Point cloud lane line marking method and device, storage medium and electronic equipment

Abstract

The invention provides a point cloud lane line marking method and device, a storage medium and electronic equipment, and is applied to the technical field of intelligent driving. According to the method, the point cloud lane line map is constructed by integrating the multi-frame image data acquired by the target positioning point, the lane line data of each image frame is accurately acquired and converted by using the positioning information, and batch marking of the point cloud lane lines is realized on the cloud marking platform, so that the precision and efficiency of point cloud lane line marking are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent driving, and particularly to a method, device, storage medium and electronic device for labeling point cloud lane lines. Background Art

[0002] In the field of autonomous driving, accurate environmental perception is one of the key factors to ensure safe navigation, and point cloud lane line labeling plays a crucial role in it. It not only helps the Light Detection and Ranging (LiDAR) system to accurately detect the surrounding environment of the vehicle, but also is the basis for map construction.

[0003] However, most of the current lane line labeling processes adopt a frame-by-frame labeling method, which is not only inefficient, but also due to the inconsistency of the overlapping areas of lane lines in adjacent frames, it will cause fluctuations in the accuracy of the labeled ground truth, affecting the reliability of the final result.

[0004] Therefore, how to improve the efficiency and accuracy of point cloud lane line labeling has become a technical problem that needs to be solved urgently by those skilled in the art. Summary of the Invention

[0005] In view of the above problems, the present invention provides a method, device, storage medium and electronic device for labeling point cloud lane lines that overcome the above problems or at least partially solve the above problems. The technical solutions are as follows:

[0006] A method for labeling point cloud lane lines includes:

[0007] Obtaining a point cloud lane line map, where the point cloud lane line map is a point cloud map constructed based on the lane line perception information of multiple image frames collected at target positioning points;

[0008] Using the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map to obtain the first point cloud lane line data of a single image frame, where the positioning information is the position information corresponding to the image frame of the target positioning point in the point cloud lane line map, and the first point cloud lane line data includes multiple target lane line measurement points in a positioning point coordinate system with the target positioning point as the coordinate origin, and the target lane line measurement points are lane line measurement points within a preset distance range from the target positioning point;

[0009] Converting the target lane line measurement points in the first point cloud lane line data of multiple image frames from the positioning point coordinate system to the same target coordinate system to obtain the second point cloud lane line data of each image frame;

[0010] Using the positioning information of multiple image frames and the second point cloud lane line data, perform batch annotation of lane lines on the cloud annotation platform to obtain the local map of the point cloud lane line of the target positioning point in each image frame.

[0011] Optionally, obtaining the first point cloud lane line data of a single image frame by using the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map includes:

[0012] Using the positioning information corresponding to a single image frame of the target positioning point in the point cloud lane line map, search for target lane line measurement points within a preset distance range from the target positioning point in the point cloud lane line map;

[0013] Convert the target lane line measurement points to the positioning point coordinate system with the target positioning point as the coordinate origin to obtain the first point cloud lane line data of a single image frame.

[0014] Optionally, the operation of performing batch annotation of lane lines on the cloud annotation platform by using the positioning information of multiple image frames and the second point cloud lane line data to obtain the local map of the point cloud lane line of the target positioning point in each image frame includes:

[0015] Based on the positioning information of multiple image frames and the second point cloud lane line data, the cloud annotation platform loads the corresponding first point cloud lane line local map;

[0016] Perform lane line aggregation annotation operation on each lane line measurement point in the first point cloud lane line local map to obtain a second point cloud lane line local map, where each lane line measurement point in the second point cloud lane line local map corresponds to lane line cluster identification information.

[0017] Optionally, before performing the lane line aggregation annotation operation on each lane line measurement point in the first point cloud lane line local map to obtain the second point cloud lane line local map, the method further includes:

[0018] Add missing first lane line measurement points in the first point cloud lane line local map;

[0019] And / or, delete incorrect second lane line measurement points in the first point cloud lane line local map.

[0020] Optionally, after performing the lane line aggregation annotation operation on each lane line measurement point in the first point cloud lane line local map to obtain the second point cloud lane line local map, the method further includes:

[0021] Using the positioning information of the target positioning point in the local map of the second point cloud lane line, obtain the third point cloud lane line data of a single image frame, where the third point cloud lane line data includes the target lane line measurement points converted back from the target coordinate system to the positioning point coordinate system;

[0022] Using the third point cloud lane line data and the lane line cluster identification information corresponding to the target lane line measurement points, perform lane line fitting on each lane line in a single image frame to obtain a lane line fitting result;

[0023] Correspondingly save the lane line fitting result and the positioning information corresponding to the single image frame.

[0024] Optionally, after using the positioning information of the target positioning point in the local map of the second point cloud lane line to obtain the third point cloud lane line data of a single image frame, the method further includes:

[0025] Using the positioning information and the third point cloud lane line data of multiple image frames, re-perform batch annotation of lane lines on the cloud annotation platform to obtain the local map of the third point cloud lane line of the target positioning point in each image frame, where each lane line measurement point in the local map of the third point cloud lane line corresponds to lane line cluster identification information.

[0026] Optionally, the third point cloud lane line data is saved as a PCD format file, and the lane line fitting result and the positioning information corresponding to the single image frame are saved as a JSON format file.

[0027] A point cloud lane line annotation device includes: a point cloud lane line map obtaining unit, a first point cloud lane line data obtaining unit, a second point cloud lane line data obtaining unit, and a lane line annotation unit,

[0028] The point cloud lane line map obtaining unit is used to obtain a point cloud lane line map, where the point cloud lane line map is a point cloud map constructed based on the lane line perception information of multiple image frames collected by a target positioning point;

[0029] The first point cloud lane line data obtaining unit is used to use the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map to obtain the first point cloud lane line data of a single image frame, where the positioning information is the position information corresponding to the image frame of the target positioning point in the point cloud lane line map, and the first point cloud lane line data includes multiple target lane line measurement points in a positioning point coordinate system with the target positioning point as the coordinate origin, and the target lane line measurement points are lane line measurement points within a preset distance range from the target positioning point;

[0030] The second point cloud lane line data acquisition unit is configured to convert the target lane line measurement points in the first point cloud lane line data of multiple image frames from the positioning point coordinate system to the same target coordinate system, so as to obtain the second point cloud lane line data of each image frame;

[0031] The lane line annotation unit is configured to use the positioning information and the second point cloud lane line data of multiple image frames to perform batch annotation of lane lines on a cloud annotation platform, so as to obtain the point cloud lane line local map of the target positioning point in each image frame.

