Road mapping method, system and equipment based on vehicle-mounted camera and medium

Abstract

The invention provides a road mapping method, system and device based on a vehicle-mounted camera and a medium, and the method comprises the steps: collecting a plurality of groups of all-road-section driving data, each group of all-road-section driving data comprising a front view image data stream and vehicle position information; semantic feature recognition is carried out on the front view image data stream to obtain lane line pixel points of each frame of front view image, and the lane line pixel points are converted into 3D point cloud data under a world coordinate system through vehicle position information to obtain whole-course lane line 3D point cloud; and performing clustering and fitting processing on the whole-course lane line 3D point clouds of each group of whole-road-section driving data, overlapping and fusing all obtained vectorization sub-lane lines of each lane line to obtain a vectorization lane line of each lane line, and obtaining a road network of the target road section. The method provided by the invention can solve the technical problem that the accuracy and integrity of the road network are influenced by the factors such as shielding and illumination in the existing road mapping method, and can greatly reduce the mapping cost at the same time.

Description

Technical Field

[0001] The present invention belongs to the field of autonomous driving and relates to high-precision map technology, and specifically to a road mapping method, system, equipment and medium based on a vehicle-mounted camera. Background Art

[0002] High-precision maps for autonomous driving contain rich geometric and semantic features required for autonomous driving. Features such as road information and traffic signals are road information during the driving process. Road information ultimately supports the tasks and motion planning of autonomous driving vehicles. Therefore, mapping of road information within the autonomous driving neighborhood is a relatively important link.

[0003] Road mapping refers to the creation of diagrams of road networks, which is usually used in urban planning, traffic management, and navigation systems. Current road mapping methods include those based on aerial / satellite images and those based on vehicle-mounted lidar point clouds.

[0004] The mapping method based on aerial / satellite images is suitable for obtaining road network information over a large area. It extracts road information from aerial / satellite images through manual labeling or designed algorithms for mapping. The precision and accuracy of mapping depend largely on the image quality and are affected by factors such as weather and lighting, building occlusion, and terrain. As a result, it is difficult to ensure the accuracy and completeness of the road network. In addition, the extracted road network does not contain depth or elevation information.

[0005] The mapping method based on vehicle-mounted LiDAR point cloud is suitable for obtaining road information in a small range. It uses the vehicle-mounted LiDAR to scan the 3D point cloud of the surrounding environment of the road, distinguishes different objects by the intensity characteristics of the point cloud of different objects, and then extracts the road-related point cloud and processes it to obtain road network information. This method also faces the problems of occlusion and point density fluctuation when extracting road markings, making the road extraction inaccurate and incomplete, and the cost of LiDAR scanning acquisition is usually high. Summary of the invention

[0006] In order to solve the technical problem that factors such as occlusion and illumination affect the accuracy and integrity of the road network in the existing road mapping methods, the present invention discloses a road mapping method based on a vehicle-mounted camera, the method comprising the following steps:

[0007] S1. Collecting multiple sets of full-section driving data on a target road section, each set of full-section driving data includes a forward-view image data stream and vehicle position information;

[0008] S2, performing semantic feature recognition on the front view image data stream to obtain lane line pixel points of each frame of the front view image, and converting the lane line pixel points into 3D point cloud data in a world coordinate system through the vehicle position information corresponding to each frame of the front view image to obtain a full-course lane line 3D point cloud of each set of the full-section driving data;

[0009] S3. Clustering and fitting the 3D point cloud of the entire lane line of each group of the entire road section driving data, superimposing and fusing all the vectorized sub-lane lines of each lane line to obtain the vectorized lane line of each lane line, and obtaining the road network of the target section.

[0010] In an optional embodiment of the above step S2, the step of performing semantic feature recognition on the front view image data stream to obtain lane line pixel points of each frame of the front view image includes:

[0011] S21, using the trained semantic segmentation algorithm to perform semantic segmentation extraction on the front view image data stream to obtain lane line pixel semantic features of each frame of the front view image in the front view image data stream;

[0012] S22, extracting the lane line segmentation midpoint of the lane line pixel semantic features of each frame of the front view image to obtain the lane line pixel points of each frame of the front view image.

