A lane line vectorization method and system for camera image and laser point cloud fusion

Abstract

The application provides a lane line vectorization method and system based on camera image and laser point cloud fusion, and the method comprises the following steps: extracting ground information based on laser point cloud data; obtaining a lane line point cloud data set through image perception lane line semantic information fusion segmentation and laser point cloud ground information; dividing the lane line point cloud data set into lane broken line point cloud data and lane solid line point cloud data according to the lane line attribute information of each three-dimensional point cloud; processing the lane broken line point cloud data and the lane solid line point cloud data respectively, extracting the skeleton lines of the lane broken line point cloud data and the lane solid line point cloud data, and performing lane line fitting on the skeleton lines to obtain a lane vectorization result. The method provided by the application directly uses the lane line point cloud data obtained through image semantic information fusion and laser point cloud to perform vectorization, and can effectively solve the uncertainty of lane line vectorization at a lane intersection in the prior art.

Description

Technical Field

[0001] This invention relates to the field of image laser processing, and more specifically, to a method and system for lane line vectorization by fusing camera images and laser point clouds. Background Technology

[0002] With the rapid development of autonomous driving technology, the demand for high-definition maps is increasing, as they provide reliable and robust prior information on the static aspects of the traffic environment. High-definition maps have many layers, including various linear road elements.

[0003] The lower layers consist of physically existing elements (e.g., road boundaries, road curbs, and lane lines), while the higher layers contain virtual elements (e.g., road lane centerlines). All the aforementioned high-resolution map layers are recorded in vector data format (e.g., graphics with vertices and edges).

[0004] Lane lines, as a crucial element of high-definition maps, are typically used to prevent potential collisions and ensure vehicle safety. They contain all the topological information of the road, making them significant for downstream tasks such as vehicle planning, prediction, and control. Manually annotating lane lines in high-definition maps is labor-intensive, expensive, and inefficient, severely limiting the widespread application of autonomous driving systems. Due to the complexity of lane topology, especially in scenarios such as lane intersections and median strips, previous research has rarely addressed the vectorization of lane lines in these complex scenarios. Summary of the Invention

[0005] This invention addresses the technical problems existing in the prior art by providing a method and system for vectorizing lane lines by fusing camera images and laser point clouds.

[0006] According to a first aspect of the present invention, a method for lane line vectorization by fusing camera images and laser point clouds is provided, comprising:

[0007] Ground information is extracted based on laser point cloud data;

[0008] Based on the fusion and segmentation of image-perceived lane line semantic information and laser point cloud ground information, a lane line point cloud data set is obtained, wherein each three-dimensional point cloud in the lane line point cloud data set is marked with lane line attribute information.

[0009] Based on the lane line attribute information of each 3D point cloud, the lane line point cloud data set is divided into lane dashed line point cloud data and lane solid line point cloud data.

[0010] The dashed lane point cloud data is processed to extract the first skeleton line of the dashed lane point cloud data, and lane lines are fitted based on the first skeleton line to obtain the vectorized result of the dashed lane line; and the solid lane point cloud data is processed to extract the second skeleton line of the solid lane point cloud data, and lane lines are fitted based on the second skeleton line to obtain the vectorized result of the solid lane line.

[0011] Based on the above technical solution, the present invention can also be improved as follows.

[0012] Optionally, the extraction of ground information based on laser point cloud data includes:

[0013] Based on the average elevation of the ground and the ground model, ground information within the lane range is extracted from the laser point cloud data.

[0014] Optionally, the step of fusing and segmenting lane line semantic information from image perception with laser point cloud ground information to obtain a lane line point cloud data set includes:

[0015] The laser point cloud data is projected onto the image-aware lane line semantic segmentation map. Based on the lane line semantic category, the laser point cloud falling within the lane semantic image block is assigned the corresponding lane dashed line attribute information and lane solid line attribute information, thus completing the lane line point cloud segmentation of the laser point cloud and obtaining the lane line point cloud data set.

[0016] Optionally, the step of fusing and segmenting the lane line semantic information perceived from the image with the laser point cloud ground information to obtain the lane line point cloud data set further includes:

[0017] Based on the ground information, noise filtering is performed on the lane line point cloud in the laser point cloud data set to remove non-lane line point cloud data.

