A mapping method, device, map making system, medium and equipment

Abstract

The application discloses a mapping method and device, a map making system, a medium and equipment, and belongs to the technical field of data processing. The method mainly comprises the following steps: performing line feature extraction on a 2D image corresponding to a pre-established 3D point cloud map to obtain a 2D line segment; respectively extending a line of sight from a visual focus to both ends of the 2D line segment to obtain intersection points corresponding to both ends of the 2D line segment in the 3D point cloud map; generating an intersection line segment by using the intersection points, projecting the intersection line segment into the 2D image, and obtaining a target line segment with a perpendicular distance to the 2D line segment less than or equal to a first preset distance threshold; and establishing a line feature map by using the target line segment. A method based on a view cone point cloud is provided to restore the depth information of the line feature extracted from the image, avoid converting the 2D line feature into a 3D line feature through line feature triangulation, reduce the computation amount of the system, simplify the operation process, improve the mapping efficiency, obtain more accurate depth information in combination with laser point clouds, and improve the map precision.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a mapping method, apparatus, map production system, medium and equipment. Background Technology

[0002] High-precision positioning is a problem that modern autonomous driving systems need to solve. As an important step in high-precision positioning, the accuracy of 3D mapping has a crucial impact on the accuracy of high-precision positioning.

[0003] In existing technologies, Plück coordinates are often used, where the endpoints of a line feature are represented by Plück coordinates. This means that a line feature needs to be represented by two Plück coordinates, which means that the line feature is six-dimensional data. Furthermore, line feature triangulation is required during the mapping process to ensure mapping accuracy.

[0004] However, the six-dimensional data of this line feature occupies a large amount of storage space. When optimizing the completed 3D map, the six-dimensional data will have parameterization issues during the optimization process. Furthermore, due to the large amount of data, the system's computational load increases and the computation process becomes more complex. Summary of the Invention

[0005] In view of the problems of over-parameterization and increased computational load and complexity caused by large amounts of data in existing technologies, this application mainly provides a mapping method, apparatus, map production system, medium and equipment.

[0006] In a first aspect, embodiments of this application provide a mapping method, comprising: extracting line features from a 2D image corresponding to a pre-established 3D point cloud map to obtain 2D line segments corresponding to map features; extending the line of sight to both ends of the 2D line segments according to the visual focus to obtain the intersection points of the two ends of the 2D line segments in the 3D point cloud map; generating intersection line segments using the intersection points and projecting the intersection line segments onto the 2D image to obtain target line segments whose vertical distance to the 2D line segments is less than or equal to a first preset distance threshold; and establishing a line feature map using the target line segments.

[0007] Secondly, embodiments of this application provide a mapping apparatus, comprising: a 2D line segment acquisition module, which extracts line features from a 2D image corresponding to a pre-established 3D point cloud map to acquire 2D line segments corresponding to map features; an intersection point acquisition module, which extends the line of sight to both ends of the 2D line segment according to the visual focus to acquire the intersection points corresponding to the two ends of the 2D line segment in the 3D point cloud map; a target line segment acquisition module, which generates intersection point line segments using the intersection points and projects the intersection point line segments onto the 2D image to acquire target line segments whose vertical distance from the 2D line segments is less than or equal to a first preset distance threshold; and a line feature mapping module, which builds a line feature map using the target line segments.

[0008] Thirdly, embodiments of this application provide a map production system, which includes a mapping device, wherein the mapping device includes: a 2D line segment acquisition module, which extracts line features from a 2D image corresponding to a pre-established 3D point cloud map to acquire 2D line segments corresponding to map features; an intersection point acquisition module, which extends the line of sight to both ends of the 2D line segment according to the visual focus to acquire the intersection points corresponding to the two ends of the 2D line segment in the 3D point cloud map; a target line segment acquisition module, which generates intersection point line segments using the intersection points and projects the intersection point line segments onto the 2D image to acquire target line segments whose vertical distance from the 2D line segments is less than or equal to a first preset distance threshold; and a line feature mapping module, which builds a line feature map using the target line segments.

