Radar split point detection method and device, vehicle-mounted terminal and vehicle

Abstract

The embodiment of the invention is suitable for the technical field of computers, and provides a radar split point detection method and device, a vehicle-mounted terminal and a vehicle, and the method comprises the steps: obtaining radar detection information and camera detection information; if the radar detection information meets the split point detection condition, generating a plurality of radar point coordinates and a detection frame coordinate matrix under a vehicle coordinate system constructed by taking the vehicle as an original point based on the coordinate transformation matrix; inputting any two radar point coordinates and the detection frame coordinate matrix into a preset split point detection algorithm to generate split point detection results corresponding to any two radar point coordinates; and performing target fusion on the radar detection information and the camera detection information based on the split point detection result to generate a target detection result. According to the method provided by the embodiment of the invention, the vehicle-mounted terminal performs split point detection on the radar detection information so as to improve the accuracy of target fusion.

Description

Technical Field

[0001] This application belongs to the field of computer technology, and in particular relates to a radar split point detection method, device, vehicle terminal and storage medium. Background Technology

[0002] In autonomous driving technology, accurately acquiring information about vehicles, pedestrians, and other targets on the road often requires sensors such as millimeter-wave radar, cameras, and lidar to obtain multi-source information about the targets. In existing technologies, millimeter-wave radar has relatively low detection accuracy due to limitations in its hardware or software algorithms; for the same target, it often corresponds to two or more radar points on the radar map. Cameras, on the other hand, have higher detection accuracy, typically outputting only one detection box for a single target. Therefore, in a one-to-one target fusion system, when millimeter-wave radar outputs multiple radar points for a single target, directly performing target fusion can easily lead to inaccurate target information in the fused output, or even prevent normal target fusion. Summary of the Invention

[0003] In view of this, embodiments of this application provide a radar split point detection method, apparatus, vehicle-mounted terminal, and vehicle to improve the accuracy of target fusion.

[0004] The first aspect of this application provides a radar split point detection method, including:

[0005] Acquire radar detection information and camera detection information;

[0006] If the radar detection information meets the preset split point detection conditions, then based on the preset coordinate transformation matrix, a matrix of multiple radar point coordinates and detection box coordinates in a vehicle coordinate system constructed with the vehicle as the origin is generated; the radar point coordinates are obtained by transforming the radar detection information based on the coordinate transformation matrix; the detection box coordinate matrix is ​​obtained by transforming the camera detection information based on the coordinate transformation matrix.

[0007] Input the coordinates of any two radar points and the detection box coordinate matrix into a preset split point detection algorithm to generate the split point detection result corresponding to the coordinates of any two radar points.

[0008] Based on the split point detection results, the radar detection information and the camera detection information are fused to generate target detection results.

[0009] In one possible implementation of the first aspect, the split point detection result includes a first detection result and a second detection result;

[0010] The step of inputting the coordinates of any two radar points and the detection box coordinate matrix into a preset split point detection algorithm to generate split point detection results corresponding to the coordinates of the any two radar points includes:

[0011] Obtain the velocity difference and distance difference between the coordinates of any two radar points;

[0012] Input the coordinates of any two radar points and the detection box coordinate matrix into a preset overlap rate algorithm to generate the overlap rate between the coordinates of any two radar points;

[0013] If the velocity difference, the distance difference, and the overlap rate all satisfy the preset splitting conditions, a first detection result is generated; the first detection result is used to indicate that the coordinates of any two radar points are radar detection points of the same target.

[0014] If at least one of the velocity difference, the distance difference, and the overlap rate does not satisfy the splitting condition, a second detection result is generated; the second detection result is used to indicate that the coordinates of any two radar points are radar detection points of different targets.

[0015] In one possible implementation of the first aspect, the splitting condition includes an overlap rate threshold;

[0016] The step of inputting the coordinates of any two radar points and the detection box coordinate matrix into a preset overlap rate algorithm to generate the overlap rate between the coordinates of any two radar points includes:

[0017] Input the coordinates of any two radar points into the first distance formula to determine the offset distance between the coordinates of any two radar points.

[0018] The overlap rate is calculated based on the offset distance and the overlap distance; the overlap distance is the mapping length of the line segment formed by the coordinates of any two radar points within the detection frame corresponding to the detection frame coordinate matrix;

[0019] If the velocity difference, the distance difference, and the overlap rate all satisfy a preset splitting condition, a first detection result is generated, including:

[0020] If the overlap rate is greater than the overlap rate threshold, then the overlap rate is determined to satisfy the splitting condition.

[0021] In one possible implementation of the first aspect, the distance difference includes a first mapped distance difference corresponding to the horizontal axis and a second mapped distance difference corresponding to the vertical axis in the vehicle coordinate system;

[0022] The step of obtaining the velocity difference and distance difference between any two radar point coordinates includes:

[0023] Input the coordinates of any two radar points into the second distance formula to determine the first mapping distance difference and the second mapping distance difference;

[0024] If the velocity difference, the distance difference, and the overlap rate all satisfy a preset splitting condition, a first detection result is generated, including:

[0025] The target length and target width are obtained based on the coordinate matrix of the detection frame.

[0026] A width threshold is determined based on the target width and the overlap rate;

[0027] If the second mapping distance difference is less than the target length and the first mapping distance difference is less than the width threshold, then the distance difference is determined to satisfy the splitting condition.

[0028] In one possible implementation of the first aspect, the split point detection result includes a second detection result; the second detection result is used to indicate that the coordinates of any two radar points are radar detection points of different targets;

[0029] The step of fusing the radar detection information and the camera detection information based on the split point detection results to generate target detection results includes:

[0030] When the split point detection result is the second detection result, for any radar point coordinate, the radar point coordinate is input into a preset first similarity algorithm to determine the angular similarity between the radar point coordinate and the detection box coordinate matrix;

[0031] For any radar point coordinate, the radar point coordinate is input into a preset second similarity algorithm to determine the distance similarity between the radar point coordinate and the detection box coordinate matrix;

[0032] Based on the angle similarity and the distance similarity, the target similarity between the coordinates of any radar point and the detection box coordinate matrix is ​​determined;

[0033] If the target similarity satisfies the preset similarity condition, then the radar point coordinates corresponding to the target similarity that satisfies the similarity condition are taken as the target radar point corresponding to the detection box coordinate matrix;

[0034] The radar detection information corresponding to the target radar point is fused with the camera detection information to generate the target detection result.

[0035] In one possible implementation of the first aspect, the step of inputting the radar point coordinates into a preset first similarity algorithm to generate an angular similarity between the radar point coordinates and the detection box coordinate matrix for any radar point coordinates includes:

[0036] The detection box coordinate matrix is ​​input into a preset first angle function to generate a first angle value; the first angle value is used to represent the maximum angle between each reference point of the detection box corresponding to the detection box coordinate matrix; the reference point is a vertex or the midpoint of an edge of the detection box;

[0037] Input the radar point coordinates and any reference point into a preset second angle function to generate a second angle value;

[0038] Based on the first angle value and the second angle value, generate the angle similarity between the radar point coordinates and any reference point;

[0039] For any radar point coordinate, the radar point coordinates are input into a preset second similarity algorithm to generate a distance similarity between the radar point coordinates and the detection box coordinate matrix, including:

[0040] The radar point coordinates are input into a preset third distance formula to generate the radar point distance between the radar point coordinates and the vehicle.

[0041] Input the coordinates of any reference point and the radar point coordinates into the fourth distance formula to generate the projected distance; the projected distance is the distance between the projected point corresponding to the radar point coordinates and the vehicle; the projected point is the point formed when the radar point coordinates are projected onto the line connecting any reference point and the origin;

[0042] The distance similarity is generated based on the radar point distance and the projected distance;

[0043] The generation of target similarity based on the angle similarity and the distance similarity includes:

[0044] The sum of the angle similarity and the distance similarity is taken as the reference point similarity between the coordinates of any reference point and the radar point.

