A vehicle body orientation identification method, device and equipment

Abstract

The application provides a vehicle body orientation identification method, device and equipment, and relates to the technical field of computers. The method comprises the following steps: determining whether the point cloud cluster distribution of a target vehicle is L-shaped and incomplete; in the case where it is determined that the point cloud cluster distribution is L-shaped and incomplete, determining a target point cloud region of the point cloud cluster based on the point cloud cluster distribution; and performing trunk component analysis on the target point cloud region to determine the vehicle body orientation of the target vehicle. The scheme of the application solves the problems of high labor cost and insufficient detection accuracy of the existing vehicle body orientation identification method.

Description

Technical Field

[0001] This invention relates to the field of communication technology, and in particular to a method, apparatus and device for vehicle orientation recognition. Background Technology

[0002] In the field of roadside perception, due to the influence of complex weather conditions and variable light intensity, a single sensor cannot handle all scenarios. Therefore, a multi-sensor fusion perception technology is generally chosen to improve the system's ability to handle complex environments. Among them, LiDAR has become an essential sensor for roadside fusion perception because it can accurately acquire the three-dimensional information of objects and has the characteristic of stable performance in different scenarios. However, in intersection scenarios, the point cloud data of vehicles scanned by LiDAR often exhibits an L-shaped distribution, that is, the side of the vehicle facing the LiDAR has scanned point cloud data, while the other side is empty, which affects the calculation of the vehicle's orientation and thus affects the detection and tracking effect of vehicles at intersections.

[0003] Existing technologies mainly calculate vehicle orientation through end-to-end vehicle orientation calculation methods, principal component analysis, and methods that obtain vehicle orientation based on the angle at which the vehicle point cloud on the xy plane is minimized by rotation. These methods have the following main problems:

[0004] First, the end-to-end vehicle orientation calculation method requires manual annotation of a large amount of point cloud data, which has high manpower requirements.

[0005] Second, the direction with the largest variance after projecting the target vehicle point cloud cluster in the xy plane by directly using principal component analysis to obtain the point cloud cluster is inconsistent with the actual orientation of the vehicle, which leads to problems such as subsequent target object classification errors.

[0006] Third, the method of obtaining the vehicle orientation based on the angle when the vehicle point cloud has the minimum area on the xy plane by rotation is not applicable to point cloud clusters with severe incompleteness and has redundant calculations.

[0007] In summary, existing vehicle orientation recognition methods suffer from high labor costs and insufficient detection accuracy, failing to meet the goal of reliable detection and tracking of vehicles at intersections. Summary of the Invention

[0008] The purpose of this invention is to provide a vehicle orientation recognition method, device, and equipment, which solves the problems of high labor costs and insufficient detection accuracy in existing vehicle orientation recognition methods.

[0009] To achieve the above objectives, embodiments of the present invention provide a vehicle orientation recognition method, comprising:

[0010] Determine whether the point cloud cluster distribution of the target vehicle is L-shaped and incomplete;

[0011] If the distribution of the point cloud cluster is determined to be L-shaped and incomplete, the target point cloud region of the point cloud cluster is determined based on the distribution of the point cloud cluster.

[0012] The target point cloud region is subjected to backbone component analysis to determine the vehicle's orientation.

[0013] Optionally, determining whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete includes:

[0014] Based on principal component analysis, the primary and secondary directions of the point cloud cluster are obtained; wherein the secondary direction is perpendicular to the primary direction.

[0015] Determine the coordinates of the center point of the point cloud cluster, wherein the center point coordinates are the coordinates of the point cloud located at the center position of the main direction and the secondary direction;

[0016] Based on the center point coordinates, the primary direction, and the secondary direction, determine whether the point cloud cluster distribution of the target vehicle is L-shaped and incomplete.

[0017] Optionally, determining whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete based on the center point coordinates, the primary direction, and the secondary direction includes:

[0018] The point cloud cluster is transformed from the radar coordinate system to the first rectangular coordinate system; wherein the first rectangular coordinate system takes the center point coordinates as the origin, the main direction as the X-axis, and the secondary direction as the Y-axis;

[0019] Obtain the cloud coordinates of the first point located in the first coordinate region, the cloud coordinates of the second point located in the second coordinate region, the cloud coordinates of the third point located in the third coordinate region, and the cloud coordinates of the fourth point located in the fourth coordinate region;

[0020] Based on the first point cloud coordinates, the second point cloud coordinates, the third point cloud coordinates, and the fourth point cloud coordinates, determine whether the point cloud cluster distribution of the target vehicle is L-shaped and incomplete.

[0021] Wherein, the first coordinate region includes a first quadrant and a second quadrant; the second coordinate region includes a second quadrant and a third quadrant; the third coordinate region includes a third quadrant and a fourth quadrant; the fourth coordinate region includes a first quadrant and a fourth quadrant; the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant are regions within the first rectangular coordinate system.

[0022] Optionally, determining whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete based on the first point cloud coordinates, the second point cloud coordinates, the third point cloud coordinates, and the fourth point cloud coordinates includes:

[0023] Based on the first point cloud coordinates and the third point cloud coordinates, calculate the first distribution gap of the point cloud cluster in the secondary direction;

[0024] Based on the second point cloud coordinates and the fourth point cloud coordinates, calculate the second distribution gap of the point cloud cluster in the main direction;

[0025] When the first distribution gap is greater than the first threshold and the second distribution gap is greater than the second threshold, or when the first distribution gap is greater than the second threshold and the second distribution gap is greater than the first threshold, it is determined that the point cloud cluster distribution of the target vehicle is L-shaped incomplete.