[0009] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer instructions that are operated to perform the mapping method in Scheme 1.

[0010] Fifthly, embodiments of this application provide a computer device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores computer instructions executable by the at least one processor, and the at least one processor operates the computer instructions to execute the mapping method in Scheme 1.

[0011] The technical solution of this application provides a method based on view frustum point clouds to recover the depth information of line features extracted from images. This avoids converting 2D line features into 3D line features through line feature triangulation, reducing the computational load of the system, simplifying the computation process, and improving mapping efficiency. Furthermore, by combining laser point clouds, more accurate depth information is obtained, improving map accuracy. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of an optional implementation of a mapping method according to this application;

[0014] Figure 2 This paper illustrates an optional example of obtaining the intersection point of one end of a line segment in a mapping method of this application;

[0015] Figure 3 This is a schematic diagram of an optional embodiment of a mapping apparatus according to this application.

[0016] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0017] The preferred embodiments of this application will now be described in detail with reference to the accompanying drawings, so that the advantages and features of this application can be more easily understood by those skilled in the art, thereby providing a clearer and more definite definition of the scope of protection of this application.

[0018] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0019] This application primarily provides a mapping method, apparatus, map production system, medium, and device. The mapping method includes: extracting line features from a 2D image corresponding to a pre-established 3D point cloud map to obtain 2D line segments corresponding to the map features; extending the line of sight to both ends of the 2D line segments according to the visual focus to obtain the intersection points of the two ends of the 2D line segments in the 3D point cloud map; generating intersection line segments using the intersection points and projecting the intersection line segments onto the 2D image to obtain target line segments whose vertical distance to the 2D line segments is less than or equal to a first preset distance threshold; and establishing a line feature map using the target line segments.

[0020] By providing a frustum-based approach, depth information of line features extracted from images can be recovered, thus avoiding the need to convert 2D line features into 3D line features through line feature triangulation. This reduces the computational load, simplifies the computation process, and improves mapping efficiency. Furthermore, combining this with laser point clouds allows for the acquisition of more accurate depth information, further enhancing map precision.

[0021] The technical solutions of this application and how they solve the aforementioned technical problems will be described in detail below with specific embodiments. The specific embodiments described below can be combined with each other to form new embodiments. The same or similar ideas or processes described in one embodiment may not be repeated in other embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0022] Figure 1 An alternative implementation of a mapping method according to this application is shown.

[0023] exist Figure 1 In the optional implementation shown, the mapping method mainly includes step S101, which involves extracting line features from the 2D image corresponding to the pre-established 3D point cloud map to obtain the 2D line segments corresponding to the map features.

[0024] In this optional implementation, the LiDAR installed on the autonomous vehicle collects 3D point clouds of the surrounding environment of the autonomous vehicle, establishes a corresponding 3D point cloud map based on the 3D point cloud, converts the 3D point cloud map into a corresponding 2D image, and extracts line segments of map features from the 2D image. The method of extracting feature points based on images saves the computational load of the system and occupies less memory, thereby improving the user experience.

[0025] In an optional embodiment of this application, the pre-establishment of a 3D point cloud map further includes: when the map features are ground features, such as lane lines, stop lines, curbs, speed bumps, etc., multi-frame 3D ground point cloud maps centered on the radar mounted on the vehicle are stitched together to generate a 3D point cloud map; when the map features are non-ground features, such as buildings, bridges, trees, etc., multi-frame 3D non-ground point clouds collected by the radar within a preset time period are clustered to generate a 3D point cloud map.