[0032] A computer-readable storage medium, on which a program is stored, and when the program is executed by a processor, the point cloud lane line annotation method described in any one of the above is implemented.

[0033] An electronic device, the electronic device includes at least one processor, at least one memory connected to the processor, and a bus; wherein, the processor and the memory communicate with each other through the bus; the processor is configured to call program instructions in the memory to execute the point cloud lane line annotation method described in any one of the above.

[0034] By means of the above technical solutions, a point cloud lane line annotation method, device, storage medium and electronic device provided by the present invention obtain a point cloud lane line map, wherein the point cloud lane line map is a point cloud map constructed based on the lane line perception information of multiple image frames collected at a target positioning point; using the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map, the first point cloud lane line data of a single image frame is obtained, wherein the positioning information is the position information corresponding to the image frame of the target positioning point in the point cloud lane line map, and the first point cloud lane line data includes multiple target lane line measurement points in a positioning point coordinate system with the target positioning point as the coordinate origin, and the target lane line measurement points are lane line measurement points within a preset distance range from the target positioning point; converting the target lane line measurement points in the first point cloud lane line data of multiple image frames from the positioning point coordinate system to the same target coordinate system to obtain the second point cloud lane line data of each image frame; using the positioning information and the second point cloud lane line data of multiple image frames to perform batch annotation of lane lines on a cloud annotation platform to obtain the point cloud lane line local map of the target positioning point in each image frame. The present invention utilizes a high-precision point cloud lane line map and positioning information to realize batch annotation of multi-frame point cloud lane lines, making the point cloud lane line annotation process more efficient and automated, and improving the efficiency and accuracy of point cloud lane line annotation.

[0035] The above description is only an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features and advantages of the present invention more obvious and understandable, the following specifically illustrates the embodiments of the present invention. Description of the Drawings

[0036] By reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present invention. Moreover, throughout the drawings, the same reference numerals are used to represent the same components. In the drawings:

[0037] Figure 1 A flowchart showing an embodiment of the point cloud lane line annotation method provided by an embodiment of the present invention;

[0038] Figure 2 A flowchart showing the construction process of the point cloud lane line map provided by an embodiment of the present invention;

[0039] Figure 3 A schematic diagram of the annotation interface of the cloud annotation platform provided by an embodiment of the present invention;

[0040] Figure 4 A flowchart showing another embodiment of the point cloud lane line annotation method provided by an embodiment of the present invention;

[0041] Figure 5 A flowchart showing another embodiment of the point cloud lane line annotation method provided by an embodiment of the present invention;

[0042] Figure 6 A schematic diagram of the system architecture of the point cloud lane line annotation system provided by an embodiment of the present invention;

[0043] Figure 7 A schematic diagram of the structure of the point cloud lane line annotation device provided by an embodiment of the present invention;

[0044] Figure 8 A schematic diagram of the structure of the electronic device provided by an embodiment of the present invention. Detailed Embodiments

[0045] The exemplary embodiments of the present invention will be described in more detail below with reference to the drawings. Although the exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present invention can be more thoroughly understood and the scope of the present invention can be fully conveyed to those skilled in the art.

[0046] In the development process of autonomous driving technology, accurate environmental perception ability is one of the core capabilities for achieving safe navigation. As an important aspect of environmental perception, point cloud lane line annotation is crucial for lane line detection and map construction in a Light Detection and Ranging (LiDAR) perception system. However, the current common point cloud lane line annotation methods face a series of challenges and limitations.

[0047] First, traditional annotation methods mainly rely on frame-by-frame annotation. This process is not only inefficient but also results in unstable accuracy of the overall annotation ground truth due to the inconsistency of annotation results in the overlapping areas of lane lines between adjacent frames. In this method, annotators need to judge the true position of lane lines based on the color differences caused by the reflection intensity in the point cloud data, which is both time-consuming and error-prone.

[0048] Second, although the annotation accuracy can be improved by assisting with the annotation results of image lane lines and transforming them into the point cloud coordinate system, this method highly depends on the spatio-temporal synchronization accuracy between the image and the point cloud data. If the synchronization accuracy is not high, the resulting point cloud lane line annotation may deviate significantly from the actual situation. In addition, since the detection distance of image lane line annotation is usually much smaller than that of point cloud lane lines, the method relying on image assistance may also weaken the detection performance of LiDAR.

[0049] Third, the maintenance condition of the road is also an important factor affecting the accuracy of point cloud lane line annotation. If the road has not been maintained for a long time, resulting in low reflection intensity, it is difficult for annotators to find the true point cloud of lane lines through the color differences in the point cloud. Although this problem can be solved by combining image lane line annotation, as mentioned above, this method still faces the limitation of spatio-temporal synchronization accuracy between the image and the point cloud data.

[0050] In summary, although point cloud lane line annotation is crucial for improving the environmental perception ability of autonomous vehicles, the current methods have problems such as low efficiency, unstable accuracy, and strong dependence on scene or data synchronization accuracy. The existence of these problems severely restricts the further development of autonomous driving technology. Therefore, it is particularly important to research and develop more efficient, more stable, and less scene-dependent point cloud lane line annotation technologies.

[0051] Based on this, in the embodiments of the present invention, a point cloud lane line annotation method is provided. First, a point cloud lane line map is constructed through multiple frames of images collected by target positioning points. Then, the lane line data of each image frame is extracted and transformed into a unified coordinate system using the map and positioning information. Finally, batch annotation is performed through a cloud annotation platform to generate the local point cloud lane line map of each image frame, thereby achieving efficient and accurate lane line annotation.