[0026] Wherein, the first threshold is greater than the second threshold, and the second threshold is greater than zero.

[0027] Optionally, determining the coordinates of the center point of the point cloud cluster includes:

[0028] The point cloud cluster is rotated from the radar coordinate system to the second rectangular coordinate system; wherein the X-axis of the second rectangular coordinate system is parallel to the primary direction, and the Y-axis is parallel to the secondary direction.

[0029] The point cloud coordinates located at the center position in the X-axis and Y-axis directions of the second rectangular coordinate system are determined as the center point coordinates.

[0030] Optionally, rotating the point cloud cluster from the radar coordinate system to the second Cartesian coordinate system includes:

[0031] Obtain a first matrix, which is a two-dimensional matrix including the first set of coordinate values;

[0032] Obtain a second matrix, which is a two-dimensional matrix including the primary direction and the secondary direction;

[0033] By multiplying the first matrix with the second matrix, the point cloud coordinates of the point cloud cluster in the radar coordinate system are rotated to the second rectangular coordinate system.

[0034] Optionally, determining the target point cloud region of the point cloud cluster based on the point cloud cluster distribution includes:

[0035] Transform the point cloud cluster's point cloud coordinates in the radar coordinate system to the first rectangular coordinate system;

[0036] The intersection of the first point cloud region and the second point cloud region is taken as the target point cloud region; wherein the first point cloud region is the point cloud region on both sides of the X-axis of the first rectangular coordinate system that includes more point cloud coordinates; the second point cloud region is the point cloud region on both sides of the Y-axis of the first rectangular coordinate system that includes fewer point cloud coordinates.

[0037] The first rectangular coordinate system has the center point coordinates of the point cloud cluster as the origin, the main direction of the point cloud cluster as the X-axis, and the secondary direction of the point cloud cluster as the Y-axis; the center point coordinates are the coordinates of the point cloud located at the center position of the main direction and the secondary direction, and the secondary direction is perpendicular to the main direction.

[0038] To achieve the above objectives, embodiments of the present invention provide a vehicle orientation recognition device, comprising:

[0039] The first processing module is used to determine whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete;

[0040] The second processing module is used to determine the target point cloud region of the point cloud cluster based on the point cloud cluster distribution when it is determined that the point cloud cluster distribution is L-shaped incomplete.

[0041] The third processing module is used to perform backbone component analysis on the target point cloud region to determine the vehicle's orientation.

[0042] To achieve the above objectives, embodiments of the present invention provide a vehicle orientation recognition device, comprising: a transceiver and a processor; wherein the processor is configured to determine whether the point cloud cluster distribution of a target vehicle is L-shaped incomplete; if the point cloud cluster distribution is determined to be L-shaped incomplete, a target point cloud region of the point cloud cluster is determined based on the point cloud cluster distribution; and backbone component analysis is performed on the target point cloud region to determine the vehicle orientation of the target vehicle.

[0043] To achieve the above objectives, embodiments of the present invention provide a vehicle orientation recognition device, including a transceiver, a processor, a memory, and a program or instructions stored in the memory and executable on the processor; when the processor executes the program or instructions, it implements the vehicle orientation recognition method as described above.

[0044] To achieve the above objectives, embodiments of the present invention provide a readable storage medium having a program or instructions stored thereon, which, when executed by a processor, implement the steps in the vehicle orientation recognition method as described above.

[0045] The beneficial effects of the above-described technical solution of the present invention are as follows:

[0046] The method of this invention determines whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete; and if the point cloud cluster distribution is determined to be L-shaped incomplete, it determines the target point cloud region of the point cloud cluster based on the point cloud cluster distribution; and performs principal component analysis on the target point cloud region to determine the vehicle's orientation. This avoids the problems of inconsistency between the direction of the largest variance after projection of points within the target vehicle's point cloud cluster in the xy plane obtained from direct principal component analysis and the actual orientation of the vehicle, leading to subsequent errors in target object classification and inaccurate detection. It is suitable for point cloud clusters with severe incompleteness, does not require manual annotation of large amounts of point cloud data, has low computational load, and low cost. Attached Figure Description

[0047] Figure 1 This is a flowchart of the vehicle orientation recognition method according to an embodiment of the present invention;

[0048] Figure 2 This is a schematic diagram of point cloud cluster distribution according to an embodiment of the present invention;

[0049] Figure 3 This is a flowchart of a vehicle orientation recognition method according to another embodiment of the present invention;

[0050] Figure 4 This is a structural block diagram of the vehicle orientation recognition device according to an embodiment of the present invention;

[0051] Figure 5 This is one of the structural block diagrams of the vehicle orientation recognition device according to an embodiment of the present invention;

[0052] Figure 6 This is the second structural block diagram of the vehicle orientation recognition device according to an embodiment of the present invention. Detailed Implementation

[0053] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0054] It should be understood that the phrase "one embodiment" or "an embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of the invention. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification do not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0055] In various embodiments of the present invention, it should be understood that the sequence number of each process described below 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 the present invention.

[0056] In addition, the terms "system" and "network" are often used interchangeably in this article.

[0057] In the embodiments provided in this application, it should be understood that "B corresponding to A" means that B is associated with A, and B can be determined based on A. However, it should also be understood that determining B based on A does not mean determining B solely based on A; B can also be determined based on A and / or other information.

