A calibration method for multi-radar data fusion in maritime radar systems

Abstract

The present invention belongs to the field of maritime radar and navigation and positioning technology, and specifically discloses a calibration method for multi-radar data fusion of a maritime radar system. The method combines the characteristics of pulse compression radar and electronic scanning radar, and adopts rigid transformation methods such as iterative nearest point algorithm to accurately calibrate the original data of different radars, ensuring the accurate alignment of data in the global positioning system GPS coordinate system, thereby significantly improving the conversion accuracy and consistency of radar data. The method of the present invention significantly improves the stability of the system in complex environments, reduces system deviations caused by inconsistent data, and ensures the reliable operation of ship navigation and collision avoidance systems. The method of the present invention has powerful real-time processing capabilities, can effectively cope with rapid changes in highly dynamic marine environments, and ensures that real-time conversion and fusion of radar data can be achieved when the ship is moving at high speed, ensuring the safe navigation of ships in complex and dynamic environments.

Description

Technical Field

[0001] The invention belongs to the technical field of maritime radar, navigation and positioning, and in particular relates to a calibration method for multi-radar data fusion of a maritime radar system. Background Art

[0002] With the rapid development of modern maritime transportation systems, the application of radar technology in ship navigation, navigation safety and collision avoidance systems has become increasingly important. Pulse compression radar (referred to as pulse compression radar) and electronically scanned array radar (referred to as electronically scanned radar), as key components of maritime radar systems, play a vital role in target detection, positioning and tracking. However, due to the different working principles of pulse compression radar and electronically scanned radar, there are significant differences in the output data format and coordinate system they generate. This difference has brought great challenges to the fusion and unified calibration of multi-radar data. In a dynamic maritime environment, especially when ships are sailing, high-precision radar data fusion is an important means to ensure safe navigation and collision avoidance.

[0003] Most of the current calibration technologies are designed for land environments, such as the joint calibration of LiDAR and cameras. These methods are mainly used in application scenarios such as autonomous driving, relying on the integration of road sensors and positioning systems such as GPS, and are relatively mature. Although these technologies have achieved remarkable results in the field of land transportation, there are still many challenges in directly applying them to maritime environments.

[0004] At present, there is no efficient calibration method specifically for maritime applications. Therefore, how to accurately convert the data of maritime sensors such as pulse pressure radar and electronic scanning radar in maritime radar systems into a unified GPS coordinate system is still a technical difficulty. In addition, due to the differences in working principles and output data formats of different types of radar systems, directly converting their coordinate data into a unified GPS coordinate system will involve complex mathematical transformations, and the existing methods have low computational efficiency and cannot meet the needs of real-time applications, especially when the ship is moving at high speed. In addition, in the dynamic environment at sea, the speed and heading of the target ship change frequently, which puts higher precision requirements on the fusion of multi-radar data. However, the existing calibration methods have low accuracy in high-dynamic environments, resulting in unsatisfactory radar data fusion effects, thereby affecting the overall accuracy of ship navigation and positioning.

[0005] In the maritime environment, the detection distance and accuracy of radar are required to be high, and the motion characteristics of the target are more complex. At present, there is a lack of special calibration methods for marine sensors (such as pulse compression radar and electronically scanned radar) in the maritime environment. Therefore, how to achieve effective calibration and fusion of pulse compression radar and electronically scanned radar data with GPS system in the maritime field, and then achieve unified ship target positioning in the global coordinate system, has become a technical problem that needs to be solved urgently. Summary of the invention

[0006] In view of the difficulty in fusing multiple maritime radar data in a maritime environment, the present invention proposes a calibration method for multi-radar data fusion of a maritime radar system, which can accurately convert the coordinate data of pulse compression radar and electronic scanning radar into the global positioning system GPS coordinate system, thereby facilitating the unified processing and fusion of different types of radar data.

[0007] In order to achieve the above object, the present invention adopts the following technical scheme:

[0008] A calibration method for multi-radar data fusion of a maritime radar system comprises the following steps:

[0009] Step 1. Obtain relevant data information of the target ship from pulse pressure radar, electronic scanning radar and global positioning system GPS;

[0010] Step 2. Extract key fields, convert formats, filter and clean, perform linear interpolation processing, and perform time alignment preprocessing on the relevant data information collected by pulse pressure radar, electronic scanning radar, and global positioning system GPS;

[0011] The key fields extracted from pulse pressure radar and electronic scanning radar include track ID, longitude, latitude, timestamp, speed over ground and course over ground information; the key fields extracted from GPS include the longitude, latitude and timestamp data of the ship;

[0012] Step 3. Use timestamps and track IDs to match to ensure that the tracks from different data sources are correctly aligned, ensuring that the data from pulse pressure radar, electronic scanning radar and GPS accurately point to the track of the same ship;

[0013] Step 4. Calculate the optimal transformation matrix for transforming the data of pulse pressure radar and electronic scanning radar into the GPS coordinate system respectively;

[0014] Step 5. After obtaining the optimal conversion matrices for converting the pulse compression radar and the electronic scanning radar to the GPS coordinate system, use the two optimal conversion matrices to convert the original latitude and longitude data of the corresponding radars to the unified GPS coordinate system.