[0052] Such asFigure 1 As shown in the figure, a flowchart of an implementation manner of the point cloud lane line annotation method provided by an embodiment of the present invention, the method may include:

[0053] S100. Obtain a point cloud lane line map, where the point cloud lane line map is a point cloud map constructed based on the lane line perception information of multiple image frames collected based on a target positioning point.

[0054] Specifically, an embodiment of the present invention may use the lane line perception information of multiple image frames detected by a perception module mounted on a vehicle, and calculate the positioning information with the vehicle as the target positioning point, where the lane line perception information includes the coordinate information of the three-dimensional dynamic target and the image lane line detected by the perception module.

[0055] Among them, the perception module is a multi-modal input device that integrates laser point cloud data, camera image data, inertial navigation positioning data, etc. It can not only perceive dynamic targets, but also detect lane lines in images and calculate positioning information. Dynamic targets include various objects that move over time, such as cars, pedestrians, and bicycles.

[0056] An embodiment of the present invention may transform both the three-dimensional dynamic target and the coordinate information of the image lane line in the lane line perception information to the positioning point coordinate system with the target positioning point as the coordinate origin, where the target positioning point may be the center of the rear axle of the vehicle carrying the perception module.

[0057] An embodiment of the present invention may synchronize the lane line perception information and the positioning information according to the timestamp information of the image frame.

[0058] Optionally, as Figure 2 shown, the construction process of the point cloud lane line map provided by an embodiment of the present invention may include:

[0059] S200. Use the perception module to detect the image lane line and dynamic target in the image frame, remove the dynamic target in the image frame, and obtain the lane line attribute information of the image lane line, where the lane line attribute information includes the position, type, and color of the lane line.

[0060] S210. According to the projection relationship between the camera image and the lidar point cloud, project the lane line attribute information detected in the image frame onto the lidar point cloud.

[0061] S220. In the lidar point cloud, filter out the lane line measurement points belonging to the projection area range of the image lane line in the point cloud, and construct a point cloud lane line from the lane line measurement points to obtain a point cloud lane line map.

[0062] S110. Obtain the first point cloud lane line data of a single image frame by using the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map.

[0063] Among them, the positioning information is the position information of the target positioning point corresponding to the image frame in the point cloud lane line map.

[0064] Among them, the first point cloud lane line data includes multiple target lane line measurement points in the positioning point coordinate system with the target positioning point as the coordinate origin. The first point cloud lane line data usually includes the position coordinates and reflection intensities of the target lane line measurement points.

[0065] Among them, the target lane line measurement points are lane line measurement points within a preset distance range from the target positioning point.

[0066] Specifically, in the embodiment of the present invention, the lane line measurement points close to the target positioning point can be found in the point cloud lane line map through the positioning information as the target lane line measurement points, and the target lane line measurement points are converted from the map global coordinate system to the positioning point coordinate system with the target positioning point as the coordinate origin to generate the first point cloud lane line data of a single image frame, and the first point cloud lane line data is saved in a PCD (Point Cloud Data) format file. At the same time, the positioning information is saved as a JSON (JavaScript Object Notation) format file. The PCD format file and the JSON format file corresponding to the same picture frame share the same frame number for data synchronization.

[0067] S120. Convert the target lane line measurement points in the first point cloud lane line data of multiple image frames to the same target coordinate system to obtain the second point cloud lane line data of each image frame.

[0068] Specifically, in the embodiment of the present invention, by specifying the indexes of the starting frame and the ending frame, the PCD format files containing the first point cloud lane line data and the JSON format files containing the positioning information within the index range are selected. Then, the target lane line measurement points in the first point cloud lane line data of all selected PCD format files are uniformly converted to the same coordinate system to obtain the second point cloud lane line data, that is, the PCD format file currently saves the second point cloud lane line data.

[0069] The second point cloud lane line data usually includes the position coordinates and reflection intensities of the target lane line measurement points.

[0070] In the embodiment of the present invention, in the first point cloud lane line data, the laser point cloud data of each image frame has been transformed from the map global coordinate system to the positioning point coordinate system of the corresponding image frame, that is, it has been localized relative to the target positioning point of each image frame. However, since the target positioning point moves over time, the point coordinate systems of different image frames are different relative to the global coordinate system. Therefore, in order to splice the point cloud data of multiple image frames, it is necessary to transform the target lane line measurement points in the first point cloud lane line data of multiple image frames from the positioning point coordinate system to the same target coordinate system, so that the point cloud lane line data from different time points can be correctly aligned in the same reference frame, ensuring that the splicing of multi-frame point cloud lane lines can be completed subsequently.

[0071] Optionally, the target coordinate system can be the positioning point coordinate system with the target positioning point of any image frame as the coordinate origin.

[0072] Optionally, after the embodiment of the present invention transforms the target lane line measurement points in the first point cloud lane line data of multiple image frames from the positioning point coordinate system to the same target coordinate system, duplicate first point cloud lane line data can be de-duplicated. The embodiment of the present invention eliminates redundant information in the data set through de-duplication processing, thereby reducing noise, improving the accuracy of lane line detection and annotation, and further accelerating the processing speed of point cloud lane lines.

[0073] S130. Use the positioning information of multiple image frames and the second point cloud lane line data to perform batch annotation of lane lines on the cloud annotation platform to obtain the local map of the point cloud lane line of the target positioning point in each image frame.

[0074] The embodiment of the present invention can upload PCD format files containing the second point cloud lane line data and JSON format files containing positioning information of several image frames to the cloud annotation platform.

[0075] The annotation interface of the cloud annotation platform can be as Figure 3 shown. The cloud annotation platform has the ability to load the positioning information and point cloud lane line data related to multiple image frames and generate the corresponding local map of the point cloud lane line. Users or automated lane line detection programs can detect lane lines on the local map of the point cloud lane line loaded by the cloud annotation platform and perform annotation operations on the lane lines in the local map of the point cloud lane line based on the detection results, thereby realizing batch annotation of lane lines in multiple frames of images, and further improving the efficiency and accuracy of lane line annotation.