[0026] In this optional embodiment, before pre-building the 3D point cloud map, ground features are first distinguished into ground features and non-ground features. The 3D point clouds acquired by radar are then categorized according to different features, classifying them into 3D ground point clouds and 3D non-ground point clouds. Different 3D point cloud maps are then built based on the type of point cloud. During the mapping process, since the single frame or a few frames of 3D ground point clouds acquired by radar are relatively sparse, pre-built multi-frame 3D ground point cloud maps are used to further refine these maps. The system stitches together the data to obtain a 3D point cloud map corresponding to the ground features. Compared to single-frame or a few-frame 3D ground point clouds, 3D non-ground point clouds are sufficiently dense. Therefore, single-frame or a few-frame 3D non-ground point clouds acquired by radar are used directly to build a 3D point cloud map corresponding to the non-ground features. In this embodiment, ground features and non-ground features are distinguished and mapped separately. This ensures that the built 3D point cloud map saves system computation and reduces memory usage, while maintaining sufficient accuracy to guarantee the precision of the extracted line segments, thus improving the user experience.

[0027] In one optional example of this application, when distinguishing between ground features and non-ground features, a 3D ground point cloud is collected separately using a LiDAR, and 3D mapping is performed on multiple frames of the 3D ground point cloud. The 3D map of the current autonomous vehicle's surrounding environment is then stitched together to obtain a 3D point cloud map corresponding to the ground features. A 3D non-ground point cloud is collected separately using a LiDAR, and 3D mapping is performed on multiple frames of the 3D non-ground point cloud to obtain a 3D point cloud map corresponding to the non-ground features.

[0028] In one optional embodiment of this application, 3D point clouds of the environment surrounding the autonomous vehicle are acquired using LiDAR. Each 3D point cloud is then evaluated. If the height value of a 3D point cloud is less than or equal to a preset height threshold, the 3D point cloud is identified as a 3D ground point cloud; if the height value is greater than the preset height threshold, the 3D point cloud is identified as a 3D non-ground point cloud. 3D mapping is performed on multiple frames of the 3D ground point cloud, and the 3D map of the current environment surrounding the autonomous vehicle is stitched together to obtain a 3D point cloud map corresponding to the ground features. Similarly, 3D mapping is performed on multiple frames of the 3D non-ground point cloud to obtain a 3D point cloud map corresponding to the non-ground features.

[0029] exist Figure 1 In the optional embodiment shown, the mapping method further includes step S102, which involves extending the line of sight to both ends of the 2D line segment according to the visual focus, and obtaining the intersection points of the two ends of the 2D line segment in the 3D point cloud map.

[0030] In this optional implementation, the line of sight is extended towards one end of the 2D line segment according to the direction of the visual focus, so that the line of sight intersects with the 3D point cloud map, thereby obtaining the intersection point at one end of the 2D line segment. The intersection point at the other end of the 2D line segment is obtained using the same method. The visual focus is the focus of the visual sensor installed in the autonomous vehicle, and the visual sensor includes, but is not limited to, a visual odometry system. This solution uses vision to obtain the intersection points at both ends of the 2D line segment, avoiding the complex process of converting 2D line features into 3D line features using line feature triangulation, reducing the computational load of the system, and accelerating the efficiency of line feature mapping.

[0031] In one optional embodiment of this application, the method of extending the line of sight from the visual focus to both ends of the 2D line segment to obtain the intersection points of the two ends of the 2D line segment in the 3D point cloud map further includes: emitting rays from the visual focus to both ends of the 2D line segment to obtain regional point cloud data within a preset area where the rays intersect in the 3D point cloud map; performing planar fitting on the regional point cloud data to generate a regional fitting plane; and calculating the intersection points of the rays on the regional fitting plane based on the plane equation of the regional fitting plane and the line equation of the ray.

[0032] In this optional embodiment, since the line of sight is not a line but a cone-shaped ray, when the line of sight extends to one end of the 2D line segment and intersects with the 3D point cloud map, the intersection area, i.e., the preset area, is obtained. The point cloud data in the preset area is fitted with a plane to obtain the area fitting plane in the direction of the line of sight. The plane equation of the area fitting plane is calculated using the coordinate information of the point cloud data in the area fitting plane. The straight line equation of the central ray is calculated using the position of the visual focus and the position of one end of the 2D line segment. The intersection point is obtained by combining the plane equation and the straight line equation. Optionally, the angle of the cone-shaped ray is 15°.