[0015] The present invention has the following advantages:

[0016] As described above, the present invention relates to a calibration method for multi-radar data fusion for maritime radar systems. The method combines the characteristics of pulse compression radar and electronic scanning radar, and adopts coordinate conversion methods such as iterative closest point ICP algorithm, least squares method (LSM) or point cloud alignment based on Procrustes analysis to accurately calibrate the original data of different radars. This method can ensure the accurate alignment of data in the global positioning system GPS coordinate system, thereby significantly improving the conversion accuracy and consistency of radar data and reducing error accumulation. This high-precision radar coordinate conversion is crucial to improving the performance of the overall navigation and positioning system. Through accurate coordinate conversion algorithms and optimized calibration methods, the present invention significantly improves the stability of the system in complex environments, reduces system deviations caused by inconsistent data, and ensures the reliable operation of ship navigation and collision avoidance systems. In particular, the introduction of automated time synchronization and trajectory matching mechanisms effectively reduces potential problems caused by differences in data sources and improves the stability and safety of the system. This enhanced stability not only improves the overall performance of the system, but also ensures the safe operation of ships in complex environments. The calibration method provided by the present invention has powerful real-time processing capabilities and can effectively cope with rapid changes in highly dynamic marine environments. Through efficient algorithm design, it ensures that real-time conversion and fusion of radar data can be achieved even when the ship is moving at high speed. This meets the strict requirements for navigation and positioning accuracy in maritime transportation and ensures the safe navigation of ships in complex and dynamic environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a flow chart of a calibration method for multi-radar data fusion for maritime radar in Example 1 of the present invention;

[0018] Figure 2 This is a flowchart of an iterative closest point ICP algorithm in Embodiment 1 of the present invention;

[0019] Figure 3 It is the original trajectory map of pulse pressure radar, electronic scanning radar and GPS;

[0020] Figure 4 This is a comparison chart of the electronic scanning radar and GPS tracks before and after the conversion;

[0021] Figure 5 This is a comparison of the pulse compression radar and GPS tracks before and after the conversion;

[0022] Figure 6 This is the effect diagram of the pulse pressure radar, electronic scanning radar and GPS trajectory alignment after coordinate conversion. DETAILED DESCRIPTION

[0023] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0024] Example 1

[0025] This embodiment describes a calibration method for multi-radar data fusion for maritime radar. The method obtains the target ship trajectory based on GPS as a reference, and converts the original longitude and latitude data of pulse compression radar and electronic scanning radar into the GPS coordinate system through matrix operation, thereby realizing data calibration of pulse compression radar and electronic scanning radar respectively, thereby ensuring the unified mapping of multiple radar data in the maritime environment in the GPS global coordinate system, and realizing the accurate fusion and calibration of multiple radar perception data.

[0026] The method of the present invention includes five steps: data acquisition, data preprocessing, trajectory matching, conversion matrix calculation and data conversion. Through these five steps, radar data from different sources can be converted to a unified GPS coordinate system, thereby solving the problem of multiple sensor data fusion. Figure 1 The calibration method in this embodiment is described in detail.

[0027] like Figure 1 As shown, a calibration method for multi-radar data fusion of a maritime radar system comprises the following steps:

[0028] Step 1. Data collection.

[0029] Obtain relevant data information of the target ship from pulse pressure radar, electronic scanning radar and global positioning system GPS. Pulse pressure radar and electronic scanning radar respectively collect information such as longitude, latitude, track ID, timestamp, ground speed (speed) and ground course (direction) of the ship; at the same time, GPS collects data such as longitude, latitude and timestamp of the ship as a global coordinate reference.

[0030] Step 2. Data preprocessing.

[0031] Since the data formats from pulse pressure radar, electronically scanned radar and GPS are inconsistent and may contain redundant information, they need to be processed in a unified manner. The data obtained by pulse pressure radar, electronically scanned radar and GPS are converted into a unified format. Key fields, including longitude, latitude, track ID and timestamp, are extracted and formatted. Through time synchronization technology, multi-source data are time-aligned to ensure that the data obtained by different radars and GPS systems are consistent at the same time point.

[0032] The specific steps of step 2 are:

[0033] Step 2.1. First, extract key fields from the data collected by pulse compression radar and electronic scanning radar, including track ID, longitude, latitude, timestamp, ground speed and ground heading information.

[0034] During the extraction process, a rule-based parsing method is used to extract key information from the original data file using regular expressions and field delimiters, and generate structured data formats such as CSV or JSON for subsequent processing. Next, key fields are extracted from the data collected by GPS, including the longitude, latitude and timestamp data of the ship. The data collected by GPS is extracted in the same way as the data collected by pulse pressure radar and electronic scanning radar, ensuring that the fields extracted from all data sources are consistent.

[0035] Step 2.2. After extracting the key fields, convert the key field data format and unify the timestamps of data from different sources into UTC format to ensure time consistency. Screen and clean incomplete or redundant trajectory data, remove invalid data, and ensure that each trajectory data contains complete key information. To further enhance the comparability of the data, linear interpolation is also required to fill in the missing blank points in the ship trajectory and ensure the continuity of the ship trajectory in time.