[0045] If the similarity of the reference points meets the preset reference conditions, then the similarity of the reference points that meets the reference conditions is determined as the target similarity.

[0046] In one possible implementation of the first aspect, the split point detection result includes a first detection result; the first detection result is used to indicate that the coordinates of any two radar points are radar detection points of the same target.

[0047] The step of fusing the radar detection information and the camera detection information based on the split point detection results to generate target detection results includes:

[0048] Obtain the actual distance between the coordinates of each radar point and the vehicle;

[0049] The radar point with the smallest actual distance is determined as the target radar point corresponding to the target object;

[0050] The radar detection information corresponding to the target radar point is fused with the camera detection information to generate the target detection result.

[0051] A second aspect of this application provides a radar split point detection device, comprising:

[0052] The information acquisition module is used to acquire radar detection information and camera detection information;

[0053] The conversion module is used to generate multiple radar point coordinates and detection box coordinate matrices in a vehicle coordinate system constructed with the vehicle as the origin, based on a preset coordinate transformation matrix, if the radar detection information meets preset split point detection conditions; the radar point coordinates are obtained by converting the radar detection information based on the coordinate transformation matrix; the detection box coordinate matrix is ​​obtained by converting the camera detection information based on the coordinate transformation matrix.

[0054] The split detection module is used to input the coordinates of any two radar points and the detection frame coordinate matrix into a preset split point detection algorithm to generate the split point detection result corresponding to the coordinates of the any two radar points.

[0055] The target fusion module is used to fuse the radar detection information and the camera detection information based on the split point detection results to generate target detection results.

[0056] A third aspect of this application provides an in-vehicle terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the radar split point detection method as described in the first aspect above.

[0057] A fourth aspect of this application provides a vehicle, characterized in that it includes a millimeter-wave radar, a camera, and an in-vehicle terminal; the in-vehicle terminal is communicatively connected to the millimeter-wave radar and the camera respectively;

[0058] The millimeter-wave radar is used to acquire radar detection information;

[0059] The camera is used to acquire camera detection information;

[0060] The vehicle-mounted terminal is used to execute the radar split point detection method as described in the first aspect above, based on the radar detection information and the camera detection information.

[0061] A fifth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the radar split point detection method as described in the first aspect above.

[0062] A sixth aspect of this application provides a computer program product that, when run on a computer, causes the computer to execute the radar split point detection method described in the first aspect.

[0063] Compared with the prior art, the embodiments of this application have the following advantages:

[0064] In this embodiment, the vehicle-mounted terminal can acquire radar detection information and camera detection information, and determine whether the acquired radar detection information meets the split point detection conditions preset by the developer. If the vehicle-mounted terminal determines that the radar detection information meets the split point detection conditions, it can generate multiple radar point coordinates and detection box coordinate matrices in a vehicle coordinate system constructed with the vehicle as the origin, based on the coordinate transformation matrix preset by the developer. The radar point coordinates generated by the vehicle-mounted terminal are obtained by transforming the radar detection information based on the coordinate transformation matrix. The detection box coordinate matrix generated by the vehicle-mounted terminal is obtained by transforming the camera detection information based on the coordinate transformation matrix. After generating the detection box coordinate matrix and radar point coordinates, the vehicle-mounted terminal can input any two radar point coordinates and the detection box coordinate matrix into the split point detection algorithm preset by the developer to generate the split point detection result corresponding to the above two radar point coordinates. Finally, the vehicle-mounted terminal can perform target fusion on the radar detection information and camera detection information based on the split point detection result to generate the target detection result. In this embodiment, since the vehicle terminal can perform split point detection on the coordinates of any two radar points according to the split point detection algorithm, the accuracy of target fusion of radar detection information and camera detection information can be improved, so as to generate more accurate target detection results. Attached Figure Description

[0065] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0066] Figure 1 This is a schematic diagram of a vehicle provided in an embodiment of this application;

[0067] Figure 2 This is a schematic diagram of a radar split point detection method provided in an embodiment of this application;

[0068] Figure 3 This is a schematic diagram of another radar split point detection method provided in an embodiment of this application;

[0069] Figure 4 This is a schematic diagram of a first mapping distance difference and a second mapping distance difference provided in an embodiment of this application;

[0070] Figure 5 This is a schematic diagram of an offset distance and overlap distance provided in an embodiment of this application;

[0071] Figure 6 This is a schematic diagram of another radar split point detection method provided in an embodiment of this application;

[0072] Figure 7 This is a schematic diagram illustrating the calculation of angular similarity and distance similarity provided in an embodiment of this application;

[0073] Figure 8 This is a schematic diagram of a radar split point detection device provided in an embodiment of this application;

[0074] Figure 9 This is a schematic diagram of an in-vehicle terminal provided in an embodiment of this application. Detailed Implementation

[0075] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0076] The technical solution of this application will be described below through specific embodiments.

[0077] Reference Figure 1 The diagram illustrates a vehicle according to an embodiment of this application. Figure 1As shown, vehicle 1 may include a millimeter-wave radar 11, a camera 12, and an on-board terminal 13. The first signal output port of the millimeter-wave radar 11 can be communicatively connected to the first signal input port of the on-board terminal 13. The second signal output port of the camera 12 can be communicatively connected to the first signal input port of the on-board terminal 13. When the vehicle is running, the millimeter-wave radar 11 can acquire radar detection information and send the acquired radar detection information to the on-board terminal 13 through its first signal output port. The camera 12 can acquire camera detection information and send the acquired camera detection information to the on-board terminal 13 through its second signal output port. The on-board terminal 13 can execute the radar split point detection method as described in any of the following embodiments based on the received radar detection information and camera detection information.

[0078] Reference Figure 2 This diagram illustrates a radar split-point detection method provided in an embodiment of this application. This method can be used in an in-vehicle terminal. The in-vehicle terminal can be an Electronic Control Unit (ECU), Microcontroller Unit (MCU), Central Processing Unit (CPU), Automobile Engine Control Module (ECM), or other similar in-vehicle terminals. Specifically, the method may include the following steps:

[0079] S201. Acquire radar detection information and camera detection information.

[0080] In this embodiment, when the driver needs to use the vehicle's autonomous driving function, the driver can activate the function via the autonomous driving button on the vehicle. The driver can also activate the autonomous driving function by clicking the autonomous driving control on the vehicle's central control screen. After the driver activates the autonomous driving function, the onboard terminal can acquire radar detection information of the vehicle's surrounding environment via the vehicle's millimeter-wave radar and acquire environmental images of the surrounding environment via a camera installed on the vehicle. After acquiring the environmental images, the onboard terminal can input them into a target detection algorithm pre-set by the developers to generate camera detection information corresponding to the environmental images. The camera detection information acquired by the onboard terminal can include detection boxes corresponding to each target object in the environmental image and the type of each target object. For example, the type of target object in the camera detection information can be a small vehicle, a truck, a bicycle, a pedestrian, etc. The radar detection information acquired by the onboard terminal can include multiple radar detection points, as well as the distance, speed, and angle corresponding to each radar detection point. The acquisition time of the radar monitoring information and the camera detection information can be the same. The acquisition time of each radar detection point in the radar detection information can also be the same. For example, when there is a target A to the left rear of the vehicle, the radar detection information obtained by the vehicle terminal may include radar detection point 1 and radar detection point 2 to the left rear of the vehicle. As another example, when there is a target A to the left rear of the vehicle and a target B to the right rear of the vehicle, the radar detection information obtained by the vehicle terminal may include radar detection point 1 to the left rear of the vehicle and radar detection point 2 to the right rear of the vehicle.

[0081] S202. If the radar detection information meets the preset split point detection conditions, then based on the preset coordinate transformation matrix, a matrix of multiple radar point coordinates and detection box coordinates in a vehicle coordinate system constructed with the vehicle as the origin is generated; the radar point coordinates are obtained by transforming the radar detection information based on the coordinate transformation matrix; the detection box coordinate matrix is ​​obtained by transforming the camera detection information based on the coordinate transformation matrix.

