Lane center line extraction method and device, equipment and storage medium

By converting the road section trajectory into a trajectory feature map and identifying the lane center line, the problem of poor lane center line extraction in the existing technology is solved, and more efficient and accurate lane center line extraction is achieved, and intelligent driving applications are supported.

CN119007150APending Publication Date: 2024-11-22NAVINFO
View PDF 0 Cites 3 Cited by

Patent Information

Application Number
CN202411067664.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing lane centerline extraction method is difficult to meet the needs of applications such as autonomous driving due to the high data acquisition cost and slow update speed, especially in complex intersection environments.

Method used

By matching the preset trajectory source with the road network, the road section trajectory is obtained and converted into a trajectory feature map, including density map and heading map, thereby identifying the lane centerline in the road section scene and intersection scene, improving extraction accuracy and efficiency.

Benefits of technology

This method improves the extraction accuracy and efficiency of the lane center line, can more accurately identify the lane center line under complex road conditions, and provides more stable data to support intelligent driving safety.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119007150A_ABST
    Figure CN119007150A_ABST
Patent Text Reader

Abstract

The invention provides a lane center line extraction method and device, equipment and a storage medium, and the method comprises the steps: obtaining a road section track of a target region according to a matching result between a preset track source and a road network; according to attribute information of a preset track feature map, converting the road track into a track feature map; wherein the track feature map comprises a density map and / or a course map; and identifying a first lane center line related to a road section scene and a second lane center line related to an intersection scene from the track feature map, and obtaining a lane center line of the target area according to an identification result. The lane center line is extracted through the intersection and the road section scene, and the extraction precision of the lane center line can be effectively improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of electronic map technology, and in particular to a lane centerline extraction method, device, equipment and storage medium. Background Art

[0002] The lane centerline is a reference line formed by connecting the center positions of lanes. It is an important underlying data in electronic maps (such as high-precision maps). The lane centerline can be used to estimate the number of lanes, lane boundaries, lane connectivity, etc., providing empirical data for autonomous driving, etc., and providing rich virtual lane information for lane-level navigation.

[0003] The current lane centerline extraction solution mainly uses artificial intelligence (AI) technologies such as image segmentation and point cloud segmentation to identify lane markings, ground edges, etc. However, due to the high cost of data collection and slow update speed, especially at complex intersections where multiple data collection is required to cover, the lane centerline extraction effect is difficult to meet actual needs. Summary of the invention

[0004] The present application provides a lane centerline extraction method, device, equipment and storage medium to solve at least one of the above-mentioned technical problems.

[0005] According to one aspect of the present application, a lane centerline extraction method is provided, comprising:

[0006] According to the matching result between the preset trajectory source and the road network, the road section trajectory of the target area is obtained;

[0007] According to the attribute information of the preset trajectory feature map, the road segment trajectory is converted into a trajectory feature map; wherein the trajectory feature map includes a density map and / or a heading map;

[0008] A first lane centerline for a road section scene and a second lane centerline for an intersection scene are identified from the trajectory feature map, and the lane centerline of the target area is acquired according to the identification result.

[0009] In one embodiment, the method further comprises:

[0010] Obtaining a road intersection connectivity relationship group and a road section mark of the target area; wherein the road intersection connectivity relationship group includes at least one of the following road intersection connectivity relationships: an entry road intersection, a passing road intersection, and an exit road intersection, and the road section mark is used to mark the road section information between every two adjacent road intersections;

[0011] The step of obtaining the road segment trajectory of the target area according to the matching result between the preset trajectory source and the road network includes:

[0012] The road segment trajectory is obtained according to the intersection connectivity relationship group, the road segment mark and the matching result.

[0013] In one embodiment, the matching result includes the corresponding position of each trajectory in the preset trajectory source on the road network;

[0014] The step of obtaining the road segment trajectory of the target area according to the matching result between the preset trajectory source and the road network includes:

[0015] For the target area, each trajectory is interrupted according to the road network and the corresponding position to obtain a road segment trajectory of the target area.

[0016] In one embodiment, the method for obtaining the matching result between the preset trajectory source and the road network includes:

[0017] Acquire real-time dynamic measurement RTK trajectory data, and pre-process the RTK trajectory data to obtain a preset trajectory source; wherein the pre-processing includes trajectory stringing processing and trajectory cleaning processing;

[0018] Each track in the preset track source is matched with the road network to obtain the corresponding position of each track in the road network, and obtain the matching result between the preset track source and the road network.