[0036] Step 2.3. After processing in steps 2.1 and 2.2, the data from pulse compression radar, electronic scanning radar and GPS are time aligned using an automated time synchronization algorithm, namely timestamp alignment and linear interpolation, to ensure that the ship trajectory data at the same time point can be effectively matched, laying the foundation for subsequent trajectory matching and coordinate conversion.

[0037] This series of precise data preprocessing steps ensures that radar data from different sources can be effectively integrated under unified standards, providing a solid data foundation for achieving high-precision ship navigation and positioning.

[0038] Timestamp alignment refers to correcting the raw data of different sensors (in this embodiment, pulse pressure radar, electronic scanning radar and GPS) according to time, finding the closest time point of each sensor at the same time, so as to ensure that multi-source data can be accurately matched. Since the sampling frequency and timestamp accuracy of different sensors may be different, it is not always feasible to directly align timestamps, so it is necessary to select a sensor as a benchmark and synchronize the data of other sensors according to this benchmark.

[0039] In this embodiment, the GPS data is used as a reference, and the data of the pulse pressure radar and the electronic scanning radar are synchronized based on this reference.

[0040] When the timestamps of the sensors cannot be completely aligned, this embodiment also introduces linear interpolation, which estimates the value of the intermediate point between the known data points to fill the time difference. Linear interpolation calculates the sensor data at the unmatched time point based on the data of the two closest known time points, ensuring that the trajectory information remains continuous in time.

[0041] The present invention uses an automated time synchronization algorithm to enable data from all sensors to be seamlessly integrated at the same time point, and to enable data from different radars and GPS to be accurately aligned and integrated within the same time frame.

[0042] Figure 3 This is the trajectory diagram of pulse compression radar, electronic scanning radar and GPS after data preprocessing. This figure shows the original trajectory data of pulse compression radar, electronic scanning radar and GPS before coordinate conversion. The blue trajectory line is the target ship trajectory collected by the shipborne GPS system, the yellow trajectory is the target ship trajectory scanned by the pulse compression radar, and the red trajectory line is the target ship trajectory obtained by the electronic scanning radar.

[0043] Compared with the traditional multi-radar data calibration method, the present invention optimizes the data preprocessing steps, unifies the multi-source data format, and introduces an automated time synchronization and trajectory matching mechanism, which significantly simplifies the complex calculation steps. In addition, in terms of automated time synchronization, the present invention method standardizes and aligns the timestamps of multi-source radar data, eliminates the time differences between different sensors, and ensures the consistency of multi-radar data at the same time point.

[0044] Step 3. Trajectory matching.

[0045] During the track matching phase, the main goal is to ensure that the data from pulse compression radar, electronic scanning radar and GPS accurately point to the same ship's track. Since the ship's speed or heading may change, the track ID may be inconsistent.

[0046] In this embodiment, the timestamp and track ID are used in combination for matching to ensure that the tracks from different data sources can be correctly aligned, and to ensure that the data from the pulse pressure radar, electronic scanning radar and GPS accurately point to the track of the same ship.

[0047] Step 3.1. First, perform preliminary screening based on the track ID from the data preprocessed in step 2, and group the records with the same track ID together. The track ID is usually assigned by the radar system to identify the continuous navigation track of the same ship.

[0048] This step preliminarily categorizes data from different sources to ensure that the data points to the same ship in the subsequent matching process.

[0049] Step 3.2. Unify the timestamp format of each data source (such as converting to UTC format), and set a time window (for example, ±5 seconds). Search for matching records from three different data sources, namely, pulse pressure radar, electronic scanning radar, and GPS, within this time window. Use exact matching and approximate matching strategies to compare timestamps to ensure that as many matching records as possible can be found.

[0050] Specifically, the matching criteria include the proximity of timestamps and the consistency of track IDs.

[0051] When the difference between timestamps from different data sources is within a small range, for example, within a preset range threshold (the preset range threshold may be set to 1 second), the records are directly considered to be an exact match.

[0052] If the timestamps differ greatly within the set time window, for example, they exceed the preset range threshold within the set time window, but are still within the allowable range threshold (for example, ±5 seconds), then further matching is performed in combination with the speed and heading parameters.

[0053] By comparing the actual movement trajectories of the ships, including speed and heading, it is ensured that the matching records belong to the same ship.

[0054] This step is to find the corresponding records in different data sources through timestamp comparison, ensuring that the data collected by different sensors point to the same ship at the same time, thereby laying a temporal foundation for subsequent trajectory matching and fusion.

[0055] Step 3.3. After matching the timestamp in step 3.2, the track ID and timestamp are combined to further confirm that the track data from different data sources point to the same ship. In this process, not only the time and ID matching are verified, but also multiple conditions including the consistency of latitude and longitude position, the rationality of speed and heading are further verified and matched. Those records that are obviously inconsistent in time and space will be marked as abnormal records and excluded from matching.