[0082] In this embodiment, after acquiring radar detection information, the vehicle terminal can determine whether the acquired radar detection information meets the split point detection conditions preset by the developers. If the vehicle terminal determines that the radar detection information meets the split point detection conditions, it can perform split point detection based on the radar detection information and camera detection information to determine whether there are multiple radar detection points corresponding to the same target object in the radar detection information, that is, to determine whether there are multiple radar detection points split from the same target object in the radar detection information. At this time, the vehicle terminal can first determine the camera detection information within the angle range where the multiple radar detection points that meet the split conditions are located based on the angles corresponding to each radar detection point in the radar detection information, and then perform split point detection on the multiple radar detection points that meet the split conditions based on the camera detection information within the angle range. For example, when multiple radar detection points that meet the split conditions in the radar detection information are all located within the range of 45 degrees from the left front to the right front of the vehicle, the vehicle terminal can acquire the camera detection information of the front camera of the vehicle to perform split point detection on the multiple radar detection points within the range.

[0083] The vehicle-mounted terminal can perform coordinate transformation on radar and camera detection information using a coordinate transformation matrix pre-set by developers. This coordinate transformation matrix can be set by developers based on the relationships between the pixel coordinate system, camera coordinate system, and vehicle coordinate system. The vehicle-mounted terminal can generate multiple radar point coordinates and detection box coordinate matrices within a vehicle coordinate system constructed with the vehicle as the origin. The radar point coordinates are obtained by transforming each radar detection point in the radar detection information based on the coordinate transformation matrix and the corresponding distances and angles. The detection box coordinate matrix is ​​obtained by transforming vehicle-type detection boxes in the camera detection information using the coordinate transformation matrix.

[0084] In one possible implementation, the vehicle-mounted terminal can determine whether the radar detection information meets the split-point detection condition by the total number of radar points in the radar detection information. When the vehicle-mounted terminal determines that the total number of radar points is greater than or equal to a pre-set threshold by the developer, the vehicle-mounted terminal can determine that the radar detection information meets the split-point detection condition. When the vehicle-mounted terminal determines that the total number of radar points is less than the pre-set threshold by the developer, the vehicle-mounted terminal can determine that the radar detection information does not meet the split-point detection condition. For example, the threshold can be 2.

[0085] In one possible implementation, the vehicle-mounted terminal can determine whether the radar detection information meets the split-point detection condition by combining the total number of detection frames in the camera detection information and the total number of radar points in the radar detection information. When the vehicle-mounted terminal determines that the total number of radar points is greater than the total number of detection frames and the difference between the total number of radar points and the total number of detection frames is greater than or equal to a pre-set threshold, the vehicle-mounted terminal can determine that the radar detection information meets the split-point detection condition. When the vehicle-mounted terminal determines that the total number of radar points is less than or equal to the total number of detection frames, or when the vehicle-mounted terminal determines that the total number of radar points is greater than the total number of detection frames and the difference between the total number of radar points and the total number of detection frames is less than a pre-set threshold, the vehicle-mounted terminal can determine that the radar detection information does not meet the split-point detection condition.

[0086] In one possible implementation, before inputting the radar point coordinates and the detection box coordinate matrix into the split point detection algorithm, the vehicle terminal can perform a preliminary detection of the two radar detection points by determining whether the line connecting the two radar detection points intersects with the detection box. This preliminary detection determines whether the two radar detection points need to undergo split point detection, i.e., whether they meet the split point detection conditions. If the vehicle terminal determines that there is an intersection between the line connecting the two radar point coordinates and the rectangle formed by the detection box coordinate matrix, then the vehicle terminal can determine that the two radar detection points need to undergo split point detection, i.e., they meet the split point detection conditions. If the vehicle terminal determines that there is no intersection between the line connecting the two radar point coordinates and the rectangle formed by the detection box coordinate matrix, then the vehicle terminal can determine that the two radar detection points do not need to undergo split point detection, and the vehicle terminal can directly generate the split point detection result of the two radar point coordinates based on the relationship between the two radar point coordinates and the detection box coordinate matrix.

[0087] When the vehicle terminal determines that there is no intersection between the line connecting the coordinates of the two current radar points and the rectangle formed by the detection frame coordinate matrix, the vehicle terminal can determine whether the coordinates of the two current radar points are within the rectangle formed by the detection frame coordinate matrix.

[0088] If the vehicle-mounted terminal determines that the coordinates of the two current radar points are both within the rectangle formed by the detection frame coordinate matrix, then the vehicle-mounted terminal can determine that the radar detection points corresponding to the current two radar point coordinates are two points split from the same target object, that is, the vehicle-mounted terminal can determine that the two radar coordinates satisfy the first detection result. At this time, the vehicle-mounted terminal can perform target fusion on the radar detection information and the camera detection information according to the method corresponding to the first detection result. For the specific process of the vehicle-mounted terminal performing target fusion according to the first detection result, please refer to the content in S204 of this embodiment, which will not be repeated here.

[0089] If the vehicle-mounted terminal determines that the coordinates of the two current radar points are both outside the rectangle formed by the detection frame coordinate matrix, then the vehicle-mounted terminal can determine that the radar detection points corresponding to the current two radar point coordinates correspond to different targets, that is, the vehicle-mounted terminal can determine that the two radar coordinates satisfy the second detection result. At this time, the vehicle-mounted terminal can perform target fusion on the radar detection information and the camera detection information according to the method corresponding to the second detection result. For the specific process of the vehicle-mounted terminal performing target fusion according to the second detection result, please refer to the content of the third embodiment of this application, which will not be repeated here.

[0090] In this embodiment, when the line connecting the coordinates of two radar points does not intersect with the detection box coordinate matrix, the vehicle terminal can directly generate the split point detection result between the coordinates of the two radar points based on the positional relationship between the coordinates of the two radar points and the detection box coordinate matrix. Therefore, the method provided in this embodiment can reduce the computational load of the system and improve the split point detection efficiency.

[0091] S203. Input the coordinates of any two radar points and the detection frame coordinate matrix into a preset split point detection algorithm to generate the split point detection result corresponding to the coordinates of the any two radar points.

[0092] In this embodiment, after the vehicle terminal obtains the radar point coordinates corresponding to multiple radar detection information that meet the preset split point detection conditions, and the detection frame coordinate matrix corresponding to the camera detection information acquired at the same time as the radar detection information, it can input any two radar point coordinates and detection frame coordinate matrices that meet the preset split point detection conditions into the split point detection algorithm pre-set by the developers to generate the split point detection result corresponding to the above two radar point coordinates.

[0093] In one possible implementation, before inputting the radar point coordinates and the detection box coordinate matrix into the split point detection algorithm, the vehicle-mounted terminal can perform a preliminary detection of the two radar detection points by checking whether there is an intersection between the line connecting the two radar detection points and the detection box. For details regarding the preliminary detection of the radar detection points by the vehicle-mounted terminal, please refer to S202 of this embodiment; it will not be repeated here.

[0094] S204. Based on the split point detection results, the radar detection information and the camera detection information are fused to generate target detection results.

[0095] In this embodiment, after generating the split point detection results between all pairs of radar detection points in the radar detection information, the vehicle-mounted terminal can perform target fusion on the radar detection information and camera detection information based on the generated split point detection results to generate target detection results. The target detection results generated by the vehicle-mounted terminal after target fusion may include the detection bounding box, type, angle, speed, and distance of the target object.

[0096] In this embodiment, when the radar detection information meets the split detection conditions, the vehicle terminal performs split point detection using a split point detection algorithm. Since this embodiment does not limit the specific type of millimeter-wave radar, this solution has strong versatility and can adapt to targets input from different sensors, realizing platform-based development of target fusion. Furthermore, using the split point detection algorithm avoids the vehicle terminal fusing camera detection information with split points, improving the accuracy of target fusion and thus providing accurate decision-making information for autonomous driving decisions.