[0013] In an optional embodiment of the above step S2, converting the lane line pixel points into 3D point cloud data in a world coordinate system by using the vehicle position information corresponding to each frame of the front view image includes:

[0014] S23, performing coordinate system conversion on the lane line pixel points of each frame of the front view image by using the camera pose information obtained after calibrating the vehicle-mounted camera, to obtain the lane line points in the vehicle coordinate system;

[0015] S24. The lane line points in the self-vehicle coordinate system are converted into 3D point cloud data in the world coordinate system through the latitude, longitude and attitude angle of the self-vehicle position information corresponding to each frame of the front view image, wherein the height information in the 3D point cloud data is the elevation of the self-vehicle corresponding to each frame of the front view image.

[0016] In an optional embodiment of the above step S23, the vehicle-mounted camera is calibrated using a computer vision library or MATLAB camera calibration software to obtain the camera pose information, where the camera pose information includes a rotation matrix and a translation vector of the camera relative to the vehicle coordinate system.

[0017] In an optional embodiment of the above step S3, clustering and fitting the 3D point cloud of the entire lane line of each group of the entire road segment driving data includes:

[0018] S31, clustering the 3D point cloud of the entire lane line using a clustering algorithm to obtain a 3D point cloud set of each lane line in the target road section;

[0019] S32, performing piecewise straight line and piecewise curve fitting processing on the 3D point cloud set of each lane line to obtain a 3D lane line equation for each lane line;

[0020] S33, uniformly sampling the 3D lane line equation of each lane line, and connecting the obtained ordered points to obtain a vectorized sub-lane line of each lane line.

[0021] In an optional embodiment of the above step S33, uniformly sampling the 3D lane line equation of each lane line and connecting the obtained ordered points to obtain the vectorized sub-lane line of each lane line includes:

[0022] According to the starting point and the ending point of each lane line, each lane line is uniformly sampled to obtain the ordered points of the lane line, and the obtained ordered points are connected to obtain the vectorized sub-lane lines of each lane line.

[0023] An embodiment of the present invention further provides a road mapping system based on a vehicle-mounted camera, comprising a data acquisition module, a semantic feature segmentation module, a 3D point cloud conversion module and a point cloud processing and mapping module.

[0024] Among them, the data acquisition module is used to collect multiple groups of full-section driving data on the target section, each group of the full-section driving data includes a front-view image data stream and the vehicle position information, and the data acquisition module includes a vehicle-mounted camera, a GPS module and an IMU sensor.

[0025] The semantic feature segmentation module is used to perform semantic feature recognition on the front view image data stream to obtain lane line pixel points of each frame of the front view image;

[0026] The 3D point cloud conversion module is used to convert the lane line pixel points into 3D point cloud data in the world coordinate system through the vehicle position information corresponding to each frame of the front view image, so as to obtain the full lane line 3D point cloud of each set of the full road section driving data;

[0027] The point cloud processing and mapping module is used to cluster and fit the 3D point cloud of the entire lane line of each group of the entire road section driving data, and superimpose and fuse all the vectorized sub-lane lines of each lane line to obtain the vectorized lane line of each lane line, so as to obtain the road network of the target section.

[0028] In an optional embodiment of the above-mentioned point cloud processing and mapping module, the point cloud processing and mapping module includes a vectorized sub-lane line acquisition module and a fusion mapping module.

[0029] The vectorized sub-lane line acquisition module is used to cluster the 3D point cloud of the entire lane line using a clustering algorithm to obtain a 3D point cloud set of each lane line in the target road section; perform piecewise straight line and piecewise curve fitting processing on the 3D point cloud set of each lane line to obtain a 3D lane line equation of each lane line; uniformly sample the 3D lane line equation of each lane line, and connect the obtained ordered points to obtain a vectorized sub-lane line of each lane line;

[0030] The fusion mapping module is used to superimpose and fuse all the vectorized sub-lane lines of each lane line to obtain the vectorized lane line of each lane line.

[0031] An embodiment of the present invention further provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, any of the above-mentioned road mapping methods based on a vehicle-mounted camera is implemented to solve the technical problem that the accuracy and integrity of the road network are affected by factors such as occlusion and illumination in the existing road mapping methods.

[0032] An embodiment of the present invention further provides a computer-readable storage medium storing a computer program for executing any of the above-mentioned road mapping methods based on vehicle-mounted cameras, so as to solve the technical problem that the accuracy and integrity of the road network are affected by factors such as occlusion and illumination in the existing road mapping methods.