[0018] Optionally, the step of processing the lane dashed line point cloud data, extracting the first skeleton line of the lane dashed line point cloud data, and performing lane line fitting based on the first skeleton line to obtain the lane dashed line vectorization result includes:

[0019] Extract the first lane vector based on the lane dashed line data;

[0020] Calculate a first rotation matrix based on the first lane vector and the vehicle's driving direction, and rotate the lane dashed line to the vehicle's driving direction based on the first rotation matrix;

[0021] The geometric topological relationship of lane dashed lines is calculated based on the histogram statistical method to obtain the set of lane dashed lines on the same side of the same lane;

[0022] Extract the first skeleton line of the set of dashed lines of the lanes on the same side, and rotate the first skeleton line back to its original direction according to the first rotation matrix;

[0023] Lane line fitting is performed based on the first skeleton line rotated back to its original direction to obtain the vectorized result of the lane dashed line.

[0024] Optionally, the process of processing the lane solid line point cloud data, extracting the second skeleton line of the lane solid line point cloud data, and fitting lane lines based on the second skeleton line to obtain the lane solid line vectorization result includes:

[0025] Principal component analysis was performed on the solid lane point cloud data to distinguish between straight and curved segments based on the eigenvalue relationships.

[0026] If it is a curved segment, based on the geometric positional relationship between the centroid of the solid lane point cloud data and the line connecting the near and far endpoints of the vehicle's driving direction, the curved segment is divided into multiple sub-segments. Each sub-segment is processed in the same way as a straight segment to extract a second skeleton line, and lane lines are fitted based on the second skeleton line to obtain the vectorized result of the solid lane line.

[0027] Optionally, the step of extracting the second skeleton line according to the processing method of straight line segments, and performing lane line fitting based on the second skeleton line to obtain the lane solid line vectorization result includes:

[0028] Extract the corresponding second lane vector from the solid lane point cloud data, and compare whether the angle between the second lane vector and the vehicle's driving direction is greater than a preset threshold.

[0029] If the value is greater than a preset threshold, a second rotation matrix is ​​calculated based on the second lane vector and the vehicle's driving direction, and the lane solid line point cloud data is rotated to the vehicle's driving direction based on the second rotation matrix.

[0030] Extract the second skeleton line from the solid lane point cloud data, and rotate the second skeleton line to its original direction according to the second rotation matrix;

[0031] The second skeleton line rotated to its original direction is fitted with lane lines to obtain the solid line vectorization result of the lane lines;

[0032] If the value is less than or equal to a preset threshold, the third skeleton line of the solid lane point cloud data is extracted, and lane line fitting is performed on the third skeleton line to obtain the vectorized result of the solid lane line.

[0033] According to a second aspect of the present invention, a lane line vectorization system for fusing camera images and laser point clouds is provided, comprising:

[0034] The extraction module is used to extract ground information based on laser point cloud data;

[0035] The fusion segmentation module is used to fuse and segment lane line semantic information and laser point cloud ground information based on image perception to obtain a lane line point cloud data set, wherein each three-dimensional point cloud in the lane line point cloud data set is marked with lane line attribute information.

[0036] The partitioning module is used to partition the lane line point cloud data set into lane dashed line point cloud data and lane solid line point cloud data according to the lane line attribute information of each 3D point cloud.

[0037] The vectorization processing module is used to process the lane dashed line point cloud data, extract the first skeleton line of the lane dashed line point cloud data, and perform lane line fitting based on the first skeleton line to obtain the lane dashed line vectorization result; and to process the lane solid line point cloud data, extract the second skeleton line of the lane solid line point cloud data, and perform lane line fitting based on the second skeleton line to obtain the lane solid line vectorization result.

[0038] According to a third aspect of the present invention, an electronic device is provided, including a memory and a processor, the processor being configured to execute a computer management program stored in the memory, and to implement a lane line vectorization method for fusing camera images and laser point clouds.

[0039] According to a fourth aspect of the present invention, a computer-readable storage medium is provided having a computer management class program stored thereon, which, when executed by a processor, implements the steps of a lane line vectorization method for fusing camera images and laser point clouds.