[0097] In one possible implementation, the split point detection results generated by the vehicle-mounted terminal can include a first detection result and a second detection result. The first detection result can be used to indicate that the current two radar point coordinates are two radar detection points split from the same target object. The second detection result can be used to indicate that the current two radar point coordinates are two radar detection points corresponding to different targets.

[0098] When the split point detection result generated by the vehicle terminal is the first detection result, that is, when the two current radar detection points are split from the same target object, the vehicle terminal can determine the actual distance between the two radar detection points and the vehicle based on the radar point coordinates corresponding to the two radar detection points respectively. The vehicle terminal can confirm the radar point coordinates with the smallest actual distance as the target coordinates corresponding to the detection frame coordinate matrix, and determine the radar detection point corresponding to the radar point coordinates with the smallest actual distance as the target radar point corresponding to the target object in the camera detection information. After determining the target radar point corresponding to the target object in the camera detection information, the vehicle terminal can perform target fusion with the radar detection information corresponding to the target radar point and the camera detection information to generate a target detection result. The vehicle terminal can perform target fusion with the radar detection information corresponding to the target radar point and the camera detection information using any target fusion algorithm known to those skilled in the art. This application embodiment is not intended to specifically limit the target fusion method.

[0099] In this embodiment, when two radar detection points are points split from the same target, the closer the distance, the higher the accuracy of radar detection. Therefore, in this embodiment, using the radar point coordinates with the smallest actual distance as the target radar point corresponding to the target can effectively improve the accuracy of vehicle target fusion.

[0100] When the split point detection result between two radar point coordinates is the second detection result, that is, when the two are two radar detection points corresponding to different targets, the vehicle terminal can execute the method provided in the second embodiment below to determine the target radar point from the two radar detection points. For details, please refer to the content in the third embodiment of this application, which will not be repeated here.

[0101] Figure 3 A flowchart illustrating the specific implementation of a radar split-point detection method S203 provided in the second embodiment of this application is shown. See also... Figure 3 Compared to Figure 2 In the embodiment provided, S203 of the radar split point detection method includes: S2031 to S2034, which are detailed below:

[0102] S2031. Obtain the velocity difference and distance difference between the coordinates of any two radar points.

[0103] In this embodiment, the vehicle-mounted terminal can determine whether two radar detection points are points split from the same target object based on the speed difference and distance difference between them. First, the speed difference and distance difference between the coordinates of the two radar points are obtained. Specifically, the vehicle-mounted terminal can calculate the speed difference between the coordinates of the two radar points based on the speeds of the radar detection points corresponding to the coordinates of the two radar points in the radar detection information.

[0104] In one possible implementation, the distance difference calculated by the vehicle-mounted terminal may include a first mapped distance difference and a second mapped distance difference. The first mapped distance difference can be the difference between the coordinates of the two radar points along the horizontal axis of the vehicle coordinate system, i.e., the lateral distance difference between the two radar point coordinates. The second mapped distance difference can be the difference between the coordinates of the two radar points along the vertical axis of the vehicle coordinate system, i.e., the longitudinal distance difference between the two radar point coordinates. The vehicle-mounted terminal can input the coordinates of the two radar points into a second distance formula to determine the first mapped distance difference and the second mapped distance difference between the coordinates of the two radar points.

[0105] In one possible implementation, the second distance formula can be as follows.

[0106] D1=|d x1 -d x2 |

[0107] D2=|d y1 -d y2 |

[0108] Where D1 can represent the first mapping distance difference, and D2 can represent the second mapping distance difference. x1 This can represent the lateral distance between the first radar point and the vehicle, i.e., the x-coordinate value of the first radar point. x2 This can represent the lateral distance between the second radar point and the vehicle, i.e., the x-coordinate value of the second radar point. y1 This can represent the longitudinal distance between the first radar point and the vehicle, i.e., the ordinate value of the first radar point. y2 This can represent the longitudinal distance between the second radar point coordinate and the vehicle, i.e., the longitudinal coordinate value of the second radar point coordinate.

[0109] See Figure 4 This diagram illustrates a first mapping distance difference and a second mapping distance difference provided in an embodiment of this application. Figure 4 As shown, the difference between the first radar detection point and the second radar detection point along the horizontal axis is the first mapping distance difference D1 between the two radar detection points. The difference between the first radar detection point and the second radar detection point along the vertical axis is the second mapping distance difference D2 between the two radar detection points.

[0110] S2032. Input the coordinates of any two radar points and the detection frame coordinate matrix into a preset overlap rate algorithm to generate the overlap rate between the coordinates of any two radar points.

[0111] In this embodiment, the vehicle-mounted terminal can also determine whether two radar detection points are points split from the same target object based on the overlap rate between the two radar detection points and the detection frame. The vehicle-mounted terminal can input the radar point coordinates corresponding to the two radar detection points that need to be detected and the detection frame coordinate matrix corresponding to the detection frame into the overlap rate algorithm pre-set by the developers, so as to calculate the overlap rate between the two radar point coordinates and the detection frame coordinate matrix through the overlap rate algorithm.

[0112] In one possible implementation, the vehicle terminal can input the coordinates of the two radar points into a first distance formula pre-set by the developer to determine the offset distance between the coordinates of the two radar points.

[0113] In one possible implementation, the first distance formula can be as follows.

[0114]

[0115] Where L1 represents the offset distance between the coordinates of two radar points. x1 represents the x-coordinate of the first radar point. x2 represents the x-coordinate of the second radar point. y1 represents the y-coordinate of the first radar point. y2 represents the y-coordinate of the second radar point.

[0116] In one possible implementation, the vehicle-mounted terminal can also calculate the overlap distance based on the coordinates of the two radar points and the detection frame coordinate matrix. The overlap distance calculated by the vehicle-mounted terminal can be the length of the line segment formed by the coordinates of the two radar points within the detection frame corresponding to the detection frame coordinate matrix. Specifically, the terminal device can generate a line segment function corresponding to the line connecting the two radar point coordinates. Then, the vehicle-mounted terminal can jointly solve the line segment function and the detection frame coordinate matrix to find the intersection coordinates of the line connecting the two radar point coordinates and the detection frame. The vehicle-mounted terminal can then input the radar point coordinates within the detection frame and the intersection coordinates into a fifth distance formula to calculate the overlap distance. The fifth distance formula can be any formula known to those skilled in the art for calculating the length of a line segment within the detection frame; this embodiment is not intended to specifically limit the fifth distance formula.

[0117] like Figure 5 The diagram illustrates an offset distance and overlap distance provided in an embodiment of this application. See also... Figure 5 The vehicle coordinate system may include a detection frame, a first radar detection point located within the detection frame, and a second radar detection point located outside the detection frame. The detection frame can be represented in the vehicle coordinate system as a detection frame coordinate matrix. The first radar detection point can be represented in the vehicle coordinate system as the coordinates of a first radar point. The second radar detection point can be represented in the vehicle coordinate system as the coordinates of a second radar point. The length of the line segment connecting the first and second radar points can be the offset distance between the first and second radar points. The length of the line segment connecting the first and second radar points within the detection frame can be the overlap distance between the two radar points and the detection frame.

[0118] After calculating the offset distance and overlap distance, the vehicle terminal can input the offset distance and overlap distance into the overlap rate calculation formula preset by the developers to calculate the overlap rate.

[0119] In one possible implementation, the overlap rate calculation formula can be as follows.

[0120]

[0121] Here, overlap represents the overlap rate between the line connecting the coordinates of two radar points and the detection box coordinate matrix. L2 represents the overlap distance of the line segment formed by the coordinates of two radar points within the detection box corresponding to the detection box coordinate matrix. L1 represents the offset distance between the coordinates of two radar points.