[0076] A method for labeling point cloud lane lines provided by the present invention obtains a point cloud lane line map, where the point cloud lane line map is a point cloud map constructed based on the lane line perception information of multiple image frames collected at target positioning points; using the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map, the first point cloud lane line data of a single image frame is obtained, where the positioning information is the position information corresponding to the image frame of the target positioning point in the point cloud lane line map, and the first point cloud lane line data includes multiple target lane line measurement points in a positioning point coordinate system with the target positioning point as the coordinate origin, and the target lane line measurement points are lane line measurement points within a preset distance range from the target positioning point; converting the target lane line measurement points in the first point cloud lane line data of multiple image frames from the positioning point coordinate system to the same target coordinate system to obtain the second point cloud lane line data of each image frame; using the positioning information and the second point cloud lane line data of multiple image frames to perform batch labeling of lane lines on the cloud annotation platform to obtain the local point cloud lane line map of the target positioning point in each image frame. The present invention utilizes a high-precision point cloud lane line map and positioning information to achieve batch labeling of multi-frame point cloud lane lines, making the point cloud lane line labeling process more efficient and automated, and improving the efficiency and accuracy of point cloud lane line labeling.

[0077] Optionally, based on one or more corresponding Figure 1 embodiments described above, in another optional embodiment provided by an embodiment of the present invention, obtaining the first point cloud lane line data of a single image frame by using the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map may specifically include:

[0078] Using the positioning information corresponding to a single image frame of the target positioning point in the point cloud lane line map, search for target lane line measurement points within a preset distance range from the target positioning point in the point cloud lane line map. Convert the target lane line measurement points to a positioning point coordinate system with the target positioning point as the coordinate origin to obtain the first point cloud lane line data of a single image frame.

[0079] An embodiment of the present invention can perform a nearest neighbor search in the point cloud lane line map based on the positioning information of the target positioning point in the point cloud lane line map, and select the point cloud data within a preset distance range from the target positioning point as the target lane line measurement points. Then, convert the coordinate system of the target lane line measurement points from the global map coordinate system to a positioning point coordinate system with the target positioning point as the coordinate origin to obtain the first point cloud lane line data of a single image frame, and save it as a PCD format file. At the same time, the positioning information is saved as a JSON format file. An embodiment of the present invention can synchronize the PCD format file and the JSON format file through the frame number of the image frame, that is, match the first point cloud lane line data and the positioning information of the same image frame.

[0080] It can be understood that the preset distance range can be set according to actual needs. By setting the preset distance range in the embodiments of the present invention, target lane line measurement points with a high correlation with the target positioning point can be effectively screened out, avoiding lane line measurement points with low correlation from participating in subsequent point cloud lane line annotation, and improving the efficiency of point cloud lane line annotation.

[0081] In the embodiments of the present invention, by using the target positioning point and the positioning information corresponding to a single image frame in the point cloud lane line map, target lane line measurement points within the preset distance range are searched out, and these measurement points are converted to the positioning point coordinate system with the target positioning point as the coordinate origin to obtain the point cloud lane line data of a single frame. It can ensure that each lane line measurement point is accurately positioned relative to the target positioning point. While enhancing the accuracy of the measurement data, the point cloud data converted to the positioning point coordinate system can more intuitively reflect the spatial relationship between the lane line and the vehicle, providing a consistent reference framework for the lane line data extracted from different image frames, facilitating subsequent cross-frame lane line comparison and fusion, and realizing the batch annotation function of multi-frame point cloud lane lines.

[0082] Optionally, based on Figure 1 the method shown, as Figure 4 shown, a schematic flow chart of another implementation manner of the point cloud lane line annotation method provided by the embodiments of the present invention. Step S130 may specifically include:

[0083] S400. Based on the positioning information of multiple image frames and the second point cloud lane line data, the cloud annotation platform loads the corresponding first point cloud lane line local map.

[0084] Specifically, the cloud annotation platform can confirm the synchronized positioning information and the second point cloud lane line data according to the frame number of the image frame, so that the target positioning point and the matching target lane line measurement points are accurately positioned. Then, using the point cloud fusion technology, the point cloud lane line data of multiple image frames are fused into a unified point cloud map, and various lane line features are identified and extracted from the fused point cloud lane line data, and the extracted lane line features are marked in the point cloud map to construct the lane lines in the point cloud map and generate the point cloud lane line local map.

[0085] S410. Perform a lane line aggregation annotation operation on each lane line measurement point in the first point cloud lane line local map to obtain a second point cloud lane line local map, where each lane line measurement point in the second point cloud lane line local map corresponds to lane line cluster identification information.

[0086] Among them, the lane line cluster identification information is a unique identifier for identifying the lane line to which the lane line measurement point belongs. The lane line cluster identification information can be a number, a letter, or a combination of the two.

[0087] Optionally, in the embodiments of the present invention, a lane line detection program can be used to cluster similar lane line measurement points in the local point cloud lane line map, that is, the lane line measurement points identified as the same lane line are divided into the same lane line cluster, and the same lane line cluster identification information is assigned to the lane line measurement points in the same lane line cluster.

[0088] Optionally, the embodiments of the present invention provide a manual clustering function. The user can use the "New" function of the cloud annotation platform to create a new annotation object, and in the lane line annotation, enclose all the lane line measurement points on the same lane line in a polygon envelope to form a visual boundary to ensure that each lane line is separately and accurately annotated. After the user confirms the lane line measurement points corresponding to each lane line, these lane line measurement points are divided into the same lane line cluster, and the same lane line cluster identification information is assigned to the lane line measurement points in the same lane line cluster. The lane line cluster identification information obtained after clustering will be updated to the corresponding point cloud lane line data, that is, the PCD format file.

[0089] Through the lane line aggregation annotation operation, the embodiments of the present invention can organize the lane line measurement points into a structured set, thereby accurately annotating each lane line in the local point cloud lane line map and improving the efficiency of point cloud lane line annotation.

[0090] Optionally, based on one or more corresponding embodiments described above, in another optional embodiment provided by the embodiments of the present invention, before performing the lane line aggregation annotation operation on the lane line measurement points in the first local point cloud lane line map to obtain the second local point cloud lane line map, the method may further include: Figure 4 Adding missing first lane line measurement points in the first local point cloud lane line map.