[0058] like Figure 1 As shown, an embodiment of the present invention provides a vehicle orientation recognition method, which includes the following steps:

[0059] Step 101: Determine whether the point cloud cluster distribution of the target vehicle is L-shaped and incomplete.

[0060] It should be noted that in intersection scenarios, the point cloud data of vehicles scanned by LiDAR often exhibits an L-shaped distribution, meaning that the front or rear of the vehicle, as well as some sides facing the LiDAR, have scanned point clouds, while the other side is empty, thus forming a point cloud cluster with an approximate "L" shape.

[0061] Step 102: If it is determined that the point cloud cluster distribution is incomplete in an L-shape, the target point cloud region of the point cloud cluster is determined based on the point cloud cluster distribution.

[0062] In this step, the target point cloud region can be specifically the point cloud region located at the head of the "L" in the L-shaped distribution of the point cloud cluster (e.g., Figure 2 (The point cloud region circled by the dashed line v). Taking a small car as an example, the target point cloud region is roughly the point cloud distribution corresponding to the door area.

[0063] Step 103: Perform backbone component analysis on the target point cloud region to determine the vehicle's orientation.

[0064] Figure 2 The dashed rectangles in the matrix represent the actual vehicle outline, and the dashed lines at the central axis of the matrix represent the actual vehicle orientation. The principal direction (X-axis) obtained directly from principal component analysis of the point cloud clusters deviates from the actual vehicle orientation by a certain angle. However, due to... Figure 2 As can be seen, the main direction of the point cloud region corresponding to the head of the "L" is consistent with the actual vehicle orientation. Therefore, by performing backbone component analysis on the point cloud of this target region, the obtained main direction is the accurate vehicle orientation.

[0065] It should be noted that for vehicle point cloud clusters with complete shapes, principal component analysis can be performed directly on the point cloud clusters to obtain the vehicle orientation.

[0066] As one implementation method, the acquisition of point cloud clusters of the target vehicle may include: using a pass-through filter to filter out unwanted points in the horizontal, vertical and other directions according to the required detection area to obtain the region of interest (ROI); further filtering out the ground part from the ROI and performing clustering calculations on the point cloud to obtain point cloud clusters.

[0067] Specifically, extracting the region of interest and filtering out the ground portion may include the following steps:

[0068] Step a: Randomly select 3 points from the point cloud to form a plane.

[0069] Step b: Calculate the distances from other points to the plane.

[0070] Step c: Set a threshold; points with a distance less than the threshold are considered in-group points; record the number of in-group points.

[0071] Step d, repeat steps a through c n times.

[0072] Step e: Select the plane with the most internal group points as the ground.

[0073] In step f, the least squares method is used for fine-tuning, and the point cloud of the ground portion is filtered out after obtaining the ground portion.

[0074] Specifically, performing clustering calculations on point clouds may include the following steps:

[0075] The DBSCAN clustering algorithm is used. This algorithm first divides the points in the point cloud after the first step of processing into three categories: core points, boundary points, and noise points. Core points are points that contain more than the minimum number of points min_samples within a radius min_distance; boundary points are points whose number of points within a radius min_distance is less than a set value minPts and fall within the neighborhood of core points; noise points are points that are neither core points nor boundary points and do not belong to any cluster. The main idea is as follows: Step a, extract a point p from the unlabeled point cloud; Step b, if p has not been processed (assigned to a cluster or labeled as noise), check its neighborhood. If it contains at least minPts objects, create a new cluster C and add all points in it to the candidate set N; Step c, for all unprocessed objects q in the candidate set N, check their neighborhoods. If they contain at least minPts objects, add these objects to the candidate set N; if q has not been assigned to any cluster, add q to C; Step d, repeat step c to continue checking unprocessed objects in N until the current candidate set N is empty; Step e, repeat steps a, b, c, and d until all objects have been assigned to a cluster or labeled as noise. Through clustering calculations, the point cloud is divided into different clusters, each containing different objects.

[0076] See Figure 3 In one embodiment, step 101 includes:

[0077] Step 1011: Based on principal component analysis, obtain the primary and secondary directions of the point cloud cluster; wherein the secondary direction is perpendicular to the primary direction.

[0078] like Figure 2 The diagram shows the primary and secondary directions.

[0079] Step 1012: Determine the coordinates of the center point of the point cloud cluster. The coordinates of the center point are the coordinates of the point cloud located at the center position of the main direction and the secondary direction.

[0080] In this step, the center point coordinates are the coordinates of the point cloud cluster that is located at the center of both the primary and secondary directions.

[0081] Step 1013: Based on the center point coordinates, the primary direction, and the secondary direction, determine whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete.

[0082] In one specific embodiment, step 1013 includes:

[0083] like Figure 2 As shown, the point cloud cluster's point cloud coordinates in the radar coordinate system are transformed to the first rectangular coordinate system; wherein, the first rectangular coordinate system takes the center point coordinates as the origin, the main direction as the X-axis, and the secondary direction as the Y-axis;

[0084] Obtain the cloud coordinates of the first point located in the first coordinate region, the cloud coordinates of the second point located in the second coordinate region, the cloud coordinates of the third point located in the third coordinate region, and the cloud coordinates of the fourth point located in the fourth coordinate region;

[0085] Based on the first point cloud coordinates, the second point cloud coordinates, the third point cloud coordinates, and the fourth point cloud coordinates, determine whether the point cloud cluster distribution of the target vehicle is L-shaped and incomplete.