[0033] Compared with the prior art, the beneficial effects achieved by at least one of the above technical solutions adopted in the embodiments of this specification include at least: the road mapping method of the present invention converts the pixel information of the image into a 3D point cloud by using multiple groups of forward-looking image data streams of the entire road section collected by the on-board camera of the vehicle, combined with the geometric vehicle positioning information obtained by the GPS module and IMU sensor on the vehicle, and then constructs the road network information of the vehicle driving section by superposition and fusion. The method of the present invention can effectively reduce the cost of road mapping while ensuring a certain accuracy. At the same time, the superposition and fusion of multiple groups of data can solve the technical problem that the accuracy and integrity of the road network are affected by factors such as occlusion and illumination. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0035] Figure 1 A flow chart of a road mapping method based on a vehicle-mounted camera is disclosed in an embodiment of the present invention;

[0036] Figure 2 A schematic diagram of a vectorized lane line of a certain lane line disclosed in an embodiment of the present invention;

[0037] Figure 3 This is an architecture diagram of a road mapping system based on a vehicle-mounted camera disclosed in an embodiment of the present invention;

[0038] Figure 4 A schematic diagram of a computer device disclosed in an embodiment of the present invention;

[0039] Among them, 301, data acquisition module; 302, semantic feature segmentation module; 303, 3D point cloud conversion module; 304, point cloud processing and mapping module; 401, memory; 402, processor. DETAILED DESCRIPTION

[0040] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0041] The following describes the implementation methods of the present application through specific specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The present application can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that the following embodiments and the features of the embodiments can be combined with each other in the absence of conflict. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work belong to the scope of protection of the present application.

[0042] The embodiment of the present invention discloses a road mapping method based on a vehicle-mounted camera, see Figure 1 As shown, the method comprises the following steps:

[0043] S1. Collecting multiple sets of full-section driving data on a target road section, each set of full-section driving data includes a forward-view image data stream and vehicle position information;

[0044] S2, performing semantic feature recognition on the front view image data stream to obtain lane line pixel points of each frame of the front view image, and converting the lane line pixel points into 3D point cloud data in a world coordinate system through the vehicle position information corresponding to each frame of the front view image to obtain a full-course lane line 3D point cloud of each set of the full-section driving data;

[0045] S3. Clustering and fitting the 3D point cloud of the entire lane line of each group of the entire road section driving data, superimposing and fusing all the vectorized sub-lane lines of each lane line to obtain the vectorized lane line of each lane line, and obtaining the road network of the target section.

[0046] In an optional embodiment of the above step S1, for each target road section, full-section driving data under various conditions can be collected under different lighting conditions and different road environments. Specifically, when driving on the target road section, a forward-looking image data stream can be obtained by real-time acquisition through a vehicle-mounted camera set on the vehicle, such as a forward-looking monocular vehicle-mounted camera, wherein the forward-looking image data stream includes a continuous plurality of frames of forward-looking images. At the same time, the vehicle position corresponding to each frame of the forward-looking image can be obtained by data collected by high-precision equipment such as a GPS module and an IMU sensor set on the vehicle.

[0047] In an optional embodiment of the above step S2, the step of performing semantic feature recognition on the front view image data stream to obtain lane line pixel points of each frame of the front view image includes:

[0048] S21, using the trained semantic segmentation algorithm to perform semantic segmentation extraction on the front view image data stream to obtain lane line pixel semantic features of each frame of the front view image in the front view image data stream;

[0049] S22, extracting the lane line segmentation midpoint of the lane line pixel semantic features of each frame of the front view image to obtain the lane line pixel points of each frame of the front view image.

[0050] In an optional embodiment of the above step S2, converting the lane line pixel points into 3D point cloud data in a world coordinate system by using the vehicle position information corresponding to each frame of the front view image includes:

[0051] S23, performing coordinate system conversion on the lane line pixel points of each frame of the front view image by using the camera pose information obtained after calibrating the vehicle-mounted camera, to obtain the lane line points in the vehicle coordinate system;

[0052] S24. The lane line points in the self-vehicle coordinate system are converted into 3D point cloud data in the world coordinate system through the latitude, longitude and attitude angle of the self-vehicle position information corresponding to each frame of the front view image, wherein the height information in the 3D point cloud data is the elevation of the self-vehicle corresponding to each frame of the front view image.