[0122] S2033. If the velocity difference, the distance difference, and the overlap rate all satisfy the preset splitting conditions, a first detection result is generated; the first detection result is used to indicate that the coordinates of any two radar points are radar detection points of the same target.

[0123] In this embodiment, after determining the speed difference, distance difference, and overlap rate, the vehicle-mounted terminal can determine whether each of these factors satisfies the splitting conditions preset by the developers. If the vehicle-mounted terminal determines that all three factors satisfy the splitting conditions, it can generate a first detection result. This first detection result can be used to indicate that the coordinates of the two radar points are radar detection points split from the same target object. For the specific process of target fusion based on the first detection result, please refer to section S204 of this embodiment; it will not be repeated here.

[0124] In one possible implementation, the vehicle-mounted terminal can determine whether the speed difference meets the splitting condition by judging whether the speed difference is less than or equal to a speed threshold preset by the developers. If the vehicle-mounted terminal determines that the speed difference between the two radar point coordinates is less than or equal to the speed threshold, then the vehicle-mounted terminal can determine that the speed difference meets the splitting condition. If the vehicle-mounted terminal determines that the speed difference between the two radar point coordinates is greater than the speed threshold, then the vehicle-mounted terminal can determine that the speed difference does not meet the splitting condition.

[0125] In this embodiment, since millimeter-wave radar can accurately measure the velocity of a target, when the velocity difference between two radar point coordinates is too large, the two radar point coordinates are likely not radar points from the same target split. Therefore, the velocity difference can accurately determine whether two radar point coordinates are radar points from the same target split, and the method provided in this embodiment can ensure the accuracy of split point detection.

[0126] In one possible implementation, the vehicle-mounted terminal can determine whether the overlap rate meets the splitting condition by judging whether the overlap rate is greater than or equal to an overlap threshold preset by the developers. If the vehicle-mounted terminal determines that the overlap rate is greater than or equal to the overlap threshold, it can determine that the overlap rate meets the splitting condition. If the vehicle-mounted terminal determines that the overlap rate is less than the overlap threshold, it can determine that the overlap rate does not meet the splitting condition.

[0127] In this embodiment, since the detection box in the camera detection information can accurately represent the shape of the target object, when the overlap rate between the line segment formed by the coordinates of two radar points and the coordinate matrix of the detection box is small, that is, when the shape of the object formed by the coordinates of the two radar points differs greatly from the actual shape of the target object, the two radar point coordinates are likely not radar points split from the same target object. Therefore, split point detection by overlapping rate improves the judgment accuracy of the vehicle terminal.

[0128] In one possible implementation, the distance difference calculated by the vehicle-mounted terminal can include a first mapped distance difference corresponding to the horizontal axis and a second mapped distance difference corresponding to the vertical axis. After calculating the first and second mapped distance differences, the vehicle-mounted terminal can obtain the target length and target width corresponding to the target object based on the detection box coordinate matrix. The target length can be the length of the detection box corresponding to the detection box coordinate matrix, and the target width can be the width of the detection box corresponding to the detection box coordinate matrix. After calculating the target width, the vehicle-mounted terminal can also calculate a width threshold based on the target width and the overlap rate.

[0129] In one possible implementation, the formula for calculating the width threshold can be as follows.

[0130] η = overlap * width

[0131] Where η can represent the width threshold, overlap can represent the overlap rate, and width can represent the target width.

[0132] After calculating the target length and width thresholds, the vehicle-mounted terminal can determine whether the difference in the second mapped distance between the two radar point coordinates on the vertical axis is less than the target length, and whether the difference in the first mapped distance between the two radar point coordinates on the horizontal axis is less than the width threshold. If the vehicle-mounted terminal determines that the difference in the second mapped distance is less than the target length and the difference in the first mapped distance is less than the width threshold, then the vehicle-mounted terminal can determine that the distance difference meets the splitting condition. If the vehicle-mounted terminal determines that the difference in the second mapped distance is greater than or equal to the target length, and / or the difference in the first mapped distance is greater than or equal to the width threshold, then the vehicle-mounted terminal can determine that the distance difference does not meet the splitting condition.

[0133] In this embodiment, since the detection box in the camera detection information can accurately represent the size of the target object, when the longitudinal distance between two radar point coordinates is greater than the length of the detection box or the lateral distance is greater than the width of the detection box, the two radar point coordinates are likely not radar points split from the same target object. Therefore, further combining the first mapping distance difference and the second mapping distance difference for split point detection can improve the accuracy of the vehicle terminal's judgment. In addition, in this embodiment, the vehicle terminal calculates the width threshold by combining the overlap rate, which can effectively avoid misidentifying nearby guardrail targets as split points, thereby increasing the accuracy of split point detection. As for the length, since there are few other obstacles at close range in the front and rear directions during vehicle operation, the vehicle terminal can directly use the target length of the detection box as the length threshold for split point judgment.

[0134] S2034. If at least one of the velocity difference, the distance difference, and the overlap rate does not satisfy the splitting condition, a second detection result is generated; the second detection result is used to indicate that the coordinates of any two radar points are radar detection points of different targets.

[0135] In this embodiment, if the vehicle-mounted terminal determines that any one or more of the speed difference, distance difference, and overlap rate do not meet the splitting condition, the vehicle-mounted terminal can generate a second detection result. This second detection result can be used to indicate that the two radar points represent radar detection points for different targets.

[0136] In this embodiment, the vehicle terminal performs split point detection by combining three aspects: speed, distance, and overlap rate. This ensures the detection accuracy of the vehicle terminal, thereby improving the accuracy of target fusion and providing a guarantee for the safety of autonomous driving functions.

[0137] Figure 6 A flowchart illustrating the specific implementation of a radar split-point detection method S204 provided in the third embodiment of this application is shown. See also... Figure 6 Compared to Figure 2 In the embodiment provided, S203 of the radar split point detection method includes: S601 to S605, which are detailed below:

[0138] S601. When the split point detection result is the second detection result, for any radar point coordinate, the radar point coordinate is input into a preset first similarity algorithm to determine the angular similarity between the coordinate of any radar point and the detection box coordinate matrix.

[0139] In this embodiment, when the vehicle terminal determines that the split point detection result between the current radar point coordinates is the second detection result, that is, when the vehicle terminal can determine that the current two radar point coordinates are two radar detection points corresponding to different targets, the vehicle terminal can calculate the target similarity between each radar point coordinate in the second detection result and each detection box coordinate matrix to match the target radar point corresponding to each detection box coordinate matrix. For any radar point coordinate that needs to be calculated for similarity, the terminal device can input the radar point coordinate into the first similarity algorithm pre-set by the developers to determine the angular similarity corresponding to the radar point coordinate through the first similarity algorithm.

[0140] In one possible implementation, the vehicle-mounted terminal can input the detection box coordinate matrix into a first angle function pre-set by the developer to generate a first angle value. This first angle value represents the maximum angle between reference points of the detection box corresponding to the detection box coordinate matrix. Reference points in the detection box can be vertices or midpoints of edges. Then, the vehicle-mounted terminal can input the radar point coordinates for which similarity calculation is needed and any reference point in the detection box into a second angle function pre-set by the developer to generate a second angle value. This second angle value is the angle between the radar point coordinates and the current reference point. After calculating the first and second angle values, the vehicle-mounted terminal can generate an angular similarity score between the radar point coordinates and the current reference point based on these scores.

[0141] In one possible implementation, the formula for calculating angular similarity can be as follows.

[0142]

[0143] Here, δ can represent angular similarity. It can represent the maximum angle between each reference point of the detection box corresponding to the detection box coordinate matrix, that is, the first angle value. The angle similarity between the radar point coordinates and the current reference point can be represented by the second angle value. The first and second angle values ​​are divided to eliminate the influence of dimensions, making the angle similarity between the two points independent of the measurement unit of the original data, i.e., independent of the measurement scale. This is equivalent to a normalization operation. Therefore, the method provided in this embodiment can simply and quickly calculate the angle similarity.