[0086] Wherein, the first coordinate region includes a first quadrant and a second quadrant; the second coordinate region includes a second quadrant and a third quadrant; the third coordinate region includes a third quadrant and a fourth quadrant; the fourth coordinate region includes a first quadrant and a fourth quadrant; the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant are regions within the first rectangular coordinate system.

[0087] Here, the first point cloud coordinates refer to the number of point clouds located in the first coordinate region; the second point cloud coordinates refer to the number of point clouds located in the second coordinate region; the third point cloud coordinates refer to the number of point clouds located in the third coordinate region; and the fourth point cloud coordinates refer to the number of point clouds located in the fourth coordinate region.

[0088] Further, in one embodiment, determining whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete based on the first point cloud coordinates, the second point cloud coordinates, the third point cloud coordinates, and the fourth point cloud coordinates includes:

[0089] Based on the first point cloud coordinates and the third point cloud coordinates, calculate the first distribution gap of the point cloud cluster in the secondary direction;

[0090] Based on the second point cloud coordinates and the fourth point cloud coordinates, calculate the second distribution gap of the point cloud cluster in the main direction;

[0091] When the first distribution gap is greater than the first threshold and the second distribution gap is greater than the second threshold, or when the first distribution gap is greater than the second threshold and the second distribution gap is greater than the first threshold, the point cloud cluster distribution of the target vehicle is determined to be L-shaped incomplete; wherein, the first threshold is greater than the second threshold and the second threshold is greater than zero.

[0092] For example, the first distribution gap is calculated as follows: jud_y = |A1-A2| / A2; A1 refers to the number of coordinates of the first point cloud, A2 refers to the number of coordinates of the third point cloud, and jud_y refers to the first distribution gap.

[0093] The second distribution gap is calculated as follows: jud_x = |B1-B2| / B2; B1 refers to the number of coordinates of the second point cloud, B2 refers to the number of coordinates of the fourth point cloud, and jud_x refers to the second distribution gap.

[0094] If jud_x > max and jud_y > min; or jud_y > max and jud_x > min, then the point cloud cluster distribution of the target vehicle is determined to be L-shaped incomplete; where max is the first threshold and min is the second threshold.

[0095] It should be noted that the first and second thresholds can be set according to requirements, but cannot be 0. They are used to determine whether there are different degrees of uneven point cloud distribution on both sides of the main direction and both sides of the secondary direction, and whether the unevenness is greater or less than the other. Based on this, it is determined whether it is an L-shaped defect. If the point cloud cluster is complete or only includes the point cloud distribution of one side of the vehicle (such as the head or tail), then it does not meet the discriminant.

[0096] Specifically, in one embodiment, step 1012 includes:

[0097] The point cloud cluster is rotated from the radar coordinate system to the second rectangular coordinate system; wherein the X-axis of the second rectangular coordinate system is parallel to the primary direction, and the Y-axis is parallel to the secondary direction.

[0098] The point cloud coordinates located at the center position in the X-axis and Y-axis directions of the second rectangular coordinate system are determined as the center point coordinates.

[0099] It should be noted that the point cloud coordinates of the point cloud cluster in the radar coordinate system in this embodiment are obtained by performing PCA dimensionality reduction on the point cloud cluster, removing the height direction information, and obtaining the x-axis and y-axis coordinates of each point cloud.

[0100] In this embodiment, the second rectangular coordinate system is obtained by rotating the radar coordinate system. By rotating the point cloud cluster's point cloud coordinates in the radar coordinate system to the second rectangular coordinate system, it is easier to calculate the center point coordinates and reduce the amount of computation.

[0101] Specifically, rotating the point cloud cluster from the radar coordinate system to the second Cartesian coordinate system includes:

[0102] Obtain a first matrix E, which is a two-dimensional matrix including the set of coordinate values ​​of the point cloud cluster in the radar coordinate system;

[0103] Obtain a second matrix M, which is a two-dimensional matrix including the primary direction and the secondary direction;

[0104] By multiplying the first matrix with the second matrix, the point cloud coordinates of the point cloud cluster in the radar coordinate system are rotated to the second rectangular coordinate system.

[0105] In this embodiment, the first matrix E includes the x-axis and y-axis coordinates of each point cloud cluster in the radar coordinate system.

[0106] In one embodiment, the method further includes:

[0107] Estimate the size of the target vehicle based on the primary and secondary directions; filter out the point cloud corresponding to non-vehicle objects based on the size of the target vehicle.

[0108] In one embodiment, step 102 includes:

[0109] Transform the point cloud cluster's point cloud coordinates in the radar coordinate system to the first rectangular coordinate system;

[0110] The intersection of the first point cloud region and the second point cloud region is taken as the target point cloud region; wherein the first point cloud region is the point cloud region on both sides of the X-axis of the first rectangular coordinate system that includes more point cloud coordinates; the second point cloud region is the point cloud region on both sides of the Y-axis of the first rectangular coordinate system that includes fewer point cloud coordinates.

[0111] The first rectangular coordinate system has the center point coordinates of the point cloud cluster as the origin, the main direction of the point cloud cluster as the X-axis, and the secondary direction of the point cloud cluster as the Y-axis; the center point coordinates are the coordinates of the point cloud located at the center position of the main direction and the secondary direction, and the secondary direction is perpendicular to the main direction.

[0112] For example, with Figure 2 For example, it shows the first rectangular coordinate system, and the target point cloud region is the point cloud region in the first quadrant (the region where both the X-axis and Y-axis are positive) of the first rectangular coordinate system.