[0040] This invention provides a method and system for lane line vectorization by fusing camera images and laser point clouds. The method involves fusing and segmenting lane line semantic information from image perception with ground information from laser point clouds to obtain a lane line point cloud dataset. Based on the lane line attribute information of each 3D point cloud, the dataset is divided into dashed lane line point cloud data and solid lane line point cloud data. The dashed and solid lane line point cloud data are processed separately to extract skeleton lines, which are then fitted to lane lines to obtain the vectorized lane line result. This method directly utilizes the lane line point cloud data obtained by fusing image semantic information with laser point clouds for vectorization, effectively addressing the uncertainties in lane line vectorization at lane intersections using existing technologies. Attached Figure Description

[0041] Figure 1 A flowchart of a lane line vectorization method for fusing camera images and laser point clouds provided by the present invention;

[0042] Figure 2 A schematic diagram of the overall process for vectorizing lane lines by fusing camera images and laser point clouds;

[0043] Figure 3 A schematic diagram of a lane line vectorization system for fusing camera images and laser point clouds provided by the present invention;

[0044] Figure 4 A schematic diagram of the hardware structure of a possible electronic device provided by the present invention;

[0045] Figure 5 This is a schematic diagram of the hardware structure of a possible computer-readable storage medium provided by the present invention. Detailed Implementation

[0046] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. In addition, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined with each other to form feasible technical solutions. Such combinations are not constrained by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0047] Figure 1 A flowchart of a lane line vectorization method for fusing camera images and laser point clouds provided by the present invention is shown below. Figure 1 As shown, the method includes:

[0048] Step 1: Extract ground information based on laser point cloud data.

[0049] Understandably, in this invention, three-dimensional laser point cloud data and image-aware semantic data of the ground are acquired simultaneously, and ground information is extracted from the laser point cloud data. Specifically, ground information within the lane range is extracted from the laser point cloud data based on the average elevation of the ground and the ground model.

[0050] Step 2: Based on the fusion and segmentation of the image-perceived lane line semantic information and the laser point cloud ground information, a lane line point cloud data set is obtained, wherein each three-dimensional point cloud in the lane line point cloud data set is marked with lane line attribute information.

[0051] Understandably, the laser point cloud data is projected onto the image-aware lane line semantic segmentation map. Then, based on the lane line semantic category, the laser point clouds falling within the lane semantic image block are assigned corresponding lane dashed line attribute information and lane solid line attribute information, thus completing the lane line point cloud segmentation of the laser point cloud. Next, the lane line point cloud is denoised and filtered based on ground information to remove non-lane point clouds, resulting in a lane line point cloud data set, where each 3D point cloud in the set carries the dashed and solid attribute information of the lane line.

[0052] Step 3: Based on the lane line attribute information of each 3D point cloud, divide the lane line point cloud data set into lane dashed line point cloud data and lane solid line point cloud data.

[0053] Understandably, since dashed lane lines are discontinuous and much shorter than solid lane lines, lane line vectorization requires processing in two separate pipelines: dashed and solid lane lines. Based on the lane line attribute information of each 3D point cloud in the lane line point cloud dataset, the dataset is divided into dashed lane point cloud data and solid lane point cloud data, and then processed separately for each.

[0054] Step 4: Process the lane dashed line point cloud data, extract the first skeleton line of the lane dashed line point cloud data, and perform lane line fitting based on the first skeleton line to obtain the lane dashed line vectorization result; and process the lane solid line point cloud data, extract the second skeleton line of the lane solid line point cloud data, and perform lane line fitting based on the second skeleton line to obtain the lane solid line vectorization result.

[0055] Among them, see Figure 2 This is a flowchart of the overall process for lane line vectorization. As an example, the process involves processing the lane dashed line point cloud data, extracting the first skeleton line from the lane dashed line point cloud data, and fitting lane lines based on the first skeleton line to obtain the lane dashed line vectorization result. This includes: extracting a first lane vector based on the lane dashed line data; calculating a first rotation matrix based on the first lane vector and the vehicle's driving direction; rotating the lane dashed lines to the vehicle's driving direction based on the first rotation matrix; calculating the geometric topological relationship of the lane dashed lines based on histogram statistics to obtain a set of lane dashed lines on the same side of the same lane; extracting the first skeleton line from the set of lane dashed lines on the same side, and rotating the first skeleton line back to its original direction based on the first rotation matrix; and fitting lane lines based on the first skeleton line rotated back to its original direction to obtain the lane dashed line vectorization result.