[0144] See Figure 7 This diagram illustrates a calculation schematic of angular similarity and distance similarity provided in an embodiment of this application. Figure 7As shown, A, B, and C are all reference points on the detection box, the origin O is the vehicle, F is the radar detection point corresponding to the radar point coordinates that need to be calculated for similarity, and H is the projection point of radar detection point F on the line connecting reference point B and origin O. Therefore, the first angle value corresponding to the detection box in the figure can be ∠AOC, and the second angle value between radar detection point F and reference point B can be ∠FOB.

[0145] In this embodiment, when a radar detection point is the target radar point corresponding to the detection frame, the angle between the radar monitoring point and any reference point on the detection frame should be less than the maximum angle between the reference points in the detection frame. Therefore, through the method provided in this embodiment, the vehicle terminal can accurately determine the similarity between the radar detection point and the detection frame based on the angle similarity, thereby improving the accuracy of the target radar point selection.

[0146] S602. For any radar point coordinate, input the radar point coordinate into a preset second similarity algorithm to determine the distance similarity between the radar point coordinate and the detection box coordinate matrix.

[0147] In this embodiment, for any radar point coordinates that require similarity calculation, the terminal device can input the radar point coordinates into a second similarity algorithm pre-set by the developers, so as to determine the distance similarity corresponding to the radar point coordinates through the second similarity algorithm.

[0148] In one possible implementation, the vehicle terminal can input the aforementioned radar point coordinates into a third distance formula pre-set by the developers to generate the radar point distance between the radar point coordinates and the vehicle.

[0149] In one possible implementation, the third distance formula can be as follows.

[0150]

[0151] Where D3 can be the distance to the radar point, x can be the x-coordinate of the radar point's coordinates, and y can be the y-coordinate of the radar point's coordinates.

[0152] The vehicle-mounted terminal can also input the coordinates of the reference point and the radar point into the fourth distance formula to calculate the projected distance. The projected distance calculated by the vehicle-mounted terminal can be the distance between the projected point corresponding to the radar point coordinates and the vehicle. The projected point corresponding to the radar point coordinates can be the point formed when the radar point coordinates are projected onto the line connecting the current reference point and the origin.

[0153] After calculating the radar point distance and the projected distance, the vehicle-mounted terminal can generate the distance similarity based on these distances. The specific formula for calculating the distance similarity is as follows.

[0154]

[0155] Where α represents the distance similarity between the radar point coordinates and the reference point. D3 represents the radar point distance between the radar point coordinates and the vehicle. D4 represents the projected distance of the radar point coordinates.

[0156] See Figure 7 This diagram illustrates a calculation schematic of angular similarity and distance similarity provided in an embodiment of this application. Figure 7 As shown, A, B, and C are all reference points on the detection frame, O is the vehicle, F is the radar detection point corresponding to the radar point coordinates for which similarity calculation is needed, and H is the projection point of radar detection point F onto the line connecting reference point B and origin O. Therefore, the radar point distance between radar detection point F and vehicle O in the figure can be the length of line segment OF, and the projection distance between radar detection point F and reference point B can be the length of line segment OH.

[0157] In this embodiment, the vehicle terminal calculates the distance similarity between the radar point coordinates and the reference point by projecting distance and radar point distance, and determines whether the current radar point coordinates are the target radar point based on the distance similarity, which can improve the accuracy of target radar point selection.

[0158] S603. Based on the angle similarity and the distance similarity, determine the target similarity between the coordinates of any radar point and the detection box coordinate matrix.

[0159] In this embodiment, after calculating the angular similarity and distance similarity between the radar point coordinates and a certain reference point, the vehicle-mounted terminal can use the sum of the angular similarity and distance similarity as the reference point similarity between the radar point coordinates and the reference point. After calculating the reference point similarity between the radar point coordinates and all reference points on the detection box, the vehicle-mounted terminal can determine whether the similarity of each reference point meets the reference conditions preset by the developers. If the vehicle-mounted terminal determines that the similarity of a certain reference point meets the reference conditions, then the vehicle-mounted terminal can determine that the reference point similarity that meets the reference conditions is the target similarity between the radar point coordinates and the detection box coordinate matrix. If the vehicle-mounted terminal determines that the similarity of a certain reference point does not meet the reference conditions, then the vehicle-mounted terminal can determine that the similarity of that reference point is not the target similarity. Specifically, the vehicle-mounted terminal can determine that the minimum reference point similarity among all reference point similarities meets the reference conditions, and determine that the minimum reference point similarity is the target similarity between the radar point coordinates and the detection box coordinate matrix, and the reference point corresponding to the minimum reference point similarity is the target reference point between the radar point coordinates and the detection box coordinate matrix.

[0160] In one possible implementation, the developers can pre-set the angle weights corresponding to angle similarity and the distance weights corresponding to distance similarity based on experimental results. The vehicle-mounted terminal can then perform a weighted sum of the angle similarity and distance similarity based on the angle weights and distance weights to calculate the reference point similarity for each reference point.

[0161] S604. If the target similarity satisfies a preset similarity condition, then the radar point coordinates corresponding to the target similarity that satisfies the similarity condition are taken as the target radar point corresponding to the detection box coordinate matrix.

[0162] In this embodiment, after determining the target similarity between the radar point coordinates and the detection box coordinate matrix, the vehicle-mounted terminal can determine whether the target similarity meets the similarity conditions preset by the developers. Specifically, the vehicle-mounted terminal can first determine whether the target similarity is less than the similarity threshold preset by the developers. If the vehicle-mounted terminal determines that the target similarity is less than the similarity threshold, then the vehicle-mounted terminal can determine that the target similarity meets the similarity conditions. If the vehicle-mounted terminal determines that the target similarity is greater than or equal to the similarity threshold preset by the developers, then the vehicle-mounted terminal can determine that the target similarity does not meet the similarity conditions.

[0163] In one possible implementation, the vehicle-mounted terminal can further determine whether the number of target similarities satisfying the similarity condition is greater than 1. If the vehicle-mounted terminal determines that the number of target similarities satisfying the similarity condition is greater than 1, then the vehicle-mounted terminal can determine the radar point with the smallest target similarity among multiple radar point coordinates with target similarities less than a similarity threshold as the target radar point corresponding to the detection box coordinate matrix. If the vehicle-mounted terminal determines that the number of target similarities satisfying the similarity condition is equal to 1, then the vehicle-mounted terminal can determine the radar point coordinates corresponding to the target similarity currently satisfying the similarity condition as the target radar point corresponding to the detection box coordinate matrix. If the vehicle-mounted terminal determines that the number of target similarities satisfying the similarity condition is less than 1, then the vehicle-mounted terminal can output an error message to indicate that a target detection error has occurred.

[0164] S605. The radar detection information corresponding to the target radar point is fused with the camera detection information to generate the target detection result.

[0165] In this embodiment, after determining the target radar point, the vehicle-mounted terminal can fuse the radar detection information corresponding to the target radar point with the camera detection information to generate a target detection result. Specifically, the vehicle-mounted terminal can fuse the speed, distance, and angle corresponding to the target radar point with the camera detection information to generate a target detection result. The vehicle-mounted terminal can use any target fusion algorithm known to those skilled in the art to fuse the radar detection information corresponding to the target radar point with the camera detection information. This embodiment is not intended to specifically limit the target fusion method.

[0166] In this embodiment, the vehicle-mounted terminal determines the target radar point based on distance and angle similarity. This considers both the similarity in length and angle between line segments, compensating for the limitations of considering information from a single dimension and effectively preventing erroneous associations caused by single-dimensional judgments, thereby improving the accuracy of target radar point selection. Furthermore, determining the target radar point based on similarity conditions allows for accurate and timely selection of the optimal associated target, effectively eliminating noise measurements from sensors and interference from measurements of other targets. Finally, the improved accuracy of target radar point selection enables the vehicle-mounted terminal to provide more precise information for autonomous driving functions, thereby enhancing the safety level of autonomous driving, improving the accuracy of autonomous driving decisions, and achieving higher-level safety requirements for autonomous driving functions.