[0033] Figure 2 This illustrates an optional example of obtaining the intersection point of one end of a line segment in a mapping method of this application.

[0034] exist Figure 2In the example shown, the solid black circle represents the focal point of the line of sight. The straight lines emanating from the focal point diagonally upward and downward represent the direction of the line of sight. These straight lines include solid lines and dashed lines. The solid lines represent the direction of the line of sight from the outermost edge of the cone ray, while the dashed lines represent the central ray of the line of sight. The vertical dark line to the right of the focal point of the line of sight represents the previously extracted 2D line segment. In the ellipse to the right of the 2D line segment, the relatively larger ellipse represents the 3D point cloud map, while the two relatively smaller ellipses are the region fitting planes corresponding to the areas where the cone ray emanating from the focal point of the line of sight intersects with the 3D point cloud map. When the line of sight extending from the cone-shaped ray to one end of the 2D line segment intersects with the 3D point cloud map, the intersection area, i.e., the preset area, is obtained. Plane fitting is performed on the point cloud data within this preset area to obtain the area fitting plane along the line of sight, which consists of two relatively small ellipses. The plane equation of the area fitting plane is calculated using the coordinate information of the point cloud data in the area fitting plane. The straight line equation of the central ray is calculated using the position of the visual focus and the position of one end of the 2D line segment. The intersection point is obtained by combining the plane equation and the straight line equation.

[0035] exist Figure 1 In the optional embodiment shown, the mapping method further includes step S103, generating intersection line segments using intersection points, and projecting the intersection line segments into a 2D image to obtain target line segments whose vertical distance from the 2D line segments is less than or equal to a first preset distance threshold.

[0036] In this optional implementation, the intersection points determined in the fitting planes of the two regions are connected to obtain the 3D line segment corresponding to the 2D line segment, namely the intersection line segment. The intersection line segment is then projected back into the 2D image. Based on the positional relationship between the 2D line segment and the projected intersection line segment, when the vertical distance between the projected intersection line segment and the 2D line segment is within a first preset domain, the intersection line segment is determined to be the target line segment. Here, the image-based method for identifying the target line segment has fast recognition efficiency and low system computation.

[0037] In an optional embodiment of this application, generating intersection line segments using intersection points further includes: when the map feature is a non-ground feature, determining the number of intersection points determined at both ends of the 2D line segment; when the number of intersection points at both ends is equal to a preset number, generating intersection line segments using the intersection points; when the number of intersection points at one end is greater than the preset number, selecting intersection line segments from the connecting line segments formed by connecting the intersection points.

[0038] In this optional embodiment, since the line segments corresponding to non-ground features in the 2D image overlap, the intersection points obtained according to the aforementioned method may be the intersection points corresponding to the 2D line segments extracted from other non-ground features. For example, two pillars placed side by side, from the line of sight, only the pillar closer to the line of sight is visible, while the pillar farther from the line of sight is obscured by the pillar closer to the line of sight. However, the distance between the two pillars is 10 meters. When obtaining the intersection point of the pillar closer to the line of sight in the 3D point cloud map using the aforementioned method, there is also an intersection point corresponding to the pillar farther from the line of sight. The fitting plane in this area should calculate one intersection point, i.e., the preset number is set to 1. However, according to the above situation, the fitting plane in this area will calculate two intersection points. One intersection point needs to be removed. Therefore, the two intersection points calculated by the fitting plane in this area can be connected to the intersection points corresponding to the other end of the 2D line segments to obtain two connecting line segments. Based on the relationship between the 2D line segments and the ground, the correct intersection point line segment is selected from the two connecting line segments.