[0053] In an optional embodiment of the above step S23, the vehicle-mounted camera is calibrated using a computer vision library or MATLAB camera calibration software to obtain the camera pose information, and the camera pose information includes the rotation matrix and translation vector of the camera relative to the vehicle coordinate system. When calibrating the vehicle-mounted camera, the vehicle coordinate system used refers to a coordinate system fixed on the vehicle body, which can be formed with the center point of the rear axle of the vehicle as the coordinate origin, the vehicle direction as the x direction, the left side of the vehicle as the y direction, and the vertical upward as the z direction to form a right-handed coordinate system.

[0054] In an optional embodiment of the above step S3, clustering and fitting the 3D point cloud of the entire lane line of each group of the entire road segment driving data includes:

[0055] S31, clustering the 3D point cloud of the entire lane line using a clustering algorithm to obtain a 3D point cloud set of each lane line in the target road section;

[0056] S32, performing piecewise straight line and piecewise curve fitting processing on the 3D point cloud set of each lane line to obtain a 3D lane line equation for each lane line;

[0057] S33, uniformly sampling the 3D lane line equation of each lane line, and connecting the obtained ordered points to obtain a vectorized sub-lane line of each lane line.

[0058] In the specific implementation, for the 3D point cloud of the entire lane line corresponding to each set of full-section driving data, the dbscan clustering algorithm can be used to cluster the point set corresponding to each lane line, and the fitting equation can be obtained by piecewise straight line and curve fitting for the point set of each lane line.

[0059] In an optional embodiment of the above step S33, uniformly sampling the 3D lane line equation of each lane line and connecting the obtained ordered points to obtain the vectorized sub-lane line of each lane line includes:

[0060] According to the starting point and the ending point of each lane line, each lane line is uniformly sampled to obtain the ordered points of the lane line, and the obtained ordered points are connected to obtain the vectorized sub-lane lines of each lane line.

[0061] In a specific implementation manner of the present invention, the vectorized lane line of each lane line can be obtained by:

[0062] First, each point in the 3D point cloud of the entire lane line corresponding to each set of full-segment driving data can be expressed as {(x0, y0), (x1, y1)...(x n ,y n )}, these point sets belong to different lane lines on the target road. After the clustering algorithm, the point sets of each lane line are obtained. The number of point sets of each lane center line is not fixed. For example, the first lane line has n1 points: The second lane line has n2 points: The mth lane has n m Points: When the number of point sets is greater than a certain threshold, each point set is fitted according to the following formula: Get the lane line parameters of the corresponding point set Where k∈(1,m], thus obtaining m 3D lane line parameter equations;

[0063] Then, according to the starting and ending points of the m groups of point sets, each lane line curve is uniformly sampled, and the obtained ordered points are connected to obtain m vectorized sub-lane lines;

[0064] Finally, when there are Z groups of data, some of the data may be incomplete or unclear due to occlusion. In this case, the Z group of data can be processed by superposition and fusion to make each lane line a relatively comprehensive and clear vectorized lane line. For the vectorized lane line of a lane line, see Figure 2 As shown in the figure, the road network mapping process is completed.

[0065] Based on the same inventive concept, a road mapping system based on a vehicle-mounted camera is also provided in an embodiment of the present invention, as described in the following embodiments. Since the principle of solving the problem by the road mapping system based on a vehicle-mounted camera is similar to that of the above-mentioned road mapping method, the implementation of the road mapping system can refer to the implementation of the above-mentioned road mapping method, and the repeated parts will not be repeated. As used below, the term "unit" or "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0066] Figure 3 is a structural block diagram of a road mapping system based on a vehicle-mounted camera disclosed in an embodiment of the present invention, such as Figure 3As shown, the system includes a data acquisition module 301, a semantic feature segmentation module 302, a 3D point cloud conversion module 303 and a point cloud processing and mapping module 304, and the structure is described below.