[0056] Understandably, because lane dashed lines are relatively short and often located in the middle of a road segment, and lane widths are relatively stable, they are more representative of lane direction within a certain range. Therefore, lane vector information can be extracted based on lane dashed line data first. Since vehicles do not maintain the same driving direction as the lane during travel, directly using lane point cloud data for lane dashed line topological connections would be difficult. Therefore, rotating the lane dashed line data to the vehicle's driving direction effectively reduces the lateral span of the lane dashed line plane in 2D, which facilitates topological connections of lane dashed lines on the same side of the same lane. Next, histogram statistics are used to determine lane dashed line sets based on the lateral aggregation range of the lane dashed line point cloud, thus obtaining the topological connection relationship of the lane dashed lines. Then, based on the topological relationship, skeleton lines are extracted for each lane dashed line set, and then the skeleton line point cloud is rotated back to the original coordinate system for lane dashed line curve fitting, obtaining the vectorized lane dashed line result.

[0057] As an example, the lane solid line point cloud data is processed to extract the second skeleton line of the lane solid line point cloud data, and lane line fitting is performed based on the second skeleton line to obtain the lane solid line vectorization result. This includes: performing principal component analysis on the lane solid line point cloud data to distinguish between straight segments and curved segments based on the eigenvalue relationship; if it is a curved segment, the curved segment is divided into multiple sub-segments based on the geometric positional relationship between the centroid of the lane solid line point cloud data and the near and far endpoints of the vehicle's driving direction; the second skeleton line is extracted from each sub-segment in the same way as the straight segment, and lane line fitting is performed based on the second skeleton line to obtain the lane solid line vectorization result.

[0058] The process involves extracting a second skeleton line using the same processing method as straight line segments, and then fitting lane lines based on the second skeleton line to obtain a solid lane vectorized result. This includes: extracting the corresponding second lane vector from the solid lane point cloud data, and comparing whether the angle between the second lane vector and the vehicle's driving direction is greater than a preset threshold; if it is greater than the preset threshold, calculating a second rotation matrix based on the second lane vector and the vehicle's driving direction, and rotating the solid lane point cloud data to the vehicle's driving direction based on the second rotation matrix; extracting the second skeleton line from the solid lane point cloud data, and rotating the second skeleton line back to its original direction based on the second rotation matrix; fitting lane lines to the second skeleton line rotated to its original direction to obtain a solid lane vectorized result; if it is less than or equal to a preset threshold, extracting the third skeleton line from the solid lane point cloud data, and fitting lane lines to the third skeleton line to obtain a solid lane vectorized result.

[0059] Understandably, solid lane lines in general scenarios such as intersections, ramps, and median strips exhibit unstable lane widths, intersections of two lane lines, and irregular outer bounding boxes of median strips. By performing principal component analysis on the lane solid line point cloud, based on eigenvalue representation, these irregularities are categorized as "curves," while other cases are categorized as "straight lines." Lane solid lines categorized as "straight lines" are then assessed for rotation based on the angle between the lane line direction and the vehicle's driving direction, followed by curve fitting to obtain the vectorized lane solid line result. Lane solid lines categorized as "curves" require lane line segmentation. This segmentation primarily involves re-dividing the lane solid line point cloud into multiple subsets based on the geometric positional relationship between the centroid of the lane line point cloud data and the lines connecting the near and far endpoints of the vehicle's driving direction. These subsets are then subjected to curve fitting following the "straight line" processing logic, ultimately yielding the vectorized result.

[0060] See Figure 3 This invention provides a lane line vectorization system for fusing camera images and laser point clouds. The system includes an extraction module 301, a fusion and segmentation module 302, a division module 303, and a vectorization processing module 304, wherein:

[0061] Extraction module 301 is used to extract ground information based on laser point cloud data;

[0062] The fusion segmentation module 302 is used to fuse and segment lane line semantic information and laser point cloud ground information based on image perception to obtain a lane line point cloud data set, wherein each three-dimensional point cloud in the lane line point cloud data set is marked with lane line attribute information.