[0039] In an optional embodiment of this application, selecting intersection line segments from connecting line segments formed by connecting intersection points further includes: when the map feature is a feature perpendicular to the ground, selecting connecting line segments perpendicular to the ground as intersection line segments; when the map feature is a feature parallel to the ground, selecting connecting line segments parallel to the ground as intersection line segments.

[0040] In this optional embodiment, since the relationship between non-ground map features and the ground includes, but is not limited to, perpendicularity and parallelism, when a 2D line segment calculates two intersection points at one end and one intersection point at the other end, when the map feature is perpendicular to the ground, connecting line segments that are not perpendicular to the ground are discarded, and connecting line segments that are perpendicular to the ground are taken as intersection point segments. For example, with the aforementioned pillars, since the distance between the two pillars is 10 meters, there must be one connecting line segment perpendicular to the ground and one connecting line segment not perpendicular to the ground. Therefore, the connecting line segment perpendicular to the ground is taken as the intersection point segment of the pillar closer to the focal point of the line of sight. When the map feature is parallel to the ground, connecting line segments that are not parallel to the ground are discarded, and connecting line segments that are parallel to the ground are taken as intersection point segments, such as a beam.

[0041] In an optional embodiment of this application, projecting the intersection line segments onto a 2D image to obtain target line segments whose perpendicular distance from the 2D line segments is less than or equal to a first preset distance threshold further includes: projecting the intersection line segments onto a 2D image to obtain the projected line segments corresponding to the intersection line segments; calculating the perpendicular distance between the 2D line segments and the projected line segments based on the depth information of the intersection line segments; when the number of intersection line segments is less than or equal to a preset number, selecting the intersection line segments whose perpendicular distance is less than or equal to the first preset distance threshold as target line segments; when the number of intersection line segments is less than or equal to a preset number, optimizing the intersection line segments whose perpendicular distance is greater than the first preset distance threshold and less than or equal to a second preset distance threshold to obtain target line segments, wherein the second preset distance threshold is greater than the first preset distance threshold; when the number of intersection line segments is greater than a preset number, selecting the intersection line segments whose perpendicular distance is less than or equal to the first preset distance threshold as target line segments; when the number of intersection line segments is greater than a preset number, discarding the intersection line segments whose perpendicular distance is greater than the first preset distance threshold.

[0042] In this optional embodiment, based on the previously obtained intersection line segments, the intersection line segments are projected onto a 2D image to obtain the corresponding projected line segments. Based on the depth information of the intersection line segments, the vertical distance between the 2D line segment and the projected line segment is calculated. If the number of intersection line segments is one, i.e., when the preset number is one, the intersection line segments with a vertical distance less than or equal to a first preset distance threshold are designated as target line segments. The intersection line segments with a vertical distance greater than the first preset distance threshold and less than or equal to a second preset distance threshold are optimized, i.e., the intersection line segments are corrected so that the 2D line segment coincides with the projected intersection line segment, and the corrected intersection line segments are designated as target line segments. If the number of intersection line segments is multiple, i.e., when the preset number is greater than one, the intersection line segments with a vertical distance less than or equal to the first preset distance threshold are designated as target line segments. The intersection line segments with a vertical distance greater than the first preset distance threshold are discarded.

[0043] For example, when two intersection points are calculated at both ends of a 2D line segment, four connecting line segments are obtained by connecting the intersection points at both ends. When the map feature is perpendicular to the ground, the connecting line segments that are not perpendicular to the ground are removed, leaving two connecting line segments that are perpendicular to the ground. Both connecting line segments are taken as intersection line segments. At this time, the two intersection line segments are projected into the 2D image. Since the depth information of the two intersection line segments is different, the intersection line segment that is farther away is removed by the first preset distance threshold, and the intersection line segment that is closer is taken as the target line segment.

[0044] exist Figure 1 In the optional embodiment shown, the mapping method further includes step S104, which uses the target line segment to create a line feature map.

[0045] In this optional implementation, after obtaining the target line segment using the above method, line feature mapping is performed using the target line segment to obtain an accurate line feature map and improve the line feature map.

