executed. S5: Fusion Result Output: The fused data is stored in a database and displayed in real time on a map, recording the number of detection points for each radar. and And the number of points N after fusion, Where K is the total number of observations within the data collection period.

2. The target detection fusion method based on dual radar networking according to claim 1, characterized in that: The timestamps of the radar data are accurate to the second level, ignoring millisecond-level errors. When performing track point association determination, if the timestamp difference Δt between two radar track points is ≤5s, the time matching is considered successful, and the next step of association determination is performed; otherwise, the current track point is regarded as an unrelated target and stored in an independent track list.

3. The target detection fusion method based on dual radar networking according to claim 2, characterized in that: The conditions for determining the correlation of the track points include that the azimuth error is within three times the accuracy range and the speed difference is less than three meters per second.

4. The target detection fusion method based on dual radar networking according to claim 3, characterized in that: The fusion calculation prioritizes related points at the same time. If no data is available at the same time, the tracking points from the previous time are queried for matching.

5. The target detection fusion method based on dual radar networking according to claim 1, characterized in that: The data preprocessing uses the GeographicLib library for geographic coordinate transformation and distance calculation.

6. The target detection fusion method based on dual radar networking according to claim 1, characterized in that: The map loading uses locally stored tile data and is visualized based on the Leaflet framework.

7. The target detection fusion method based on dual radar networking according to claim 1, characterized in that: The radar data acquired is transmitted to the data processing module via UDP and stored in an SQLite database.

8. The target detection fusion method based on dual radar networking according to claim 1, characterized in that: The radar equipment parameters are adjusted to match the data rate, and extrapolation algorithms are used to supplement missing points in sparse radar detection data, thereby increasing data continuity.

9. The target detection fusion method based on dual radar networking according to claim 1, characterized in that: The map displays real-time track information including individual radar detection points, extrapolated track points, and fused track points, which are dynamically updated using different colors or markers.

Images