[0056] Records that are obviously inconsistent in time and space, for example, refer to the case where the trajectory data of two data sources exceed the preset time window in time, and the location difference in longitude and latitude (i.e., the spatial difference) is greater than 0.01 degrees.

[0057] In order to enhance the accuracy of matching, the nearest neighbor matching algorithm is used to search for the nearest point within the time window, and the changes in speed and heading are combined to ensure the consistency of matching records in time and space. The reliability of the matching results is further verified using a multiple condition matching method, that is, checking whether the geographical locations of all matching records at the same time point are consistent, ensuring that the differences in longitude and latitude are within the set tolerance range, and comparing the speed and heading of the matching records to mark any differences beyond a reasonable range.

[0058] Here, the speed and heading are beyond the reasonable range, for example, if the speed of a ship displayed in the pulse pressure radar is 10 knots at the same time stamp, and the speed displayed in the electronic scanning radar is 20 knots, then this point is marked as an abnormal point. In addition, the heading changes greatly in a short period of time, for example, it is 10° at this moment and 180° at the next moment.

[0059] Records that fail to pass the matching verification will be marked as abnormal records and excluded from the matching. All records that pass the verification (i.e., the records within the time difference, longitude and latitude, speed, and heading differences mentioned above are the verified track data) are integrated into a unified track data, including track ID, longitude, latitude, timestamp, speed, and heading information.

[0060] At this point, the data from different radars and GPS systems form a complete ship track, providing reliable input for subsequent coordinate conversion. This track matching process ensures that the ship tracks from different data sources are accurately aligned in time and space, thus forming a unified and complete track data. The data will contain information such as track ID, longitude and latitude, timestamp, speed and heading, and provide high-precision input for subsequent coordinate conversion and multi-source radar data fusion.

[0061] Through the detailed trajectory matching process in step 3 above, this embodiment can effectively integrate data from multiple source radars, improve the accuracy and reliability of data fusion, and lay a solid foundation for subsequent coordinate transformation.

[0062] Figure 4 This is a comparison chart of the converted electronic scanning radar and GPS tracks. This figure shows the comparison of the electronic scanning radar track before and after conversion, where the blue track line represents the track of the shipborne GPS system as a reference track. The red track line is the electronic scanning radar data before conversion; the yellow track line is the track line of the electronic scanning radar after coordinate conversion.

[0063] Figure 5 This is a comparison of pulse compression radar and GPS trajectory after conversion. This figure shows the comparison of pulse compression radar data with GPS trajectory before and after conversion. The blue trajectory line is the trajectory of the shipborne GPS system for reference. The yellow is the pulse compression radar trajectory before conversion, and the dark blue trajectory line is the trajectory of the pulse compression radar obtained through coordinate conversion.

[0064] Above Figure 4 and Figure 5 The trajectory changes of the electronic scanning radar and the pulse compression radar before and after matching are respectively shown. It is not difficult to see from the accompanying drawings that the present invention accurately aligns the trajectories of different data sources to ensure that the data from the radar and the GPS are consistent.

[0065] During the trajectory matching process, when the speed or heading of the ship changes, the track ID may change accordingly. The present invention ensures that the data from different sensors point to the same ship by combining the timestamp with the track ID. The trajectory matching mechanism proposed in the present invention, combined with the characteristics of the ship's motion trajectory, realizes efficient multi-source radar data matching, reduces redundant operations and complex manual intervention, significantly reduces computational complexity, and improves the operating efficiency of the overall system.

[0066] Step 4. Transformation matrix calculation.

[0067] The optimal transformation matrices for transforming the data of the pulse compression radar and the electronic scanning radar into the GPS coordinate system are calculated respectively, and each optimal transformation matrix includes an optimal rotation matrix and a translation vector.

[0068] In order to realize the conversion of different radar data to the GPS coordinate system, the improved iterative closest point (ICP) algorithm is used in this embodiment to calculate the optimal conversion matrix to realize the conversion of constant pulse pressure radar or electronic scanning radar data to the GPS coordinate system. The purpose of this algorithm is to align the source point cloud of pulse pressure radar and electronic scanning radar with the target point cloud of GPS data through the rotation matrix R and the translation vector t, and minimize the mean square error between them.

[0069] Weighted centroid calculation and adaptive weight assignment are introduced in the process to enhance the robustness and accuracy of the algorithm.

[0070] like Figure 2 The flowchart of the ICP algorithm is shown in Figure 1. This figure shows the specific steps of the ICP (Iterative Closest Point) algorithm, and details how to achieve data alignment and calibration by calculating the closest points between the source point cloud and the target point cloud.

[0071] The specific steps of using the improved iterative closest point ICP algorithm to calculate the optimal transformation matrix are as follows:

[0072] Step 4.1. Initial alignment.

[0073] Given a source point cloud in a pulse compression radar or electronically scanned radar and GPS target point cloud , and assume an initial transformation matrix .

[0074] Represents the source point cloud The point in Represents the target point cloud The point in , , m, n represent the source point cloud and target point cloud The number of points in ; represents the initial rotation matrix, Represents the initial translation vector.

[0075] Step 4.2. Closest point matching.