[0113] This embodiment can determine the target point cloud region based on the distribution of point cloud clusters, and then perform principal component analysis based on the target point cloud region to obtain the main direction, which is consistent with the vehicle body orientation.

[0114] like Figure 4 As shown, an embodiment of the present invention provides a vehicle orientation recognition device 400, comprising:

[0115] The first processing module 401 is used to determine whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete;

[0116] The second processing module 402 is used to determine the target point cloud region of the point cloud cluster based on the point cloud cluster distribution when it is determined that the point cloud cluster distribution is L-shaped incomplete.

[0117] The third processing module 403 is used to perform backbone component analysis on the target point cloud region to determine the vehicle body orientation of the target vehicle.

[0118] Optionally, the first processing module 401 includes:

[0119] The first processing submodule is used to obtain the main direction and secondary direction of the point cloud cluster based on the principal component analysis method; wherein the secondary direction is perpendicular to the main direction;

[0120] The second processing submodule is used to determine the coordinates of the center point of the point cloud cluster, wherein the coordinates of the center point are the coordinates of the point cloud located at the center position of the main direction and the secondary direction;

[0121] The third processing submodule is used to determine whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete based on the center point coordinates, the main direction, and the secondary direction.

[0122] Optionally, the third processing submodule includes:

[0123] The first processing unit is used to transform the point cloud cluster's point cloud coordinates in the radar coordinate system to the first rectangular coordinate system; wherein the first rectangular coordinate system takes the center point coordinates as the origin, the main direction as the X-axis, and the secondary direction as the Y-axis;

[0124] The second processing unit is used to obtain the first point cloud coordinates located in the first coordinate region, the second point cloud coordinates located in the second coordinate region, the third point cloud coordinates located in the third coordinate region, and the fourth point cloud coordinates located in the fourth coordinate region.

[0125] The third processing unit is used to determine whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete based on the first point cloud coordinate number, the second point cloud coordinate number, the third point cloud coordinate number and the fourth point cloud coordinate number.

[0126] Wherein, the first coordinate region includes a first quadrant and a second quadrant; the second coordinate region includes a second quadrant and a third quadrant; the third coordinate region includes a third quadrant and a fourth quadrant; the fourth coordinate region includes a first quadrant and a fourth quadrant; the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant are regions within the first rectangular coordinate system.

[0127] Optionally, the third processing unit is specifically used for:

[0128] Based on the first point cloud coordinates and the third point cloud coordinates, calculate the first distribution gap of the point cloud cluster in the secondary direction;

[0129] Based on the second point cloud coordinates and the fourth point cloud coordinates, calculate the second distribution gap of the point cloud cluster in the main direction;

[0130] When the first distribution gap is greater than the first threshold and the second distribution gap is greater than the second threshold, or when the first distribution gap is greater than the second threshold and the second distribution gap is greater than the first threshold, it is determined that the point cloud cluster distribution of the target vehicle is L-shaped incomplete.

[0131] Wherein, the first threshold is greater than the second threshold, and the second threshold is greater than zero.

[0132] Optionally, the second processing submodule includes:

[0133] The fourth processing unit is used to rotate the point cloud cluster's point cloud coordinates in the radar coordinate system to the second rectangular coordinate system; wherein, the X-axis direction of the second rectangular coordinate system is parallel to the primary direction, and the Y-axis direction is parallel to the secondary direction;

[0134] The fifth processing unit is used to determine the point cloud coordinates of the center position located in the X-axis and Y-axis directions of the second rectangular coordinate system as the center point coordinates.

[0135] Optionally, the fourth processing unit is specifically used for:

[0136] Obtain a first matrix, which is a two-dimensional matrix including the set of coordinate values ​​of the point cloud cluster in the radar coordinate system;

[0137] Obtain a second matrix, which is a two-dimensional matrix including the primary direction and the secondary direction;

[0138] By multiplying the first matrix with the second matrix, the point cloud coordinates of the point cloud cluster in the radar coordinate system are rotated to the second rectangular coordinate system.

[0139] Optionally, the second processing module 502 includes:

[0140] The fourth processing submodule is used to transform the point cloud coordinates of the point cloud cluster in the radar coordinate system to the first rectangular coordinate system;

[0141] The fifth processing submodule is used to take the intersection of the first point cloud region and the second point cloud region as the target point cloud region; wherein the first point cloud region is the point cloud region on the side of the first rectangular coordinate system with a larger number of point cloud coordinates on both sides of the X-axis; and the second point cloud region is the point cloud region on the side of the first rectangular coordinate system with a smaller number of point cloud coordinates on both sides of the Y-axis.

[0142] The first rectangular coordinate system has the center point coordinates of the point cloud cluster as the origin, the main direction of the point cloud cluster as the X-axis, and the secondary direction of the point cloud cluster as the Y-axis; the center point coordinates are the coordinates of the point cloud located at the center position of the main direction and the secondary direction, and the secondary direction is perpendicular to the main direction.

[0143] The vehicle orientation recognition device embodiment of the present invention corresponds to the above method. All implementation means in the above method embodiment are applicable to the vehicle orientation recognition device embodiment and can achieve the same technical effect.

[0144] like Figure 5 As shown, a vehicle orientation recognition device 500 according to an embodiment of the present invention includes a processor 510 and a transceiver 520, wherein,

[0145] The processor is used to determine whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete;

[0146] If the distribution of the point cloud cluster is determined to be L-shaped and incomplete, the target point cloud region of the point cloud cluster is determined based on the distribution of the point cloud cluster.