[0063] The partitioning module 303 is used to partition the lane line point cloud data set into lane dashed line point cloud data and lane solid line point cloud data according to the lane line attribute information of each three-dimensional point cloud.

[0064] The vectorization processing module 304 is used to process the lane dashed line point cloud data, extract the first skeleton line of the lane dashed line point cloud data, and perform lane line fitting based on the first skeleton line to obtain the lane dashed line vectorization result; and to process the lane solid line point cloud data, extract the second skeleton line of the lane solid line point cloud data, and perform lane line fitting based on the second skeleton line to obtain the lane solid line vectorization result.

[0065] It is understood that the lane line vectorization system for fusion of camera images and laser point clouds provided by the present invention corresponds to the lane line vectorization method for fusion of camera images and laser point clouds provided in the foregoing embodiments. The relevant technical features of the lane line vectorization system for fusion of camera images and laser point clouds can be referred to the relevant technical features of the lane line vectorization method for fusion of camera images and laser point clouds, and will not be repeated here.

[0066] Please see Figure 4 , Figure 4 This is a schematic diagram illustrating an embodiment of the electronic device provided in this invention. For example... Figure 4 As shown, an embodiment of the present invention provides an electronic device 400, including a memory 410, a processor 420, and a computer program 411 stored in the memory 410 and executable on the processor 420. When the processor 420 executes the computer program 411, it implements the steps of a lane line vectorization method for fusing camera images and laser point clouds.

[0067] Please see Figure 5 , Figure 5 This is a schematic diagram illustrating an embodiment of a computer-readable storage medium provided by the present invention. (See diagram below.) Figure 5 As shown, this embodiment provides a computer-readable storage medium 500 on which a computer program 511 is stored. When the computer program 511 is executed by a processor, it implements the steps of a lane line vectorization method for fusing camera images and laser point clouds.

[0068] This invention provides a method and system for lane line vectorization by fusing camera images and laser point clouds. By fusing image perception semantic information with laser data, a depth-accurate lane line point cloud is obtained. Then, by combining the classification processing of lane dashed lines and lane solid lines, the method can effectively solve the lane line vectorization problem in conventional road sections, as well as in complex scenarios such as intersections and traffic dividers.

[0069] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0070] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0071] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0072] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0073] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0074] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0075] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for lane line vectorization by fusing camera images and laser point clouds, characterized in that, include: Ground information is extracted based on laser point cloud data; Based on the fusion and segmentation of image-perceived lane line semantic information and laser point cloud ground information, a lane line point cloud data set is obtained, wherein each three-dimensional point cloud in the lane line point cloud data set is marked with lane line attribute information. Based on the lane line attribute information of each 3D point cloud, the lane line point cloud data set is divided into lane dashed line point cloud data and lane solid line point cloud data. The dashed lane point cloud data is processed to extract the first skeleton line of the dashed lane point cloud data, and lane lines are fitted based on the first skeleton line to obtain the vectorized result of the dashed lane; and the solid lane point cloud data is processed to extract the second skeleton line of the solid lane point cloud data, and lane lines are fitted based on the second skeleton line to obtain the vectorized result of the solid lane. The process of processing the lane dashed line point cloud data, extracting the first skeleton line of the lane dashed line point cloud data, and fitting lane lines based on the first skeleton line to obtain the lane dashed line vectorization result includes: Extract the first lane vector based on the lane dashed line data; Calculate a first rotation matrix based on the first lane vector and the vehicle's driving direction, and rotate the lane dashed line to the vehicle's driving direction based on the first rotation matrix; The geometric topological relationship of lane dashed lines is calculated based on the histogram statistical method to obtain the set of lane dashed lines on the same side of the same lane; Extract the first skeleton line of the set of dashed lines of the lanes on the same side, and rotate the first skeleton line back to its original direction according to the first rotation matrix; Lane line fitting is performed based on the first skeleton line rotated back to its original direction to obtain the vectorized result of the lane dashed line; The process of processing the lane solid line point cloud data, extracting the second skeleton line of the lane solid line point cloud data, and fitting lane lines based on the second skeleton line to obtain the lane solid line vectorization result includes: Principal component analysis was performed on the solid lane point cloud data to distinguish between straight and curved segments based on the eigenvalue relationships. If it is a curved segment, based on the geometric positional relationship between the centroid of the solid lane point cloud data and the line connecting the near and far endpoints of the vehicle's driving direction, the curved segment is divided into multiple sub-segments. Each sub-segment is processed in the same way as a straight segment to extract a second skeleton line, and lane lines are fitted based on the second skeleton line to obtain the vectorized result of the solid lane line.