[0076] For each point in the source point cloud , find the point in the target point cloud that is closest to it , find the corresponding point pair through the nearest neighbor search, that is, match .

[0077] Step 4.3. Calculate the weighted centroid.

[0078] By and target point cloud Different weights are assigned to the points in the image, and the weighted centroid is calculated to enhance the influence on important points. The calculation formula of the weighted centroid is:

[0079] , .

[0080] in, , Represent the weighted centroids of the source point cloud and the target point cloud respectively, Represents the source point cloud midpoint The weight of Represents the target point cloud midpoint The weight of .

[0081] Dynamic weights are used to calculate the weighted centroid. Points with high matching accuracy are given larger weights, and points with large errors are given smaller weights.

[0082] Step 4.4. Center the point cloud.

[0083] Subtract the weighted centroid from each point in the source point cloud and the target point cloud to obtain the centralized point cloud:

[0084] , .

[0085] in, represents the centralized source point cloud, Represents the centered target point cloud.

[0086] Step 4.5. Calculate the rotation matrix .

[0087] First calculate the covariance matrix , the calculation formula is as follows:

[0088] .

[0089] Covariance matrix Perform singular value decomposition SVD, the calculation formula is as follows:

[0090] .

[0091] in, and is a column orthogonal matrix; calculate the rotation matrix , the calculation formula is as follows:

[0092] .

[0093] Step 4.6. According to the rotation matrix and weighted centroid to calculate the translation vector :

[0094] .

[0095] Step 4.7. Combine the rotation matrix and translation vector into a new rigid transformation matrix :

[0096] .

[0097] Step 4.8. Use the new rigid transformation matrix Update the points in the source point cloud and repeat steps 4.2 to 4.7 until the rigid transformation matrix converges or the set number of iterations is reached.

[0098] Finally, by minimizing the following objective function To find the best rotation matrix and translation vectors :

[0099] ;

[0100] in, are corresponding point pairs in the source and target point clouds.

[0101] After the above step 4, an optimal conversion matrix for converting the pulse compression radar to the GPS coordinate system and an optimal conversion matrix for converting the electronic scanning radar to the GPS coordinate system are obtained respectively.

[0102] This embodiment designs an efficient coordinate conversion algorithm, which is specifically used for the conversion between pulse compression radar and electronic scanning radar. The algorithm is based on the iterative closest point ICP method and uses accurate matrix operations to uniformly convert the original latitude and longitude data of different radars into the GPS coordinate system. The algorithm further improves the alignment accuracy and consistency of different types of radar data in the process of multiple sensor data conversion through adaptive weighted centroid calculation and dynamic weight distribution mechanism, and significantly reduces the cumulative error. This design ensures high-precision coordinate conversion in multi-source radar systems and improves overall system performance. The improved ICP algorithm of adaptive weighted centroid calculation and weight distribution innovatively introduced by the present invention ensures that in a high-dynamic marine environment, radar data can adaptively adjust weights during time synchronization and coordinate fusion, and automatically compensate for errors between multiple radar data. In each iteration, the weights of the radar source point cloud and the target point cloud are dynamically adjusted according to the point cloud matching quality, ensuring that points with larger errors contribute less to the overall result, while points with high matching accuracy contribute more to the result, improving the overall fusion accuracy, and ensuring high-precision real-time fusion and calibration of multi-source radar data in complex and dynamic scenes.

[0103] Step 5. Data conversion.

[0104] After obtaining the optimal conversion matrices for converting pulse compression radar and electronic scanning radar to the GPS coordinate system respectively, these two optimal conversion matrices are used to convert the original latitude and longitude data of the corresponding radars to a unified GPS coordinate system to complete the calibration work.

[0105] The specific operation is to transform the radar latitude and longitude through the rotation matrix The translation vector t is converted into the longitude and latitude of GPS coordinates. The original longitude and latitude data of pulse pressure radar or electronic scanning radar are converted into the GPS coordinate system according to the following formula.

[0106] ;

[0107] in The table shows the original longitude and latitude of pulse compression radar or electronic scanning radar. is the converted GPS coordinate.

[0108] Figure 6 This is the alignment effect diagram of the converted trajectory. This figure shows the alignment effect of all radar data and GPS trajectory after the ICP algorithm. The three (pulse pressure radar, electronic scanning radar, GPS) show consistent trajectories in the same coordinate system. The blue trajectory line is the GPS trajectory as a reference, the red one is the trajectory after the electronic scanning radar conversion, and the dark blue one is the trajectory after the pulse pressure radar conversion.

[0109] This embodiment, through the proposed multi-radar data fusion method based on coordinate system calibration, can efficiently convert pulse pressure radar, electronic scanning radar and global positioning system (GPS) data, realize unified processing of different types of radar data, ensure that the data from two different radar devices, pulse pressure radar and electronic scanning radar, are accurately represented in a unified global coordinate system, and ensure the precise alignment of multi-source data. Finally, the calibration of different radar devices in the maritime environment is realized, so that all ship trajectory data can be fused and analyzed in a unified GPS coordinate system. This embodiment effectively combines shipborne radar data with GPS system data for different types of radar sensors through trajectory matching technology and multi-source data fusion mechanism, and enhances the data fusion capability between different radar systems. In addition, the present invention reduces manual intervention through automated time synchronization and trajectory alignment, realizes efficient fusion and calibration of multi-radar sensor data, not only improves the stability and real-time performance of ship navigation and collision avoidance systems in complex dynamic scenes, but also ensures the reliability and safety of the system in extreme environments.