[0067] Among them, the data acquisition module 301 is used to collect multiple groups of full-section driving data on the target section, each group of the full-section driving data includes a front-view image data stream and the vehicle position information, and the data acquisition module includes a vehicle-mounted camera, a GPS module and an IMU sensor, wherein the GPS module is used to obtain the latitude and longitude and elevation of the vehicle in each frame, and the IMU sensor is used to measure the attitude angle of the vehicle in each frame, including the pitch angle (pitch), roll angle (roll) and heading angle (yaw). By combining the data of GPS and IMU, using algorithms such as Kalman filters to improve the accuracy and stability of the data, correcting the positioning and attitude information, the latitude and longitude, attitude angle and elevation of the vehicle in each frame are obtained.

[0068] The semantic feature segmentation module 302 is used to perform semantic feature recognition on the front view image data stream to obtain lane line pixel points of each frame of the front view image.

[0069] The 3D point cloud conversion module 303 is used to convert the lane line pixel points into 3D point cloud data in the world coordinate system through the vehicle position information corresponding to each frame of the front view image, so as to obtain the full lane line 3D point cloud of each set of the full road section driving data.

[0070] The point cloud processing and mapping module 304 is used to cluster and fit the 3D point cloud of the entire lane line of each group of the entire road section driving data, and superimpose and fuse all the vectorized sub-lane lines of each lane line to obtain the vectorized lane line of each lane line, so as to obtain the road network of the target section.

[0071] In an optional embodiment of the above-mentioned point cloud processing and mapping module 304, the point cloud processing and mapping module 304 includes a vectorized sub-lane line acquisition module and a fusion mapping module.

[0072] The vectorized sub-lane line acquisition module is used to cluster the 3D point cloud of the entire lane line using a clustering algorithm to obtain a 3D point cloud set of each lane line in the target road section; perform piecewise straight line and piecewise curve fitting processing on the 3D point cloud set of each lane line to obtain a 3D lane line equation of each lane line; uniformly sample the 3D lane line equation of each lane line, and connect the obtained ordered points to obtain a vectorized sub-lane line of each lane line;

[0073] The fusion mapping module is used to superimpose and fuse all the vectorized sub-lane lines of each lane line to obtain the vectorized lane line of each lane line.

[0074] The road mapping method of the present invention utilizes multiple sets of forward-view image data streams of the entire road section collected by the on-board camera of the vehicle, combined with the geometric vehicle positioning information obtained by the GPS module and IMU sensor on the vehicle, to convert the pixel information of the image into a 3D point cloud, and then constructs the road network information of the vehicle driving section by superposition and fusion. The method of the present invention can effectively reduce the cost of road mapping while ensuring a certain accuracy. At the same time, the superposition and fusion of multiple sets of data can solve the technical problem that the accuracy and integrity of the road network are affected by factors such as occlusion and illumination.

[0075] In this embodiment, a computer device is provided, such as Figure 4 As shown, it includes a memory 401, a processor 402, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, any of the above-mentioned road mapping methods based on vehicle-mounted cameras is implemented.

[0076] Specifically, the computer device may be a computer terminal, a server or a similar computing device.

[0077] In this embodiment, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program for executing any of the above-mentioned road mapping methods based on a vehicle-mounted camera.

[0078] Specifically, computer-readable storage media include permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, program modules or other data. Examples of computer-readable 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 technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable storage media does not include transitory media such as modulated data signals and carrier waves.

[0079] Obviously, those skilled in the art should understand that the modules or steps of the above-mentioned embodiments of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. In this way, the embodiments of the present invention are not limited to any specific combination of hardware and software.

[0080] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the embodiments of the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A road mapping method based on a vehicle-mounted camera, characterized in that: include: Collecting multiple sets of full-segment driving data on the target road section, each set of the full-segment driving data includes a forward-looking image data stream and vehicle position information; Performing semantic feature recognition on the front view image data stream to obtain lane line pixel points of each frame of the front view image, converting the lane line pixel points into 3D point cloud data in a world coordinate system through the vehicle position information corresponding to each frame of the front view image, and obtaining a full-course lane line 3D point cloud for each set of the full-section driving data; The 3D point cloud of the entire lane line of each group of the entire road section driving data is clustered and fitted, and all the vectorized sub-lane lines of each lane line are superimposed and fused to obtain the vectorized lane line of each lane line, so as to obtain the road network of the target section.