[0091]

[0092] Optionally, in the embodiments of the present invention, a lane line detection program can be used to detect whether there are missing first lane line measurement points in the local point cloud lane line map and automatically complete the first lane line measurement points.

[0093] Specifically, in the embodiments of the present invention, a lane line detection program can detect the continuity and consistency of the lane lines in the local point cloud lane line map, identify the gap positions of the lane lines, and generate and complete the missing first lane line measurement points at the gap positions using an interpolation method.

[0094] Optionally, in the embodiments of the present invention, historical lane line data and a trajectory prediction algorithm can be used to predict the continuation of the local point cloud lane line map, identify the missing first lane line measurement points, and automatically complete the first lane line measurement points through a B-spline optimization algorithm.

[0095]

[0095] The embodiment of the present invention can automatically detect and complete the missing lane line measurement points in the point cloud lane line local map through the lane line detection program, thereby improving the integrity and accuracy of the local map.

[0096] Optionally, this embodiment of the present invention provides a manual lane measurement point addition feature. Users can utilize the "Add Point" feature of the cloud annotation platform, using the cursor or a designated input device, to manually add a new first lane measurement point in the missed area, thereby completing any missed first lane measurement points in the point cloud lane local map.

[0097] Optional, in the above Figure 4 In another optional embodiment provided by the embodiments of the present invention, based on one or more corresponding embodiments, before performing a lane line aggregation and annotation operation on each lane line measurement point in the first point cloud lane line partial map to obtain the second point cloud lane line partial map, the method may further include:

[0098] Delete the erroneous second lane line measurement points in the first point cloud lane line local map.

[0099] Optionally, the embodiment of the present invention may use a lane line detection program to detect whether there is a second lane line measurement point that is misidentified in the point cloud lane line local map, and automatically delete the second lane line measurement point.

[0100] Specifically, embodiments of the present invention utilize a lane detection program to detect the geometric consistency (e.g., curvature, smoothness, and directionality) of lane lines in a point cloud lane map, identify second lane measurement points that deviate from the normal pattern, and delete these incorrectly identified second lane measurement points. This lane detection program enables automated detection and deletion of incorrectly identified second lane measurement points in a point cloud lane map, thereby improving map data accuracy.

[0101] Optionally, embodiments of the present invention provide a manual lane line measurement point deletion function. Users can use the "Delete Point" function of the cloud annotation platform, select the erroneous lane line measurement point to be deleted using a cursor or other input device, and then delete it by clicking the Delete button or using a shortcut key. This will delete the misdetected second lane line measurement point from the point cloud lane line local map.

[0102] It is understandable that, in a round of point cloud lane line labeling tasks, embodiments of the present invention may include labeling operations of both adding missed first lane line measurement points and deleting erroneous second lane line measurement points in the first point cloud lane line local map.

[0103] It can be understood that in one round of point cloud lane line annotation tasks, the inventive embodiments can add missing first lane line measurement points multiple times and / or delete incorrect second lane line measurement points multiple times in the first local point cloud lane line map.

[0104] By adding missing first lane line measurement points and deleting incorrect second lane line measurement points in the first local point cloud lane line map, the inventive embodiments contribute to the fine adjustment and optimization of the lane line annotation in the local point cloud lane line map, enabling the local point cloud lane line map to more accurately reflect the actual road conditions.

[0105] Based on the pre-annotation results of the point cloud lane lines, the inventive embodiments can assist users in querying the annotation ground truth, improving the accuracy and robustness of the annotation. At the same time, by utilizing the point cloud lane line map and positioning information, batch annotation of multiple frames of point cloud lane lines can be achieved, enhancing the annotation efficiency.

[0106] Optionally, based on Figure 4 the method shown, as Figure 5 shown, a flowchart of another implementation manner of the point cloud lane line annotation method provided by the inventive embodiments. After step S410, the method may further include:

[0107] S500. Obtain the third point cloud lane line data of a single image frame by using the positioning information of the target positioning point in the second local point cloud lane line map, where the third point cloud lane line data includes target lane line measurement points converted back from the target coordinate system to the positioning point coordinate system.

[0108] The inventive embodiments can combine the positioning information with the annotated local point cloud lane line map to determine the target positioning point of a single image frame, and use the point cloud data within a preset distance range from the target positioning point as the target lane line measurement points. At this time, the target lane line measurement points are in the target coordinate system. At this time, it is necessary to convert the target lane line measurement points from the target coordinate system to the positioning point coordinate system with the target positioning point as the coordinate origin to obtain the third point cloud lane line data of a single image frame, and update it to a PCD format file, that is, save the third point cloud lane line data as a PCD format file.

[0109] S510. Use the third point cloud lane line data and the lane line cluster identification information corresponding to the target lane line measurement points to perform lane line fitting on each lane line in a single image frame to obtain a lane line fitting result.

[0110] S520. Correspondingly save the lane line fitting result with the positioning information corresponding to a single image frame.

[0111] Specifically, embodiments of the present invention can extract the point cloud data of each lane line cluster based on the third point cloud lane line data and the lane line cluster identification information corresponding to the target lane line measurement points, apply a lane line fitting algorithm to fit the point cloud data of each lane line cluster, calculate a lane line mathematical model (usually a multi-order equation, for example: a cubic equation in one variable) that can represent the distribution of the point cloud data, extract the coefficients defining the geometric shape of the lane line in the lane line data model, and jointly save these coefficients and the corresponding positioning information in a JSON format file, that is, save the lane line fitting result and the positioning information corresponding to a single image frame in a JSON format file, ensuring that the equation fitted with these coefficients can accurately describe the curvature and trend of the actual lane line. Among them, the lane line fitting algorithm can be the least squares method.

[0112] Embodiments of the present invention can obtain a lane line fitting result closer to the actual lane line shape through more accurate mathematical fitting of each lane line in a single image frame, so as to facilitate subsequent various applications of the lane line data. At the same time, combining and saving the lane line fitting result with the positioning information of the corresponding image frame can ensure the consistency and traceability of the data, improve the processing speed of the point cloud lane line data, and further improve the efficiency and accuracy of the point cloud lane line annotation.