[0110] Example 2

[0111] This embodiment 2 also describes a calibration method for multi-radar data fusion of maritime radar. Except for the following technical features that are different from the above embodiment 1, the remaining technical features of this method can refer to the above embodiment 1.

[0112] In this embodiment, the optimal conversion matrix is ​​calculated based on the least square method LSM, replacing the above ICP algorithm to achieve the conversion of constant pulse pressure radar or electronic scanning radar data to the GPS coordinate system; the specific steps are as follows:

[0113] The point cloud matching algorithm based on the least squares method (LSM) optimizes the distance between the source point cloud in the pulse pressure radar or electronic scanning radar and the target point cloud in the GPS to ensure accurate alignment of the radar data in the GPS coordinate system.

[0114] In the specific implementation process, after setting the initial alignment, each point in the source point cloud is aligned Find the nearest point corresponding to the target point cloud , and solve the following objective function , to obtain the optimal rotation matrix and translation vectors :

[0115] ;

[0116] in, are the corresponding points in the source point cloud and the target point cloud; the process is iterated continuously until the error converges.

[0117] Example 3

[0118] This embodiment 3 also describes a calibration method for multi-radar data fusion of maritime radar. Except for the following technical features that are different from the above embodiment 1, the remaining technical features of this method can refer to the above embodiment 1.

[0119] In this embodiment, the optimal conversion matrix is ​​calculated based on the point cloud alignment of Procrustes analysis, replacing the above ICP algorithm to realize the conversion of constant pulse pressure radar or electronic scanning radar data to the GPS coordinate system; the specific steps are as follows:

[0120] Point cloud matching algorithm based on Procrustes analysis optimizes the distance between the source point cloud in pulse pressure radar or electronic scanning radar and the target point cloud in GPS to ensure accurate alignment of radar data in GPS coordinate system;

[0121] In the specific implementation process, the source point cloud and the target point cloud are firstly centralized to eliminate the influence of position offset;

[0122] After setting the initial alignment, we find the nearest point in the target point cloud for each point in the source point cloud and use the following objective function Compute the optimal rotation matrix, translation vector, and scale factor:

[0123] ;

[0124] Where R represents the rotation matrix, represents the translation vector, s represents the scale factor, is the corresponding point in the source point cloud and the target point cloud. The process is iterated continuously until the distance difference between the source point cloud and the target point cloud after transformation, i.e., the error converges, to ensure the accurate alignment of the source point cloud and the target point cloud in the GPS coordinate system.

[0125] The alternatives in the above embodiment 2 or 3 are all designed for the application scenarios of maritime radar to ensure accurate coordinate conversion and data fusion in a complex and dynamic marine environment and meet the strict requirements of maritime navigation and positioning.

[0126] The present invention ensures the precise alignment of radar data in the spatial coordinate system by applying the calculation method of the rigid transformation matrix, such as the ICP algorithm, and significantly improves the accuracy and consistency of coordinate transformation. This method not only solves the problem that different radar devices are difficult to fuse due to different output data formats and coordinate systems, but also can maintain a high fusion accuracy in a dynamic and complex maritime environment. In addition, the present invention also optimizes the data processing flow, reduces computational redundancy, and improves the efficiency and real-time performance of the conversion process. This enables the method to be applied in a high-dynamic environment at sea, and can provide higher-precision data support for ship navigation and collision avoidance systems, thereby enhancing the overall safety and reliability of the maritime transportation system.

[0127] In addition, the method of the present invention can also adapt to various types of radar sensors, including pulse pressure radar and electronic scanning radar, and can effectively process the data differences between different sensors, realizing efficient data fusion. This design is not only applicable to a single sensor system, but can also be extended to a fusion system composed of multiple sensors. For example, as the number of sensors increases, the system can integrate data from different radar sources to further improve perception accuracy and reliability. This adaptive design meets the perception needs in complex maritime environments, thereby enhancing the reliability of ship positioning and navigation.

[0128] Of course, the above description is only a preferred embodiment of the present invention, and the present invention is not limited to the above embodiments. It should be noted that all equivalent substitutions and obvious deformation forms made by any technician familiar with the field under the guidance of this specification fall within the essential scope of this specification and should be protected by the present invention.