2. The lane line vectorization method according to claim 1, characterized in that, The extraction of ground information based on laser point cloud data includes: Based on the average elevation of the ground and the ground model, ground information within the lane range is extracted from the laser point cloud data.

3. The lane line vectorization method according to claim 1, characterized in that, The process of fusing and segmenting lane line semantic information from image perception with laser point cloud ground information to obtain a lane line point cloud data set includes: The laser point cloud data is projected onto the image-aware lane line semantic segmentation map. Based on the lane line semantic category, the laser point cloud falling within the lane semantic image block is assigned the corresponding lane dashed line attribute information and lane solid line attribute information, thus completing the lane line point cloud segmentation of the laser point cloud and obtaining the lane line point cloud data set.

4. The lane line vectorization method according to claim 3, characterized in that, The process of fusing and segmenting image-perceived lane line semantic information with laser point cloud ground information to obtain a lane line point cloud data set further includes: Based on the ground information, noise filtering is performed on the lane line point cloud in the laser point cloud data set to remove non-lane line point cloud data.

5. The lane line vectorization method according to claim 1, characterized in that, The process of extracting the second skeleton line according to the straight line segment processing method, and then fitting the lane line based on the second skeleton line to obtain the solid lane vectorization result includes: Extract the corresponding second lane vector from the solid lane point cloud data, and compare whether the angle between the second lane vector and the vehicle's driving direction is greater than a preset threshold. If the value is greater than a preset threshold, a second rotation matrix is ​​calculated based on the second lane vector and the vehicle's driving direction, and the lane solid line point cloud data is rotated to the vehicle's driving direction based on the second rotation matrix. Extract the second skeleton line from the solid lane point cloud data, and rotate the second skeleton line to its original direction according to the second rotation matrix; The second skeleton line rotated to its original direction is fitted with lane lines to obtain the solid line vectorization result of the lane lines; If the value is less than or equal to a preset threshold, the third skeleton line of the solid lane point cloud data is extracted, and lane line fitting is performed on the third skeleton line to obtain the vectorized result of the solid lane line.

6. A lane line vectorization system for fusing camera images and laser point clouds, implemented by the lane line vectorization method for fusing camera images and laser point clouds as described in any one of claims 1-5, characterized in that, include: The extraction module is used to extract ground information based on laser point cloud data; The fusion segmentation module is used to fuse and segment lane line semantic information and laser point cloud ground information based on image perception to obtain a lane line point cloud data set, wherein each three-dimensional point cloud in the lane line point cloud data set is marked with lane line attribute information. The partitioning module is used to partition the lane line point cloud data set into lane dashed line point cloud data and lane solid line point cloud data according to the lane line attribute information of each 3D point cloud. The vectorization processing module is used to process the lane dashed line point cloud data, extract the first skeleton line of the lane dashed line point cloud data, and perform lane line fitting based on the first skeleton line to obtain the lane dashed line vectorization result; and to process the lane solid line point cloud data, extract the second skeleton line of the lane solid line point cloud data, and perform lane line fitting based on the second skeleton line to obtain the lane solid line vectorization result.

7. An electronic device, characterized in that, It includes a memory and a processor, wherein the processor is used to implement the steps of the lane line vectorization method as described in any one of claims 1-5 when executing a computer management program stored in the memory.

8. A computer-readable storage medium, characterized in that, It stores a computer management program, which, when executed by a processor, implements the steps of the lane line vectorization method as described in any one of claims 1-5.

Images