[0113] Optionally, based on one or more corresponding embodiments described above Figure 5 In another optional embodiment provided by embodiments of the present invention, after obtaining the third point cloud lane line data of a single image frame by using the positioning information of the target positioning point in the local map of the second point cloud lane line, the method may further include:

[0114] Using the positioning information of multiple image frames and the third point cloud lane line data, re-perform batch annotation of lane lines on the cloud annotation platform to obtain the local map of the third point cloud lane line of the target positioning point in each image frame, where each lane line measurement point in the local map of the third point cloud lane line corresponds to lane line cluster identification information.

[0115] In practical applications, due to reasons such as insufficient annotation accuracy or changes in label information (for example, the previous type A lane line is redefined as type B), it may be necessary to re-annotate the point cloud lane line. Embodiments of the present invention can use the positioning information of a single image frame and the third point cloud lane line data to load the local map of the point cloud lane line that needs to be re-annotated in the cloud annotation platform and re-perform batch annotation of lane lines.

[0116] Embodiments of the present invention can support repeated multi-frame batch annotation of the same set of point cloud lane lines by loading the positioning information of multiple image frames and the corresponding local map of the third point cloud lane line data, thereby further improving the efficiency of point cloud lane line annotation.

[0117] Although the operations are depicted in a particular order, this should not be construed as requiring that the operations be performed in the particular order shown or in sequential order. In certain circumstances, multitasking and parallel processing may be advantageous.

[0118] It should be understood that the various steps recited in the method embodiments of the present invention can be executed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit steps shown. The scope of the present invention is not limited in this regard.

[0119] Corresponding to the above method embodiments, an embodiment of the present invention further provides a point cloud lane line annotation system, and its system architecture is as Figure 6 shown, and may include: a single-frame semantic processing module, a point cloud lane line module, and an annotation module. The single-frame semantic processing module can perform point cloud three-dimensional target detection perception, image lane line detection perception, and positioning module perception through the original input point cloud data, image data, and inertial navigation (Inertial Navigation System, INS) data, and can extract the original point cloud, three-dimensional target, image lane line, and positioning information. The point cloud lane line module obtains the point cloud lane line label and the result form of the single-frame point cloud lane line through operations such as single-frame lane line point cloud extraction, point cloud stitching, map splitting into single frames, and clustering fitting. The annotation module uses the result form to perform multiple single-frame restoration of the map, provides functions for adding, deleting, modifying, and querying point cloud data, and a manual lane line clustering function. After splitting the map into single frames, equation fitting is performed, and then the fitting result is fed back to the result form to repeat multi-frame batch annotation to improve the efficiency of point cloud lane line annotation.

[0120] The embodiment of the present invention collects the point cloud lane line information and its positioning information within a single frame through the point cloud lane line map and positioning information, and uploads this information to the annotation platform. On the annotation platform, by setting the starting frame and the ending frame, batch annotation of multi-frame point cloud lane line information is realized. This includes batch editing, modification, viewing, and clustering operations on multi-frame lane line information. After the annotation process is completed, the annotated point cloud lane line information is decomposed back into individual frames for subsequent application in data closed-loop links such as scene twinning and scene simulation. The real road scene and lane lines are simulated through the fitted lane line coefficients, and at the same time, repeated annotation operations on the same data set are supported.

[0121] Corresponding to the above method embodiments, an embodiment of the present invention further provides a point cloud lane line annotation device, and its structure is as Figure 7 shown, and may include: a point cloud lane line map obtaining unit 100, a first point cloud lane line data obtaining unit 200, a second point cloud lane line data obtaining unit 300, and a lane line annotation unit 400.

[0122] The point cloud lane line map acquisition unit 100 is configured to acquire a point cloud lane line map, where the point cloud lane line map is a point cloud map constructed based on the lane line perception information of multiple image frames collected at a target positioning point.

[0123] The first point cloud lane line data acquisition unit 200 is configured to use the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map to acquire the first point cloud lane line data of a single image frame, where the positioning information is the position information corresponding to the image frame of the target positioning point in the point cloud lane line map, and the first point cloud lane line data includes multiple target lane line measurement points in the positioning point coordinate system with the target positioning point as the coordinate origin, and the target lane line measurement points are lane line measurement points within a preset distance range from the target positioning point.

[0124] The second point cloud lane line data acquisition unit 300 is configured to convert the target lane line measurement points in the first point cloud lane line data of multiple image frames from the positioning point coordinate system to the same target coordinate system to acquire the second point cloud lane line data of each image frame.

[0125] The lane line annotation unit 400 is configured to use the positioning information and the second point cloud lane line data of multiple image frames to perform batch lane line annotation on a cloud annotation platform to acquire the local point cloud lane line map of the target positioning point in each image frame.

[0126] Optionally, the first point cloud lane line data acquisition unit 200 may specifically be configured to use the positioning information corresponding to a single image frame of the target positioning point in the point cloud lane line map to search for target lane line measurement points within a preset distance range from the target positioning point in the point cloud lane line map; and convert the target lane line measurement points to the positioning point coordinate system with the target positioning point as the coordinate origin to acquire the first point cloud lane line data of a single image frame.

[0127] Optionally, the lane line annotation unit 400 may include: a local point cloud lane line map loading subunit and a lane line aggregation annotation subunit.

[0128] The local point cloud lane line map loading subunit is configured to load the corresponding first local point cloud lane line map from the cloud annotation platform based on the positioning information and the second point cloud lane line data of multiple image frames.

[0129] The lane line aggregation annotation subunit is configured to perform a lane line aggregation annotation operation on each lane line measurement point in the first local point cloud lane line map to acquire a second local point cloud lane line map, where each lane line measurement point in the second local point cloud lane line map corresponds to lane line cluster identification information.

[0130] Optionally, the point cloud lane line annotation device may further include: a lane line measurement point addition annotation unit and a lane line measurement point deletion annotation unit.

[0131] The lane line measurement point addition annotation unit is configured to add missing first lane line measurement points to the first local point cloud lane line map before the lane line aggregation annotation subunit performs lane line aggregation annotation operations on each lane line measurement point in the first local point cloud lane line map to obtain a second local point cloud lane line map.