[0147] The target point cloud region is subjected to backbone component analysis to determine the vehicle's orientation.

[0148] The transceiver 520 is used to receive and send data under the control of the processor 510.

[0149] Optionally, the processor is specifically configured to: obtain the primary and secondary directions of the point cloud cluster based on principal component analysis; wherein the secondary direction is perpendicular to the primary direction; determine the coordinates of the center point of the point cloud cluster, wherein the center point coordinates are the coordinates of the point cloud located at the center position of the primary direction and the secondary direction; and determine whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete based on the center point coordinates, the primary direction, and the secondary direction.

[0150] Optionally, the processor 510 is further specifically used for:

[0151] The point cloud cluster is transformed from the radar coordinate system to the first rectangular coordinate system; wherein the first rectangular coordinate system takes the center point coordinates as the origin, the main direction as the X-axis, and the secondary direction as the Y-axis;

[0152] Obtain the cloud coordinates of the first point located in the first coordinate region, the cloud coordinates of the second point located in the second coordinate region, the cloud coordinates of the third point located in the third coordinate region, and the cloud coordinates of the fourth point located in the fourth coordinate region;

[0153] Based on the first point cloud coordinates, the second point cloud coordinates, the third point cloud coordinates, and the fourth point cloud coordinates, determine whether the point cloud cluster distribution of the target vehicle is L-shaped and incomplete.

[0154] Wherein, the first coordinate region includes a first quadrant and a second quadrant; the second coordinate region includes a second quadrant and a third quadrant; the third coordinate region includes a third quadrant and a fourth quadrant; the fourth coordinate region includes a first quadrant and a fourth quadrant; the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant are regions within the first rectangular coordinate system.

[0155] Optionally, the processor 510 is further specifically used for:

[0156] Based on the first point cloud coordinates and the third point cloud coordinates, calculate the first distribution gap of the point cloud cluster in the secondary direction;

[0157] Based on the second point cloud coordinates and the fourth point cloud coordinates, calculate the second distribution gap of the point cloud cluster in the main direction;

[0158] When the first distribution gap is greater than the first threshold and the second distribution gap is greater than the second threshold, or when the first distribution gap is greater than the second threshold and the second distribution gap is greater than the first threshold, it is determined that the point cloud cluster distribution of the target vehicle is L-shaped incomplete.

[0159] Wherein, the first threshold is greater than the second threshold, and the second threshold is greater than zero.

[0160] Optionally, the processor 510 is further specifically used for:

[0161] The point cloud cluster is rotated from the radar coordinate system to the second rectangular coordinate system; wherein the X-axis of the second rectangular coordinate system is parallel to the primary direction, and the Y-axis is parallel to the secondary direction.

[0162] The point cloud coordinates located at the center position in the X-axis and Y-axis directions of the second rectangular coordinate system are determined as the center point coordinates.

[0163] Optionally, the processor 510 is further specifically used for:

[0164] Obtain a first matrix, which is a two-dimensional matrix including the set of coordinate values ​​of the point cloud cluster in the radar coordinate system;

[0165] Obtain a second matrix, which is a two-dimensional matrix including the primary direction and the secondary direction;

[0166] By multiplying the first matrix with the second matrix, the point cloud coordinates of the point cloud cluster in the radar coordinate system are rotated to the second rectangular coordinate system.

[0167] Optionally, the processor 510 is specifically used for:

[0168] Transform the point cloud cluster's point cloud coordinates in the radar coordinate system to the first rectangular coordinate system;

[0169] The intersection of the first point cloud region and the second point cloud region is taken as the target point cloud region; wherein the first point cloud region is the point cloud region on both sides of the X-axis of the first rectangular coordinate system that includes more point cloud coordinates; the second point cloud region is the point cloud region on both sides of the Y-axis of the first rectangular coordinate system that includes fewer point cloud coordinates.

[0170] The first rectangular coordinate system has the center point coordinates of the point cloud cluster as the origin, the main direction of the point cloud cluster as the X-axis, and the secondary direction of the point cloud cluster as the Y-axis; the center point coordinates are the coordinates of the point cloud located at the center position of the main direction and the secondary direction, and the secondary direction is perpendicular to the main direction.

[0171] Another embodiment of the present invention provides a vehicle orientation recognition device, such as... Figure 6 As shown, it includes a transceiver 610, a processor 600, a memory 620, and a program or instructions stored in the memory 620 and executable on the processor 600; when the processor 600 executes the program or instructions, it implements the above-mentioned method for vehicle orientation recognition.

[0172] The transceiver 610 is used to receive and send data under the control of the processor 600.

[0173] Among them, Figure 6 In this context, the bus architecture can include any number of interconnected buses and bridges, specifically linking various circuits of one or more processors represented by processor 600 and memory represented by memory 620 together. The bus architecture can also link various other circuits such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. The bus interface provides an interface. Transceiver 610 can be multiple elements, including transmitters and receivers, providing a unit for communicating with various other devices over a transmission medium. For different user equipment, user interface 630 can also be an interface capable of connecting external or internal devices, including but not limited to keypads, displays, speakers, microphones, joysticks, etc.

[0174] The processor 600 is responsible for managing the bus architecture and general processing, while the memory 620 can store the data used by the processor 600 when performing operations.