[0167] It should be noted that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0168] Reference Figure 8 The diagram illustrates a radar split point detection device according to an embodiment of this application, which may specifically include an information acquisition module 801, a conversion module 802, a split detection module 803, and a target fusion module 804, wherein:

[0169] The information acquisition module 801 is used to acquire radar detection information and camera detection information;

[0170] The conversion module 802 is used to generate a matrix of multiple radar point coordinates and a detection frame coordinate matrix in a vehicle coordinate system with the vehicle as the origin, based on a preset coordinate transformation matrix, if the radar detection information meets preset split point detection conditions; the radar point coordinates are obtained by converting the radar detection information based on the coordinate transformation matrix; the detection frame coordinate matrix is ​​obtained by converting the camera detection information based on the coordinate transformation matrix.

[0171] The split detection module 803 is used to input the coordinates of any two radar points and the detection frame coordinate matrix into a preset split point detection algorithm to generate the split point detection result corresponding to the coordinates of the any two radar points.

[0172] The target fusion module 804 is used to fuse the radar detection information and the camera detection information based on the split point detection results to generate target detection results.

[0173] The split detection module can also be used to obtain the velocity difference and distance difference between the coordinates of any two radar points; input the coordinates of the two radar points and the detection frame coordinate matrix into a preset overlap rate algorithm to generate the overlap rate between the coordinates of the two radar points; if the velocity difference, the distance difference, and the overlap rate all satisfy the preset splitting condition, a first detection result is generated; the first detection result is used to indicate that the coordinates of the two radar points are radar detection points of the same target; if at least one of the velocity difference, the distance difference, and the overlap rate does not satisfy the splitting condition, a second detection result is generated; the second detection result is used to indicate that the coordinates of the two radar points are radar detection points of different targets.

[0174] The split detection module can also be used to input the coordinates of any two radar points into the first distance formula to determine the offset distance between the coordinates of any two radar points; calculate the overlap rate based on the offset distance and the overlap distance; the overlap distance is the mapping length of the line segment formed by the coordinates of any two radar points within the detection box corresponding to the detection box coordinate matrix; if the overlap rate is greater than the overlap rate threshold, then it is determined that the overlap rate satisfies the split condition.

[0175] The split detection module can also be used to input the coordinates of any two radar points into the second distance formula to determine the first mapping distance difference and the second mapping distance difference; obtain the target length and target width corresponding to the target object based on the detection box coordinate matrix; determine the width threshold based on the target width and the overlap rate; if the second mapping distance difference is less than the target length and the first mapping distance difference is less than the width threshold, then it is determined that the distance difference satisfies the split condition.

[0176] The target fusion module can also be used to, when the split point detection result is the second detection result, input the radar point coordinates into a preset first similarity algorithm to determine the angular similarity between the radar point coordinates and the detection box coordinate matrix for any radar point coordinates; input the radar point coordinates into a preset second similarity algorithm to determine the distance similarity between the radar point coordinates and the detection box coordinate matrix for any radar point coordinates; determine the target similarity between the radar point coordinates and the detection box coordinate matrix based on the angular similarity and the distance similarity; if the target similarity satisfies a preset similarity condition, then the radar point coordinates corresponding to the target similarity that satisfies the similarity condition are taken as the target radar point corresponding to the detection box coordinate matrix; and perform target fusion with the radar detection information corresponding to the target radar point and the camera detection information to generate the target detection result.

[0177] The target fusion module can also be used to input the detection box coordinate matrix into a preset first angle function to generate a first angle value; the first angle value represents the maximum angle between each reference point of the detection box corresponding to the detection box coordinate matrix; the reference point is a vertex or the midpoint of an edge of the detection box; input the radar point coordinates and any reference point into a preset second angle function to generate a second angle value; based on the first angle value and the second angle value, generate an angle similarity between the radar point coordinates and any reference point; input the radar point coordinates into a preset third distance formula to generate the radar point distance between the radar point coordinates and the vehicle. The coordinates of any reference point and the radar point coordinates are input into the fourth distance formula to generate a projected distance. The projected distance is the distance between the projected point corresponding to the radar point coordinates and the vehicle. The projected point is the point formed when the radar point coordinates are projected onto the line connecting any reference point and the origin. Based on the radar point distance and the projected distance, the distance similarity is generated. The sum of the angle similarity and the distance similarity is used as the reference point similarity between any reference point and the radar point coordinates. If the reference point similarity meets a preset reference condition, the reference point similarity that meets the reference condition is determined as the target similarity.

[0178] The target fusion module can also be used to obtain the actual distance between the coordinates of each radar point and the vehicle; determine the radar point with the smallest actual distance as the target radar point corresponding to the target object; and perform target fusion with the radar detection information corresponding to the target radar point and the camera detection information to generate the target detection result.

[0179] As the apparatus embodiments are basically similar to the method embodiments, they are described in a relatively simple manner. For relevant details, please refer to the description in the method embodiment section.

[0180] Reference Figure 9 The diagram illustrates a vehicle-mounted terminal according to an embodiment of this application. Figure 9 As shown, the vehicle-mounted terminal 900 in this embodiment includes: a processor 910, a memory 920, and a computer program 921 stored in the memory 920 and executable on the processor 910. When the processor 910 executes the computer program 921, it implements the steps in the various embodiments of the radar split point detection method described above, for example... Figure 2 Steps S201 to S204 are shown. Alternatively, when the processor 910 executes the computer program 921, it implements the functions of each module / unit in the above-described device embodiments, for example... Figure 8 The functions of modules 801 to 804 are shown.

[0181] For example, the computer program 921 can be divided into one or more modules / units, which are stored in the memory 920 and executed by the processor 910 to complete this application. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, which can be used to describe the execution process of the computer program 921 in the vehicle terminal 900. For example, the computer program 921 can be divided into an information acquisition module, a conversion module, a split detection module, and a target fusion module, with the specific functions of each module as follows:

[0182] The information acquisition module is used to acquire radar detection information and camera detection information;

[0183] The conversion module is used to generate multiple radar point coordinates and detection box coordinate matrices in a vehicle coordinate system constructed with the vehicle as the origin, based on a preset coordinate transformation matrix, if the radar detection information meets preset split point detection conditions; the radar point coordinates are obtained by converting the radar detection information based on the coordinate transformation matrix; the detection box coordinate matrix is ​​obtained by converting the camera detection information based on the coordinate transformation matrix.

[0184] The split detection module is used to input the coordinates of any two radar points and the detection frame coordinate matrix into a preset split point detection algorithm to generate the split point detection result corresponding to the coordinates of the any two radar points.

[0185] The target fusion module is used to fuse the radar detection information and the camera detection information based on the split point detection results to generate target detection results.

[0186] The vehicle-mounted terminal 900 may include, but is not limited to, a processor 910 and a memory 920. Those skilled in the art will understand that... Figure 9 This is merely one example of the vehicle terminal 900 and does not constitute a limitation on the vehicle terminal 900. It may include more or fewer components than shown, or combine certain components, or different components. For example, the vehicle terminal 900 may also include input / output devices, network access devices, buses, etc.

[0187] The processor 910 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0188] The memory 920 can be an internal storage unit of the vehicle terminal 900, such as a hard drive or memory of the vehicle terminal 900. The memory 920 can also be an external storage device of the vehicle terminal 900, such as a plug-in hard drive, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc., equipped on the vehicle terminal 900. Furthermore, the memory 920 can include both internal storage units and external storage devices of the vehicle terminal 900. The memory 920 is used to store the computer program 921 and other programs and data required by the vehicle terminal 900. The memory 920 can also be used to temporarily store data that has been output or will be output.