Claims

1. A calibration method for multi-radar data fusion of a maritime radar system, characterized in that: The steps include: Step 1. Obtain relevant data information of the target ship from pulse pressure radar, electronic scanning radar and global positioning system GPS; Step 2. Extract key fields, convert formats, filter and clean, perform linear interpolation processing, and perform time alignment preprocessing operations on the relevant data information collected by pulse pressure radar, electronic scanning radar, and global positioning system GPS; The key fields extracted from pulse pressure radar and electronic scanning radar include track ID, longitude, latitude, timestamp, ground speed and ground heading information; the key fields extracted from GPS include the ship's longitude, latitude and timestamp data; Step 3. Use timestamps and track IDs to match to ensure that the tracks from different data sources are correctly aligned, ensuring that the data from pulse compression radar, electronic scanning radar and GPS accurately point to the track of the same ship; Step 4. Calculate the best conversion matrix for converting the data of pulse pressure radar and electronic scanning radar to the GPS coordinate system respectively; the best conversion matrix specifically refers to the best rotation matrix R and translation vector t; The improved iterative closest point ICP algorithm is used to calculate the optimal conversion matrix to achieve the conversion of constant pulse pressure radar or electronic scanning radar data to the GPS coordinate system; the specific steps are as follows: Step 4.

1. Initial alignment; Given a source point cloud Q in a pulse compression radar or electronically scanned radar, j ,…,q m } and GPS target point cloud P = {p1,p2,…,p i ,…,p n }, and assume an initial transformation matrix T0 = (R0, t0); q j represents a point in the source point cloud Q, p i represents a point in the target point cloud P, i = 1, 2, ..., n, j = 1, 2, ..., m, m, n represent the number of points in the source point cloud Q and the target point cloud P respectively; R0 represents the initial rotation matrix, t0 represents the initial translation vector; Step 4.

2. Closest point matching; For each point q in the source point cloud j , find the point p closest to it in the target point cloud i , find the corresponding point pair through the nearest neighbor search, that is, match Step 4.

3. Calculate the weighted centroid; By assigning different weights to the points in the source point cloud Q and the target point cloud P, the weighted centroid is calculated to enhance the influence on important points. The calculation formula of the weighted centroid is: Among them, C Q , C p Represent the weighted centroid of the source point cloud and the target point cloud, respectively, w j Represents the midpoint q of the source point cloud Q j The weight, w i Represents the point p in the target point cloud P i The weight of Step 4.

4. Centralize the point cloud; Subtract the weighted centroid from each point in the source point cloud and the target point cloud to obtain the centralized point cloud: in, represents the centralized source point cloud, Represents the centralized target point cloud; Step 4.

5. Calculate the rotation matrix R; First, calculate the covariance matrix H. The calculation formula is as follows: Perform singular value decomposition SVD on the covariance matrix H, and the calculation formula is as follows: H=U∑V T ; Among them, U and V are both column orthogonal matrices; calculate the rotation matrix R, the calculation formula is as follows: R=VU T ; Step 4.

6. Calculate the translation vector t based on the rotation matrix R and the weighted center of mass: t=C P -RC Q ; Step 4.

7. Combine the rotation matrix and translation vector into a new rigid transformation matrix T: Step 4.

8. Use the new rigid transformation matrix T to update the points in the source point cloud, and repeat steps 4.2 to 4.7 until the rigid transformation matrix converges or the set number of iterations is reached; Finally, the optimal rotation matrix R and translation vector t are found by minimizing the following objective function E(R,t): Among them, (q j ,p i ) are the corresponding point pairs in the source point cloud and the target point cloud; Step 5. After obtaining the optimal conversion matrices for converting the pulse compression radar and the electronic scanning radar to the GPS coordinate system, use the two optimal conversion matrices to convert the original latitude and longitude data of the corresponding radars to a unified GPS coordinate system; In step 5, after obtaining the optimal rotation matrix R and translation vector t, the original longitude and latitude data of the pulse pressure radar or the electronic scanning radar are converted to the GPS coordinate system according to the following formula; Where (x 雷达 ,y 雷达 ) represents the original longitude and latitude of the pulse pressure radar or electronic scanning radar, (x GPS ,y GPS ) are the converted GPS coordinates.

2. The calibration method for multi-radar data fusion of a maritime radar system according to claim 1, characterized in that: The step 2 is specifically as follows: Step 2.

1. First, extract key fields from the data collected by pulse pressure radar and electronic scanning radar, including track ID, longitude, latitude, timestamp, ground speed and ground heading information. In the extraction process, a rule-based parsing method is used to extract key information from the original data file using regular expressions and field delimiters, and generate a structured data format; Next, key fields are extracted from the data collected by GPS, including the longitude, latitude and timestamp data of the ship; the data collected by GPS is extracted in the same way as the data collected by pulse pressure radar and electronic scanning radar; Step 2.

2. After extracting the key fields, convert the format of the key field data and unify the timestamps of data from different sources into UTC format; at the same time, filter and clean the incomplete or redundant trajectory data, remove invalid data, and ensure that each trajectory data contains complete key information; to further enhance the comparability of the data, linear interpolation processing is also required to fill in the missing blank points in the ship trajectory and ensure the continuity of the ship trajectory in time; Step 2.

3. After processing in steps 2.1 and 2.2, the data from the pulse compression radar, electronic scanning radar and GPS are time-aligned using an automated time synchronization algorithm to ensure that the ship trajectory data at the same time point can be effectively matched.