[0046] Figure 3 An alternative embodiment of a mapping apparatus according to this application is shown.

[0047] exist Figure 3 In the optional embodiment shown, the mapping device mainly includes: a 2D line segment acquisition module 301, which extracts line features from the 2D image corresponding to the pre-established 3D point cloud map to obtain 2D line segments corresponding to map features; an intersection point acquisition module 302, which extends the line of sight to both ends of the 2D line segment according to the visual focus to obtain the intersection points corresponding to the two ends of the 2D line segment in the 3D point cloud map; a target line segment acquisition module 303, which generates intersection line segments using the intersection points of the two ends of the 3D line segment, projects the intersection line segments onto the 2D image, and obtains target line segments whose vertical distance from the 2D line segment is less than or equal to a first preset distance threshold; and a line feature mapping module 304, which establishes a line feature map using the target line segments.

[0048] In an optional embodiment of this application, the functional modules of the mapping device of this application may be directly in hardware, in software modules executed by a processor, or in a combination of both.

[0049] Software modules may reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in this art. An exemplary storage medium is coupled to the processor, enabling the processor to read information from and write information to the storage medium.

[0050] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic, discrete hardware components, or any combination thereof. A general-purpose processor can be a microprocessor, but alternatively, it can be any conventional processor, controller, microcontroller, or state machine. The processor can also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors incorporating a DSP core, or any other such configuration. Alternatively, the storage medium can be integrated with the processor. The processor and storage medium can reside in an ASIC. The ASIC can reside in the user terminal. Alternatively, the processor and storage medium can reside as discrete components in the user terminal.

[0051] The mapping apparatus provided in this application can be used to execute the mapping method described in any of the above embodiments. Its implementation principle and technical effect are similar, and will not be repeated here.

[0052] In another optional embodiment of this application, a map making system is provided, comprising a mapping device, wherein the mapping device includes: a 2D line segment acquisition module, which extracts line features from a 2D image corresponding to a pre-established 3D point cloud map to acquire 2D line segments corresponding to map features; an intersection point acquisition module, which acquires the intersection points corresponding to the two ends of the 3D line segment based on the visual focus and the 3D line segment corresponding to the 2D line segment in the 3D point cloud map; a target line segment acquisition module, which generates an intersection point line segment using the intersection points of the two ends of the 3D line segment, projects the intersection point line segment onto the 2D image, and acquires a target line segment whose vertical distance to the 2D line segment is less than or equal to a first preset distance threshold; and a line feature mapping module, which builds a line feature map using the target line segment.

[0053] The map creation system provided in this application can be used to execute the mapping method described in any of the above embodiments. Its implementation principle and technical effect are similar, and will not be repeated here.

[0054] In another alternative embodiment of this application, a computer-readable storage medium stores computer instructions that are operated to perform the mapping method described in the above embodiments.

[0055] In another optional embodiment of this application, a computer device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores computer instructions executable by the at least one processor, and the at least one processor operates the computer instructions to perform the mapping method described in the above embodiments.

[0056] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0057] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0058] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A mapping method, characterized in that, include: Line features are extracted from the 2D image corresponding to the pre-built 3D point cloud map to obtain the 2D line segments corresponding to the map features; Based on the visual focus, extend the line of sight to both ends of the 2D line segment to obtain the intersection points of the two ends of the 2D line segment in the 3D point cloud map, including: Rays are emitted from the visual focal point to both ends of the 2D line segment, and regional point cloud data are obtained within the preset area where the rays intersect with the 3D point cloud map. Perform planar fitting on the point cloud data of the region to generate a region fitting plane; Based on the plane equation of the fitted plane of the region and the line equation of the ray, calculate the intersection points of the ray on the fitted plane of the region respectively; Intersection line segments are generated using the intersection points, and the intersection line segments are projected onto the 2D image to obtain target line segments whose vertical distance from the 2D line segments is less than or equal to a first preset distance threshold. A line feature map is created using the target line segment.