[0019] In one embodiment, the attribute information includes height and width, which are determined according to a preset operating range and pixel resolution;

[0020] The step of converting the road segment trajectory into a trajectory feature map according to the attribute information of the preset trajectory feature map comprises:

[0021] Mapping the road segment trajectory into a rasterized image to convert the trajectory line corresponding to the road segment trajectory into a pixel value, wherein the pixel values ​​are accumulated at the intersection position of any two intersecting trajectory lines;

[0022] Stretching the pixel value according to the height and width to obtain the density map;

[0023] In one implementation, converting the road segment trajectory into a trajectory feature map according to the attribute information of the preset trajectory feature map includes:

[0024] Determine the size of the initial heading map according to the height and width of the preset trajectory feature map;

[0025] Calculate and determine the heading value of the track line corresponding to the road segment track, and map the heading value to the pixel position of the initial heading map;

[0026] For each pixel position, sorting the heading values ​​in the pixel position to obtain a sorting result, and obtaining the heading value of each pixel position in the initial heading map according to the sorting result;

[0027] The heading map is obtained according to the heading value of each pixel position in the initial heading map.

[0028] In one embodiment, the trajectory line is a trajectory line obtained by connecting adjacent trajectory points;

[0029] The calculating and determining the heading value of the trajectory line corresponding to the road segment trajectory includes:

[0030] The heading value between the adjacent track points is calculated according to the longitude difference and the latitude difference between the adjacent track points to obtain the heading value of the track line corresponding to the track section.

[0031] In one embodiment, the identifying, from the trajectory feature graph, a first lane centerline in a road segment scene and a second lane centerline in an intersection scene includes:

[0032] The trajectory feature map is input into a pre-trained deep learning model for processing to identify the first lane centerline and the second lane centerline from the trajectory feature map, wherein the deep learning model is based on and trained based on a sample density map and / or a sample heading map, and the lane centerlines in the sample density map and / or the sample heading map are labeled.

[0033] According to a second aspect of the present application, a lane centerline extraction device is provided, comprising:

[0034] A trajectory acquisition module, which is configured to acquire a road segment trajectory of a target area according to a matching result between a preset trajectory source and a road network;

[0035] A conversion generation module, which is configured to convert the road segment trajectory into a trajectory feature map according to the attribute information of a preset trajectory feature map; wherein the trajectory feature map includes a density map and / or a heading map;

[0036] The recognition and extraction module is configured to recognize a first lane centerline of a road section scene and a second lane centerline of an intersection scene from the trajectory feature map, and obtain the lane centerline of the target area according to the recognition result.

[0037] In one embodiment, the device further comprises:

[0038] An intersection relationship acquisition module, configured to acquire an intersection connectivity relationship group and a road section mark of the target area; wherein the intersection connectivity relationship group includes at least one of the following intersection connectivity relationships: an entry intersection, a passing intersection, and an exit intersection, and the road section mark is used to mark the road section information between every two adjacent intersections;

[0039] The trajectory acquisition module is specifically configured to acquire the road segment trajectory according to the intersection connectivity relationship group, the road segment mark and the matching result.

[0040] In one embodiment, the matching result includes the corresponding position of each trajectory in the preset trajectory source on the road network;

[0041] The trajectory acquisition module is specifically configured to interrupt each trajectory according to the road network and the corresponding position for the target area to obtain the road segment trajectory of the target area.

[0042] In one embodiment, it further includes:

[0043] An RTK data acquisition module is configured to acquire real-time dynamic measurement RTK trajectory data and preprocess the RTK trajectory data to obtain a preset trajectory source; wherein the preprocessing includes trajectory stringing processing and trajectory cleaning processing;

[0044] A matching module is configured to match each trajectory in the preset trajectory source with the road network to obtain the corresponding position of each trajectory in the road network and obtain a matching result between the preset trajectory source and the road network.

[0045] In one embodiment, the attribute information includes height and width, which are determined according to a preset operating range and pixel resolution;

[0046] The conversion generation module comprises:

[0047] a conversion unit, configured to map the road segment trajectory into a rasterized image, so as to convert the trajectory line corresponding to the road segment trajectory into a pixel value, wherein the pixel value is accumulated at the intersection position of any two intersecting trajectory lines;

[0048] a stretching unit, configured to stretch the pixel value according to the height and the width to obtain the density map;

[0049] In one implementation, the conversion generation module includes:

[0050] a size determination unit, configured to determine the size of the initial heading map according to a height and a width of a preset trajectory characteristic map;

[0051] a heading value determination unit, configured to calculate and determine a heading value of a track line corresponding to the road segment track, and map the heading value to a pixel position of the initial heading map;

[0052] a sorting unit, configured to sort the heading values ​​in each pixel position for each pixel position to obtain a sorting result, and obtain the heading value of each pixel position in the initial heading map according to the sorting result;

[0053] An acquisition unit is configured to obtain the heading map according to the heading value of each pixel position in the initial heading map.