2. The road mapping method based on a vehicle-mounted camera according to claim 1, characterized in that: The performing semantic feature recognition on the front view image data stream to obtain lane line pixel points of each frame of the front view image includes: Using the trained semantic segmentation algorithm, semantic segmentation is performed on the front view image data stream to extract the lane line pixel semantic features of each frame of the front view image in the front view image data stream; The lane line segmentation midpoint of the lane line pixel semantic feature of each frame of the front view image is extracted to obtain the lane line pixel point of each frame of the front view image.

3. The road mapping method based on a vehicle-mounted camera according to claim 1, characterized in that: The converting the lane line pixel points into 3D point cloud data in a world coordinate system by using the vehicle position information corresponding to each frame of the front view image includes: By using the camera pose information obtained after calibrating the vehicle-mounted camera, the lane line pixel points of each frame of the front view image are converted into a coordinate system to obtain the lane line points in the vehicle coordinate system; The lane line points in the vehicle coordinate system are converted into 3D point cloud data in the world coordinate system through the latitude, longitude and attitude angle of the vehicle position information corresponding to each frame of the front view image, wherein the height information in the 3D point cloud data is the elevation of the vehicle corresponding to each frame of the front view image.

4. The road mapping method based on a vehicle-mounted camera according to claim 3, characterized in that: The vehicle-mounted camera is calibrated using a computer vision library or MATLAB camera calibration software to obtain the camera pose information, which includes a rotation matrix and a translation vector of the camera relative to the vehicle coordinate system.

5. The road mapping method based on a vehicle-mounted camera according to claim 1, characterized in that: The clustering and fitting processing of the 3D point cloud of the entire lane line of each group of the entire road section driving data includes: Clustering the 3D point cloud of the entire lane line using a clustering algorithm to obtain a 3D point cloud set of each lane line in the target road section; Performing piecewise straight line and piecewise curve fitting processing on the 3D point cloud set of each lane line to obtain a 3D lane line equation for each lane line; The 3D lane line equation of each lane line is uniformly sampled, and the obtained ordered points are connected to obtain a vectorized sub-lane line of each lane line.

6. The road mapping method based on a vehicle-mounted camera according to claim 5, characterized in that: The uniformly sampling the 3D lane line equation of each lane line and connecting the obtained ordered points to obtain the vectorized sub-lane line of each lane line includes: According to the starting point and the ending point of each lane line, each lane line is uniformly sampled to obtain the ordered points of the lane line, and the obtained ordered points are connected to obtain the vectorized sub-lane lines of each lane line.

7. A road mapping system based on a vehicle-mounted camera, characterized in that: include: A data acquisition module, the data acquisition module is used to collect multiple sets of full-section driving data on the target road section, each set of the full-section driving data includes a front view image data stream and vehicle position information, and the data acquisition module includes a vehicle-mounted camera, a GPS module and an IMU sensor; A semantic feature segmentation module, the semantic feature segmentation module is used to perform semantic feature recognition on the front view image data stream to obtain lane line pixel points of each frame of the front view image; A 3D point cloud conversion module, the 3D point cloud conversion module is used to convert the lane line pixel points into 3D point cloud data in a world coordinate system through the vehicle position information corresponding to each frame of the front view image, so as to obtain a 3D point cloud of the entire lane line of each set of the entire road section driving data; A point cloud processing and mapping module is used to cluster and fit the 3D point cloud of the entire lane line of each group of the entire road section driving data, and to superimpose and fuse all the vectorized sub-lane lines of each lane line to obtain the vectorized lane line of each lane line, so as to obtain the road network of the target section.

8. The road mapping system based on a vehicle-mounted camera according to claim 7, characterized in that: The point cloud processing and mapping module includes: A vectorized sub-lane line acquisition module, which is used to cluster the 3D point cloud of the entire lane line using a clustering algorithm to obtain a 3D point cloud set of each lane line in the target road section; perform piecewise straight line and piecewise curve fitting processing on the 3D point cloud set of each lane line to obtain a 3D lane line equation of each lane line; uniformly sample the 3D lane line equation of each lane line, and connect the obtained ordered points to obtain a vectorized sub-lane line of each lane line; A fusion module is used to superimpose and fuse all the vectorized sub-lane lines of each lane line to obtain a vectorized lane line for each lane line.

9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the road mapping method based on a vehicle-mounted camera according to any one of claims 1 to 6 is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program for executing the road mapping method based on a vehicle-mounted camera according to any one of claims 1 to 6.

Images