2. The mapping method according to claim 1, characterized in that, The step of generating an intersection line segment using the intersection point further includes: When the map feature is a non-ground feature, determine the number of intersection points determined from the two ends of the 2D line segment respectively; When the number of intersection points at both ends is equal to the preset number, the intersection point segment is generated using the intersection points; When the number of intersections at one end is greater than the preset number, the intersection segments are selected from the connecting line segments formed by connecting the intersections.

3. The mapping method according to claim 2, characterized in that, The step of selecting the intersection point segments from the connecting line segments formed by the intersection points further includes: When the map feature is a feature perpendicular to the ground, the connecting line segment perpendicular to the ground is taken as the intersection line segment; When the map feature is a feature parallel to the ground, the connecting line segment parallel to the ground is taken as the intersection line segment.

4. The mapping method according to claim 1, characterized in that, The step of projecting the intersection line segment onto the 2D image to obtain a target line segment whose perpendicular distance to the 2D line segment is less than or equal to a first preset distance threshold further includes: Based on the depth information of the intersection line segment, the intersection line segment is projected onto the 2D image to obtain the projection line segment corresponding to the intersection line segment; Calculate the vertical distance between the 2D line segment and the projected line segment; When the number of line segments at the intersection point is less than or equal to a preset number, the line segments at the intersection point with a vertical distance less than or equal to a first preset distance threshold are designated as the target line segments. When the number of line segments at the intersection point is less than or equal to a preset number, the line segments at the intersection point whose vertical distance is greater than a first preset distance threshold and less than or equal to a second preset distance threshold are optimized to obtain the target line segment, wherein the second preset distance threshold is greater than the first preset distance threshold. When the number of line segments at the intersection point is greater than the preset number, the line segments at the intersection point whose vertical distance is less than or equal to the first preset distance threshold are designated as the target line segments. When the number of line segments at the intersection point is greater than the preset number, the line segments at the intersection point whose vertical distance is greater than the first preset distance threshold are removed.

5. The mapping method according to claim 1, characterized in that, The method for generating the pre-built 3D point cloud map includes: When the map feature is a ground feature, the 3D ground point cloud map is generated by stitching together multiple frames of 3D ground point cloud map centered on the radar mounted on the vehicle. When the map features are non-ground features, the multiple frames of 3D non-ground point clouds collected by the radar within a preset time period are clustered to generate the 3D point cloud map.

6. A mapping device, characterized in that, include: The 2D line segment acquisition module extracts line features from the 2D image corresponding to the pre-established 3D point cloud map and acquires the 2D line segments corresponding to the map features. The intersection point acquisition module extends the line of sight to both ends of the 2D line segment based on the visual focus, and acquires the intersection points of the two ends of the 2D line segment in the 3D point cloud map, respectively, including: Rays are emitted from the visual focal point to both ends of the 2D line segment, and regional point cloud data are obtained within the preset area where the rays intersect with the 3D point cloud map. Perform planar fitting on the point cloud data of the region to generate a region fitting plane; Based on the plane equation of the fitted plane of the region and the line equation of the ray, calculate the intersection points of the ray on the fitted plane of the region respectively; The target line segment acquisition module generates an intersection line segment using the intersection point and projects the intersection line segment onto the 2D image to acquire a target line segment whose vertical distance from the 2D line segment is less than or equal to a first preset distance threshold. The line feature mapping module uses the target line segment to create a line feature map.

7. A map creation system, characterized in that, The map creation system includes the mapping apparatus as described in claim 6.

8. A computer-readable storage medium storing computer instructions, characterized in that, The computer instructions are operated to perform the mapping method according to any one of claims 1-5.

9. A computer device, characterized in that, include: At least one processor; as well as A memory that is communicatively connected to the at least one processor; The memory stores computer instructions executable by the at least one processor, which operates the computer instructions to perform the mapping method as described in any one of claims 1-5.

Images