[0175] This invention provides a readable storage medium storing a program or instructions. When executed by a processor, the program or instructions implement the steps in the vehicle orientation recognition method described above and achieve the same technical effect. To avoid repetition, further details are omitted here. The computer-readable storage medium may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0176] It should be further noted that the devices described in this specification include, but are not limited to, smartphones, tablets, etc., and many of the described functional components are referred to as modules in order to more specifically emphasize the independence of their implementation.

[0177] In this embodiment of the invention, the module can be implemented in software so that it can be executed by various types of processors. For example, an identified executable code module may include one or more physical or logical blocks of computer instructions, which may be constructed as objects, procedures, or functions. Nevertheless, the executable code of the identified module does not need to be physically located together, but may include different instructions stored in different bits, which, when logically combined, constitute the module and achieve the module's intended purpose.

[0178] In practice, an executable code module can be a single instruction or many instructions, and can even be distributed across multiple different code segments, different programs, and across multiple memory devices. Similarly, operational data can be identified within the module and can be implemented in any suitable form and organized within any suitable type of data structure. This operational data can be collected as a single dataset or distributed across different locations (including different storage devices), and can exist, at least in part, solely as electronic signals within the system or network.

[0179] When a module can be implemented using software, considering the current level of hardware technology, modules that can be implemented in software can be implemented using hardware circuits by those skilled in the art to achieve the corresponding functions, without considering cost. These hardware circuits include conventional very-large-scale integrated circuits (VLSI) or gate arrays, as well as existing semiconductors such as logic chips and transistors, or other discrete components. Modules can also be implemented using programmable hardware devices, such as field-programmable gate arrays, programmable array logic, and programmable logic devices.

[0180] The exemplary embodiments described above are with reference to the accompanying drawings. Many different forms and embodiments are feasible without departing from the spirit and teachings of the invention. Therefore, the invention should not be construed as limiting the exemplary embodiments set forth herein. Rather, these exemplary embodiments are provided to make the invention complete and convey the scope of the invention to those skilled in the art. In these drawings, component dimensions and relative dimensions may be exaggerated for clarity. The terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. As used herein, unless clearly indicated otherwise, the singular forms “a,” “an,” and “the” are intended to include all such forms. It will be further understood that the terms “comprising” and / or “including”, when used in this specification, indicate the presence of the stated features, integers, steps, operations, components, and / or elements, but do not exclude the presence or addition of one or more other features, integers, steps, operations, components, and / or groups thereof. Unless otherwise indicated, when stated, a range of values ​​includes the upper and lower limits of the range and any subranges in between.

[0181] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for recognizing vehicle orientation, characterized in that, include: Determine whether the point cloud cluster distribution of the target vehicle is L-shaped and incomplete; If the distribution of the point cloud cluster is determined to be L-shaped and incomplete, the target point cloud region of the point cloud cluster is determined based on the distribution of the point cloud cluster. Perform backbone component analysis on the target point cloud region to determine the vehicle's orientation. The step of determining whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete includes: Based on principal component analysis, the primary and secondary directions of the point cloud cluster are obtained; wherein the secondary direction is perpendicular to the primary direction. Determine the coordinates of the center point of the point cloud cluster, wherein the center point coordinates are the coordinates of the point cloud located at the center position of the main direction and the secondary direction; The point cloud cluster is transformed from the radar coordinate system to the first rectangular coordinate system; wherein the first rectangular coordinate system takes the center point coordinates as the origin, the main direction as the X-axis, and the secondary direction as the Y-axis; Obtain the cloud coordinates of the first point located in the first coordinate region, the cloud coordinates of the second point located in the second coordinate region, the cloud coordinates of the third point located in the third coordinate region, and the cloud coordinates of the fourth point located in the fourth coordinate region; Based on the first point cloud coordinates, the second point cloud coordinates, the third point cloud coordinates, and the fourth point cloud coordinates, determine whether the point cloud cluster distribution of the target vehicle is L-shaped and incomplete. Wherein, the first coordinate region includes a first quadrant and a second quadrant; the second coordinate region includes a second quadrant and a third quadrant; the third coordinate region includes a third quadrant and a fourth quadrant; the fourth coordinate region includes a first quadrant and a fourth quadrant; the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant are regions within the first rectangular coordinate system.

2. The vehicle orientation recognition method according to claim 1, characterized in that, Based on the first point cloud coordinates, the second point cloud coordinates, the third point cloud coordinates, and the fourth point cloud coordinates, determine whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete, including: Based on the first point cloud coordinates and the third point cloud coordinates, calculate the first distribution gap of the point cloud cluster in the secondary direction; Based on the second point cloud coordinates and the fourth point cloud coordinates, calculate the second distribution gap of the point cloud cluster in the main direction; When the first distribution gap is greater than the first threshold and the second distribution gap is greater than the second threshold, or when the first distribution gap is greater than the second threshold and the second distribution gap is greater than the first threshold, it is determined that the point cloud cluster distribution of the target vehicle is L-shaped incomplete. Wherein, the first threshold is greater than the second threshold, and the second threshold is greater than zero.

3. The vehicle orientation recognition method according to claim 1, characterized in that, Determining the coordinates of the center point of the point cloud cluster includes: The point cloud cluster is rotated from the radar coordinate system to the second rectangular coordinate system; wherein the X-axis of the second rectangular coordinate system is parallel to the primary direction, and the Y-axis is parallel to the secondary direction. The point cloud coordinates located at the center position in the X-axis and Y-axis directions of the second rectangular coordinate system are determined as the center point coordinates.