3. The calibration method for multi-radar data fusion of a maritime radar system according to claim 2, characterized in that: In step 2.3, the processing process of the automated time synchronization algorithm is as follows: First, timestamp alignment is performed to calibrate the raw data of three different sensors, namely pulse pressure radar, electronic scanning radar and GPS, according to time, and find the closest time point of each sensor at the same moment to ensure that multi-source data can be accurately matched; Since the sampling frequency and time stamp accuracy of different sensors may be different, one of the pulse pressure radar, electronic scanning radar and GPS is selected as the benchmark, and the data of the other two sensors are synchronized according to this benchmark; Secondly, when the timestamps of the sensors cannot be perfectly aligned, linear interpolation is used to fill the time difference by estimating the value of the middle point between the known data points of the ship trajectory data to ensure that the trajectory information remains continuous in time.

4. The calibration method for multi-radar data fusion of a maritime radar system according to claim 1, characterized in that: The step 3 is specifically as follows: Step 3.

1. First, from the ship track data preprocessed in step 2, preliminary screening is performed according to the track ID, and the records with the same track ID are grouped together. The track ID is commonly used to identify the continuous navigation track of the same ship; Step 3.

2. Unify the timestamp formats of the three different data sources: pulse pressure radar, electronic scanning radar and GPS, and set a time window to search for matching records of the three different data sources within the time window; When the timestamp difference from different data sources is within a preset range threshold, the records are considered to be an exact match; If the difference in timestamps from different data sources exceeds the preset range threshold, but is still within the allowable range threshold, further matching is performed in combination with speed and heading parameters to ensure that the matching records belong to the same ship by comparing the actual motion trajectory of the ship; Step 3.

3. After matching the timestamps in step 3.2, the track ID and timestamp are combined to further confirm that the track data from three different data sources, namely, pulse pressure radar, electronic scanning radar and GPS, point to the same ship; In step 3.3, not only the matching of time and track ID is verified, but also the matching result is further verified through multiple conditions including the consistency of latitude and longitude position, the rationality of speed and heading, and abnormal records are removed.

5. The calibration method for multi-radar data fusion of a maritime radar system according to claim 4, characterized in that: The step 3.3 is specifically as follows: The nearest neighbor matching algorithm is used to search for the closest point in the time window, and the changes in speed and heading are combined to ensure the consistency of the matching records in time and space. The multi-condition matching method is used to further verify the reliability of the matching results, that is, to check whether the geographical locations of all matching records at the same time point are consistent, to ensure that the differences in longitude and latitude are within the set tolerance range, and to compare the speed and heading of the matching records, marking any differences that exceed the reasonable range; Records that fail the matching verification will be marked as abnormal records and excluded from the matching; all verified records will be integrated into a unified track data, including track ID, longitude, latitude, timestamp, speed and heading information; At this point, data from different radar and GPS systems are combined to form a complete ship track.

6. The calibration method for multi-radar data fusion of a maritime radar system according to claim 1, characterized in that: In step 4, the improved iterative closest point ICP algorithm is replaced by the least square method LSM; the optimal conversion matrix is ​​calculated based on the least square method LSM to realize the conversion of the constant pulse pressure radar or electronic scanning radar data to the GPS coordinate system; the specific steps are as follows: The point cloud matching algorithm based on the least square method (LSM) optimizes the distance between the source point cloud in the pulse pressure radar or electronic scanning radar and the target point cloud in the GPS to ensure accurate alignment of the radar data in the GPS coordinate system; In the specific implementation process, after setting the initial alignment, each point q in the source point cloud is j Find the nearest point p corresponding to the target point cloud i , and solve the following objective function E(R,t) to obtain the optimal rotation matrix R and translation vector t: Among them, (q j ,p i ) are the corresponding points in the source point cloud and the target point cloud; the process is iterated continuously until the error converges.

7. The calibration method for multi-radar data fusion of a maritime radar system according to claim 1, characterized in that: In step 4, the improved iterative closest point ICP algorithm is replaced by a point cloud alignment method based on Procrustes analysis, and the optimal transformation matrix is ​​calculated based on the point cloud alignment of Procrustes analysis to realize the transformation of constant pulse pressure radar or electronic scanning radar data to the GPS coordinate system; the specific steps are as follows: Point cloud matching algorithm based on Procrustes analysis optimizes the distance between the source point cloud in pulse pressure radar or electronic scanning radar and the target point cloud in GPS to ensure accurate alignment of radar data in GPS coordinate system; In the specific implementation process, the source point cloud and the target point cloud are firstly centralized to eliminate the influence of position offset; After setting the initial alignment, for each point in the source point cloud, find the corresponding nearest point in the target point cloud, and calculate the optimal rotation matrix, translation vector and scale factor through the following objective function E(R,t): Among them, R represents the rotation matrix, t represents the translation vector, s represents the scale factor, (q j ,p i ) are the corresponding points in the source point cloud and the target point cloud; the process is iterated continuously until the distance difference between the source point cloud and the target point cloud after transformation, i.e., the error converges, to ensure the precise alignment of the source point cloud and the target point cloud in the GPS coordinate system.

Images