[0189] This application also discloses an in-vehicle terminal, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the radar split point detection method as described in the foregoing embodiments.

[0190] This application also discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements the radar split point detection method as described in the foregoing embodiments.

[0191] This application also discloses a computer program product that, when run on a computer, causes the computer to execute the radar split point detection method described in the foregoing embodiments.

[0192] The embodiments described above are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A radar split point detection method, characterized in that, include: Acquire radar detection information and camera detection information; If the radar detection information meets the preset split point detection conditions, then based on the preset coordinate transformation matrix, a matrix of multiple radar point coordinates and detection box coordinates in a vehicle coordinate system constructed with the vehicle as the origin is generated; the radar point coordinates are obtained by transforming the radar detection information based on the coordinate transformation matrix; the detection box coordinate matrix is ​​obtained by transforming the camera detection information based on the coordinate transformation matrix. Input the coordinates of any two radar points and the detection box coordinate matrix into a preset split point detection algorithm to generate the split point detection result corresponding to the coordinates of any two radar points. Based on the split point detection results, the radar detection information and the camera detection information are fused to generate target detection results.

2. The method according to claim 1, characterized in that, The split point detection results include a first detection result and a second detection result; The step of inputting the coordinates of any two radar points and the detection box coordinate matrix into a preset split point detection algorithm to generate split point detection results corresponding to the coordinates of the any two radar points includes: Obtain the velocity difference and distance difference between the coordinates of any two radar points; Input the coordinates of any two radar points and the detection box coordinate matrix into a preset overlap rate algorithm to generate the overlap rate between the coordinates of any two radar points; If the velocity difference, the distance difference, and the overlap rate all satisfy the preset splitting conditions, a first detection result is generated; the first detection result is used to indicate that the coordinates of any two radar points are radar detection points of the same target. If at least one of the velocity difference, the distance difference, and the overlap rate does not satisfy the splitting condition, a second detection result is generated; the second detection result is used to indicate that the coordinates of any two radar points are radar detection points of different targets.

3. The method according to claim 2, characterized in that, The splitting conditions include an overlap rate threshold; The step of inputting the coordinates of any two radar points and the detection box coordinate matrix into a preset overlap rate algorithm to generate the overlap rate between the coordinates of any two radar points includes: Input the coordinates of any two radar points into the first distance formula to determine the offset distance between the coordinates of any two radar points. The overlap rate is calculated based on the offset distance and the overlap distance; the overlap distance is the mapping length of the line segment formed by the coordinates of any two radar points within the detection frame corresponding to the detection frame coordinate matrix; If the velocity difference, the distance difference, and the overlap rate all satisfy a preset splitting condition, a first detection result is generated, including: If the overlap rate is greater than the overlap rate threshold, then the overlap rate is determined to satisfy the splitting condition.

4. The method according to claim 2, characterized in that, The distance difference includes the first mapped distance difference corresponding to the horizontal axis and the second mapped distance difference corresponding to the vertical axis in the vehicle coordinate system. The step of obtaining the velocity difference and distance difference between any two radar point coordinates includes: Input the coordinates of any two radar points into the second distance formula to determine the first mapping distance difference and the second mapping distance difference; If the velocity difference, the distance difference, and the overlap rate all satisfy a preset splitting condition, a first detection result is generated, including: The target length and target width are obtained based on the coordinate matrix of the detection frame. A width threshold is determined based on the target width and the overlap rate; If the second mapping distance difference is less than the target length and the first mapping distance difference is less than the width threshold, then the distance difference is determined to satisfy the splitting condition.

5. The method according to any one of claims 1-4, characterized in that, The split point detection result includes a second detection result; the second detection result is used to indicate that any two radar point coordinates are radar detection points for different targets. The step of fusing the radar detection information and the camera detection information based on the split point detection results to generate target detection results includes: When the split point detection result is the second detection result, for any radar point coordinate, the radar point coordinate is input into a preset first similarity algorithm to determine the angular similarity between the radar point coordinate and the detection box coordinate matrix; For any radar point coordinate, the radar point coordinate is input into a preset second similarity algorithm to determine the distance similarity between the radar point coordinate and the detection box coordinate matrix; Based on the angle similarity and the distance similarity, the target similarity between the coordinates of any radar point and the detection box coordinate matrix is ​​determined; If the target similarity satisfies the preset similarity condition, then the radar point coordinates corresponding to the target similarity that satisfies the similarity condition are taken as the target radar point corresponding to the detection box coordinate matrix; The radar detection information corresponding to the target radar point is fused with the camera detection information to generate the target detection result.

6. The method according to claim 5, characterized in that, For any radar point coordinate, the radar point coordinates are input into a preset first similarity algorithm to generate an angular similarity between the radar point coordinates and the detection box coordinate matrix, including: The detection box coordinate matrix is ​​input into a preset first angle function to generate a first angle value; the first angle value is used to represent the maximum angle between each reference point of the detection box corresponding to the detection box coordinate matrix; the reference point is a vertex or the midpoint of an edge of the detection box; Input the radar point coordinates and any reference point into a preset second angle function to generate a second angle value; Based on the first angle value and the second angle value, generate the angle similarity between the radar point coordinates and any reference point; For any radar point coordinate, the radar point coordinates are input into a preset second similarity algorithm to generate a distance similarity between the radar point coordinates and the detection box coordinate matrix, including: The radar point coordinates are input into a preset third distance formula to generate the radar point distance between the radar point coordinates and the vehicle. Input the coordinates of any reference point and the radar point coordinates into the fourth distance formula to generate the projected distance; the projected distance is the distance between the projected point corresponding to the radar point coordinates and the vehicle; the projected point is the point formed when the radar point coordinates are projected onto the line connecting any reference point and the origin; The distance similarity is generated based on the radar point distance and the projected distance; The generation of target similarity based on the angle similarity and the distance similarity includes: The sum of the angle similarity and the distance similarity is taken as the reference point similarity between the coordinates of any reference point and the radar point. If the similarity of the reference points meets the preset reference conditions, then the similarity of the reference points that meets the reference conditions is determined as the target similarity.

7. The method according to any one of claims 1-4, characterized in that, The split point detection result includes a first detection result; the first detection result is used to indicate that any two radar point coordinates are radar detection points of the same target object; The step of fusing the radar detection information and the camera detection information based on the split point detection results to generate target detection results includes: Obtain the actual distance between the coordinates of each radar point and the vehicle; The radar point with the smallest actual distance is determined as the target radar point corresponding to the target object; The radar detection information corresponding to the target radar point is fused with the camera detection information to generate the target detection result.

8. A radar split point detection device, characterized in that, include: The information acquisition module is used to acquire radar detection information and camera detection information; The conversion module is used to generate multiple radar point coordinates and detection box coordinate matrices in a vehicle coordinate system constructed with the vehicle as the origin, based on a preset coordinate transformation matrix, if the radar detection information meets preset split point detection conditions; the radar point coordinates are obtained by converting the radar detection information based on the coordinate transformation matrix; the detection box coordinate matrix is ​​obtained by converting the camera detection information based on the coordinate transformation matrix. The split detection module is used to input the coordinates of any two radar points and the detection frame coordinate matrix into a preset split point detection algorithm to generate the split point detection result corresponding to the coordinates of the any two radar points. The target fusion module is used to fuse the radar detection information and the camera detection information based on the split point detection results to generate target detection results.

9. A vehicle-mounted terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the radar split point detection method as described in any one of claims 1-7.

10. A vehicle, characterized in that, It includes a millimeter-wave radar, a camera, and a vehicle-mounted terminal; the vehicle-mounted terminal is communicatively connected to the millimeter-wave radar and the camera, respectively. The millimeter-wave radar is used to acquire radar detection information; The camera is used to acquire camera detection information; The vehicle-mounted terminal is used to perform the radar split point detection method as described in any one of claims 1-7 based on the radar detection information and the camera detection information.

Images