4. The vehicle orientation recognition method according to claim 3, characterized in that, The step of rotating the point cloud cluster from the radar coordinate system to the second rectangular coordinate system includes: Obtain a first matrix, which is a two-dimensional matrix including the set of coordinate values ​​of the point cloud cluster in the radar coordinate system; Obtain a second matrix, which is a two-dimensional matrix including the primary direction and the secondary direction; By multiplying the first matrix with the second matrix, the point cloud coordinates of the point cloud cluster in the radar coordinate system are rotated to the second rectangular coordinate system.

5. The vehicle orientation recognition method according to claim 1, characterized in that, The step of determining the target point cloud region of the point cloud cluster based on the point cloud cluster distribution includes: Transform the point cloud cluster's point cloud coordinates in the radar coordinate system to the first rectangular coordinate system; The intersection of the first point cloud region and the second point cloud region is taken as the target point cloud region; wherein the first point cloud region is the point cloud region on both sides of the X-axis of the first rectangular coordinate system that includes more point cloud coordinates; the second point cloud region is the point cloud region on both sides of the Y-axis of the first rectangular coordinate system that includes fewer point cloud coordinates. The first rectangular coordinate system has the center point coordinates of the point cloud cluster as the origin, the main direction of the point cloud cluster as the X-axis, and the secondary direction of the point cloud cluster as the Y-axis; the center point coordinates are the coordinates of the point cloud located at the center position of the main direction and the secondary direction, and the secondary direction is perpendicular to the main direction.

6. A vehicle orientation recognition device, characterized in that, include: The first processing module is used to determine whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete; The second processing module is used to determine the target point cloud region of the point cloud cluster based on the point cloud cluster distribution when it is determined that the point cloud cluster distribution is L-shaped incomplete. The third processing module is used to perform backbone component analysis on the target point cloud region to determine the vehicle body orientation of the target vehicle. The first processing module includes: The first processing submodule is used to obtain the main direction and secondary direction of the point cloud cluster based on the principal component analysis method; wherein the secondary direction is perpendicular to the main direction; The second processing submodule is used to determine the coordinates of the center point of the point cloud cluster, wherein the coordinates of the center point are the coordinates of the point cloud located at the center position of the main direction and the secondary direction; The third processing submodule is used to transform the point cloud coordinates of the point cloud cluster in the radar coordinate system to the first rectangular coordinate system; wherein, the first rectangular coordinate system takes the center point coordinates as the origin, the main direction as the X-axis, and the secondary direction as the Y-axis; obtains the first point cloud coordinates located in the first coordinate region, the second point cloud coordinates located in the second coordinate region, the third point cloud coordinates located in the third coordinate region, and the fourth point cloud coordinates located in the fourth coordinate region; and determines whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete based on the first point cloud coordinates, the second point cloud coordinates, the third point cloud coordinates, and the fourth point cloud coordinates; wherein, the first coordinate region includes the first quadrant and the second quadrant; the second coordinate region includes the second quadrant and the third quadrant; the third coordinate region includes the third quadrant and the fourth quadrant; the fourth coordinate region includes the first quadrant and the fourth quadrant; the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant are regions within the first rectangular coordinate system.

7. A vehicle orientation recognition device, characterized in that, include: Transceiver and processor; The processor is used to determine whether the point cloud cluster distribution of the target vehicle is L-shaped incomplete; Given that the point cloud cluster distribution is determined to be L-shaped and incomplete, the target point cloud region of the point cloud cluster is determined based on the point cloud cluster distribution; the backbone component analysis is performed on the target point cloud region to determine the vehicle body orientation of the target vehicle. Specifically, the processor is used for: Based on principal component analysis, the primary and secondary directions of the point cloud cluster are obtained; wherein the secondary direction is perpendicular to the primary direction. Determine the coordinates of the center point of the point cloud cluster, wherein the center point coordinates are the coordinates of the point cloud located at the center position of the main direction and the secondary direction; The point cloud cluster is transformed from the radar coordinate system to the first rectangular coordinate system; wherein the first rectangular coordinate system takes the center point coordinates as the origin, the main direction as the X-axis, and the secondary direction as the Y-axis; Obtain the cloud coordinates of the first point located in the first coordinate region, the cloud coordinates of the second point located in the second coordinate region, the cloud coordinates of the third point located in the third coordinate region, and the cloud coordinates of the fourth point located in the fourth coordinate region; Based on the first point cloud coordinates, the second point cloud coordinates, the third point cloud coordinates, and the fourth point cloud coordinates, determine whether the point cloud cluster distribution of the target vehicle is L-shaped and incomplete. Wherein, the first coordinate region includes a first quadrant and a second quadrant; the second coordinate region includes a second quadrant and a third quadrant; the third coordinate region includes a third quadrant and a fourth quadrant; the fourth coordinate region includes a first quadrant and a fourth quadrant; the first quadrant, the second quadrant, the third quadrant, and the fourth quadrant are regions within the first rectangular coordinate system.

8. A vehicle orientation recognition device, comprising: A transceiver, a processor, a memory, and a program or instructions stored in the memory and executable on the processor; characterized in that, when the processor executes the program or instructions, it implements the vehicle orientation recognition method as described in any one of claims 1-5.

9. A readable storage medium having a program or instructions stored thereon, characterized in that, When the program or instructions are executed by the processor, they implement the steps in the vehicle orientation recognition method as described in any one of claims 1-5.

Images