[0132] The lane line measurement point deletion annotation unit is configured to delete incorrect second lane line measurement points from the first local point cloud lane line map before the lane line aggregation annotation subunit performs lane line aggregation annotation operations on each lane line measurement point in the first local point cloud lane line map to obtain a second local point cloud lane line map.

[0133] Optionally, the point cloud lane line annotation device may further include: a third point cloud lane line data acquisition unit, a lane line fitting unit, and a storage unit.

[0134] The third point cloud lane line data acquisition unit is configured to, after the lane line aggregation annotation subunit performs lane line aggregation annotation operations on each lane line measurement point in the first local point cloud lane line map to obtain a second local point cloud lane line map, obtain third point cloud lane line data of a single image frame by using the positioning information of the target positioning point in the second local point cloud lane line map, where the third point cloud lane line data includes target lane line measurement points converted back from the target coordinate system to the positioning point coordinate system.

[0135] The lane line fitting unit is configured to perform lane line fitting on each lane line in a single image frame by using the third point cloud lane line data and the lane line cluster identification information corresponding to the target lane line measurement points to obtain a lane line fitting result.

[0136] The storage unit is configured to correspondingly store the lane line fitting result and the positioning information corresponding to the single image frame.

[0137] Optionally, the point cloud lane line annotation device may further include: a lane line re-annotation unit.

[0138] The lane line re-annotation unit is configured to, after the third point cloud lane line data acquisition unit obtains the third point cloud lane line data of a single image frame by using the positioning information of the target positioning point in the second local point cloud lane line map, re-perform batch lane line annotation on the cloud annotation platform by using the positioning information of multiple image frames and the third point cloud lane line data to obtain a third local point cloud lane line map of the target positioning point in each image frame, where each lane line measurement point in the third local point cloud lane line map corresponds to lane line cluster identification information.

[0139] Optionally, the third point cloud lane line data is saved as a PCD format file, and the lane line fitting result and the positioning information corresponding to a single image frame are saved as a JSON format file.

[0140] A point cloud lane line annotation device provided by the present invention obtains a point cloud lane line map, wherein the point cloud lane line map is a point cloud map constructed based on the lane line perception information of multiple image frames collected by a target positioning point; uses the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map to obtain the first point cloud lane line data of a single image frame, wherein the positioning information is the position information corresponding to the image frame of the target positioning point in the point cloud lane line map, and the first point cloud lane line data includes multiple target lane line measurement points in a positioning point coordinate system with the target positioning point as the coordinate origin, and the target lane line measurement points are lane line measurement points within a preset distance range from the target positioning point; converts the target lane line measurement points in the first point cloud lane line data of multiple image frames from the positioning point coordinate system to the same target coordinate system to obtain the second point cloud lane line data of each image frame; uses the positioning information and the second point cloud lane line data of multiple image frames to perform batch annotation of lane lines on a cloud annotation platform to obtain the point cloud lane line local map of the target positioning point in each image frame. The present invention uses a high-precision point cloud lane line map and positioning information to achieve batch annotation of multi-frame point cloud lane lines, making the point cloud lane line annotation process more efficient and automated, and improving the efficiency and accuracy of point cloud lane line annotation.

[0141] Regarding the device in the above embodiments, the specific manner in which each unit performs operations has been described in detail in the embodiments related to the method, and will not be elaborated here.

[0142] The point cloud lane line annotation device includes a processor and a memory. The above-mentioned point cloud lane line map obtaining unit 100, the first point cloud lane line data obtaining unit 200, the second point cloud lane line data obtaining unit 300, the lane line annotation unit 400, etc. are all stored in the memory as program units, and the processor executes the above program units stored in the memory to implement corresponding functions.

[0143] The processor contains a kernel, and the kernel retrieves the corresponding program units from the memory. One or more kernels can be set. By adjusting the kernel parameters, a point cloud lane line map is constructed by integrating multi-frame image data collected by a target positioning point, and the lane line data of each image frame is accurately obtained and converted by using the positioning information, and batch annotation of point cloud lane lines is realized on a cloud annotation platform, thereby improving the accuracy and efficiency of point cloud lane line annotation.

[0144] An embodiment of the present invention provides a computer-readable storage medium, on which a program is stored, and when the program is executed by a processor, the point cloud lane line annotation method is implemented.

[0145] An embodiment of the present invention provides a processor for running a program, wherein when the program runs, it executes the point cloud lane marking method.

[0146] As Figure 8 shown, an embodiment of the present invention provides an electronic device 1000, which includes at least one processor 1001, at least one memory 1002 connected to the processor 1001, and a bus 1003; wherein the processor 1001 and the memory 1002 communicate with each other through the bus 1003; the processor 1001 is used to call program instructions in the memory 1002 to execute the above-mentioned point cloud lane marking method. The electronic device herein may be a server, a PC, a PAD, a mobile phone, an ECU (Electronic Control Unit), a VCU (Vehicle Control Unit), an MCU (Micro Controller Unit), an HCU (Hybrid Control Unit), etc.

[0147] The present invention also provides a computer program product, which is suitable for executing a program initialized with the steps of the point cloud lane marking method when executed on an electronic device.

[0148] The present invention is described with reference to the flowcharts and / or block diagrams of methods, devices, systems, electronic devices, and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable devices generate a device for implementing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0149] In a typical configuration, an electronic device includes one or more processors (CPUs), a memory, and a bus. The electronic device may also include an input / output interface, a network interface, etc.

[0150] The memory may include non-permanent memory in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash memory (flash RAM), and the memory includes at least one storage chip. The memory is an example of a computer-readable medium.

[0151] A computer-readable medium includes permanent and non-permanent, removable and non-removable media and can implement information storage by any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassette tapes, magnetic disk storage or other magnetic storage devices, or any other non-transitory medium that can be used to store information accessible by a computing device. As defined herein, a computer-readable medium does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0152] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present invention are all information and data that have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with the relevant laws, regulations, and standards of relevant countries and regions.

[0153] In the description of the present invention, it should be understood that if terms such as "upper", "lower", "front", "rear", "left", and "right" are used to indicate the orientation or positional relationship, they are based on the orientation or positional relationship shown in the drawings. This is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the indicated position or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of the present invention.

[0154] It should be noted that in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. It should also be noted that the term "comprising", "including", or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, commodity, or device comprising a series of elements not only includes those elements, but also includes other elements not explicitly listed, or elements inherent to such process, method, commodity, or device. Without further limitation, an element defined by the statement "comprising one..." does not exclude the existence of additional identical elements in the process, method, commodity, or device comprising the element.

[0155] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0156] The above are only the embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, various changes and modifications can be made to the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the scope of the present invention.

Claims

1. A method for labeling lane lines of point clouds, characterized in that, Including: Obtain a point cloud lane line map, where the point cloud lane line map is a point cloud map constructed based on the lane line perception information of multiple image frames collected at target positioning points; Using the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map, obtain the first point cloud lane line data of a single image frame, where the positioning information is the position information corresponding to the image frame of the target positioning point in the point cloud lane line map, and the first point cloud lane line data includes multiple target lane line measurement points in a positioning point coordinate system with the target positioning point as the coordinate origin, and the target lane line measurement points are lane line measurement points within a preset distance range from the target positioning point; Convert the target lane line measurement points in the first point cloud lane line data of multiple image frames from the positioning point coordinate system to the same target coordinate system to obtain the second point cloud lane line data of each image frame; Using the positioning information and the second point cloud lane line data of multiple image frames, perform batch lane line annotation on a cloud annotation platform to obtain the point cloud lane line local map of the target positioning point in each image frame.

2. The method according to claim 1, wherein The step of using the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map to obtain the first point cloud lane line data of a single image frame includes: Using the positioning information corresponding to a single image frame of the target positioning point in the point cloud lane line map, search for target lane line measurement points within a preset distance range from the target positioning point in the point cloud lane line map; Convert the target lane line measurement points to a positioning point coordinate system with the target positioning point as the coordinate origin to obtain the first point cloud lane line data of a single image frame.

3. The method according to claim 1, wherein The step of using the positioning information and the second point cloud lane line data of multiple image frames to perform batch lane line annotation on a cloud annotation platform to obtain the point cloud lane line local map of the target positioning point in each image frame includes: Based on the positioning information and the second point cloud lane line data of multiple image frames, load the corresponding first point cloud lane line local map by the cloud annotation platform; Perform a lane line aggregation annotation operation on each lane line measurement point in the first point cloud lane line local map to obtain a second point cloud lane line local map, where each lane line measurement point in the second point cloud lane line local map corresponds to lane line cluster identification information.

4. The method according to claim 3, characterized in that, Before performing the lane line aggregation annotation operation on each lane line measurement point in the first point cloud lane line local map to obtain the second point cloud lane line local map, the method further includes: Adding missing first lane line measurement points in the first point cloud lane line local map; And / or, deleting incorrect second lane line measurement points in the first point cloud lane line local map.

5. The method according to claim 3 or 4, characterized in that, After performing the lane line aggregation annotation operation on each lane line measurement point in the first point cloud lane line local map to obtain the second point cloud lane line local map, the method further includes: Using the positioning information of the target positioning point in the second point cloud lane line local map, obtain the third point cloud lane line data of a single image frame, where the third point cloud lane line data includes the target lane line measurement points converted back from the target coordinate system to the positioning point coordinate system; Using the third point cloud lane line data and the lane line cluster identification information corresponding to the target lane line measurement points, perform lane line fitting on each lane line in a single image frame to obtain a lane line fitting result; Correspondingly save the lane line fitting result and the positioning information corresponding to a single image frame.

6. The method according to claim 5, wherein After obtaining the third point cloud lane line data of a single image frame by using the positioning information of the target positioning point in the second point cloud lane line local map, the method further includes: Using the positioning information and the third point cloud lane line data of multiple image frames, re-perform batch annotation of lane lines on the cloud annotation platform to obtain the third point cloud lane line local map of the target positioning point in each image frame, where each lane line measurement point in the third point cloud lane line local map corresponds to lane line cluster identification information.

7. The method according to claim 5, wherein The third point cloud lane line data is saved as a PCD format file, and the lane line fitting result and the positioning information corresponding to a single image frame are saved as a JSON format file.

8. A point cloud lane marking device, characterized in that, Including: A point cloud lane line map obtaining unit, a first point cloud lane line data obtaining unit, a second point cloud lane line data obtaining unit, and a lane line annotation unit, The point cloud lane line map obtaining unit is configured to obtain a point cloud lane line map, where the point cloud lane line map is a point cloud map constructed based on the lane line perception information of multiple image frames collected by a target positioning point; The first point cloud lane line data obtaining unit is configured to use the point cloud lane line map and the positioning information of the target positioning point in the point cloud lane line map to obtain the first point cloud lane line data of a single image frame, where the positioning information is the position information corresponding to the image frame of the target positioning point in the point cloud lane line map, and the first point cloud lane line data includes multiple target lane line measurement points in the positioning point coordinate system with the target positioning point as the coordinate origin, and the target lane line measurement points are lane line measurement points within a preset distance range from the target positioning point; The second point cloud lane line data obtaining unit is configured to convert the target lane line measurement points in the first point cloud lane line data of multiple image frames from the positioning point coordinate system to the same target coordinate system to obtain the second point cloud lane line data of each image frame; The lane line annotation unit is configured to use the positioning information and the second point cloud lane line data of multiple image frames to perform batch annotation of lane lines on the cloud annotation platform to obtain the point cloud lane line local map of the target positioning point in each image frame.

9. A computer-readable storage medium having a program stored thereon, characterized in that, When the program is executed by a processor, it implements the point cloud lane line annotation method according to any one of claims 1 to 7.

10. An electronic device, characterized in that, The electronic device includes at least one processor, at least one memory connected to the processor, and a bus; wherein, the processor and the memory complete communication with each other through the bus; the processor is configured to call program instructions in the memory to execute the point cloud lane line annotation method according to any one of claims 1 to 7.

Images