[0054] In one embodiment, the trajectory line is a trajectory line obtained by connecting adjacent trajectory points;

[0055] The heading value determination unit is specifically configured to calculate the heading value between the adjacent track points according to the longitude difference and latitude difference between the adjacent track points, so as to obtain the heading value of the track line corresponding to the track section.

[0056] In one embodiment, the recognition and extraction module is specifically configured to input the trajectory feature map into a pre-trained deep learning model for processing to identify the first lane centerline and the second lane centerline from the trajectory feature map, wherein the deep learning model is based on and trained based on a sample density map and / or a sample heading map, and the lane centerlines in the sample density map and / or the sample heading map are labeled.

[0057] According to the third aspect of the present application, a computer device is provided, which includes a memory and a processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory, so that the computer device executes the lane centerline extraction method provided in any one of the first aspects above.

[0058] According to a fourth aspect of the present application, a computer-readable storage medium is provided, wherein the storage medium includes computer-executable instructions, and when the computer-executable instructions are executed by a processor, the lane centerline extraction method provided in any one of the above-mentioned first aspects is used to implement.

[0059] According to a fifth aspect of the present application, a computer program product is provided, the computer program product comprising a computer program, which, when executed by a processor, implements the lane centerline extraction method provided in any one of the first aspects above.

[0060] The lane centerline extraction method, device, equipment and storage medium provided by the present application obtain the road segment trajectory of the target area according to the matching result between the preset trajectory source and the road network, convert the road segment trajectory into a trajectory feature map according to the attribute information of the preset trajectory feature map, and the trajectory feature map includes a density map and / or a heading map, and identifies the first lane centerline of the road segment scene and the second lane centerline of the intersection scene from the trajectory feature map, and obtains the lane centerline of the target area according to the recognition result. In this process, the lane centerline is extracted by replacing the trajectory clustering method in the related technology, by rendering the trajectory feature map, and identifying the road centerline of the road segment scene and the road centerline of the intersection scene from the trajectory feature map, and extracting the final road centerline according to the recognition results of different scenes, which can improve the extraction accuracy of the lane centerline and provide data support for intelligent driving safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0062] Figure 1 A possible scenario diagram provided for an embodiment of the present application;

[0063] Figure 2 A schematic diagram of a process for extracting a lane centerline provided in an embodiment of the present application;

[0064] Figure 3 A schematic diagram of a road segment trajectory in an embodiment of the present application;

[0065] Figure 4 A schematic diagram of the hue distribution of the HSV color model in an embodiment of the present application;

[0066] Figure 5a This is one of the schematic diagrams of the lane centerline of the intersection scene in the embodiment of the present application;

[0067] Figure 5b This is a second schematic diagram of the lane centerline of the intersection scene in the embodiment of the present application;

[0068] Figure 5c This is a third schematic diagram of the lane centerline of the intersection scene in the embodiment of the present application;

[0069] Figure 5d This is a fourth schematic diagram of the lane centerline of the intersection scene in the embodiment of the present application;

[0070] Figure 6a One of the schematic diagrams of the lane centerline of the road section scene in the embodiment of the present application

[0071] Figure 6bThis is a second schematic diagram of the lane centerline of the road section scene in an embodiment of the present application;

[0072] Figure 7 A schematic diagram of a process flow of another lane centerline extraction method provided in an embodiment of the present application;

[0073] Figure 8 A schematic diagram of an intersection connectivity relationship group in an embodiment of the present application;

[0074] Fig. 9 A schematic diagram of a process flow of another lane centerline extraction method provided in an embodiment of the present application;

[0075] Fig.10a A network structure diagram of a deep learning model in an embodiment of the present application;

[0076] Fig.10b for Fig.10a Schematic diagram of the structure of the encoder;

[0077] Fig.10c for Fig.10a Schematic diagram of the structure of the decoder;

[0078] Fig.10d A schematic diagram of a trajectory characteristic diagram in an embodiment of the present application;

[0079] Fig.11 A schematic flow chart of a lane centerline extraction method provided for an exemplary embodiment of the present application;

[0080] Fig.12 A schematic diagram of the structure of a lane centerline extraction device provided in an embodiment of the present application;

[0081] Fig.13 A schematic diagram of the structure of a computer device provided in an embodiment of the present application;

[0082] Fig.14 It is a block diagram of a terminal device provided by an exemplary embodiment of the present application.

[0083] The above drawings have shown clear embodiments of the present application, which will be described in more detail later. These drawings and text descriptions are not intended to limit the scope of the present application in any way, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0084] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the appended claims.

[0085] The embodiments of the present application are explained below in conjunction with application scenarios. The lane centerline extraction method provided in the embodiments of the present application can be applied to the application scenario of intelligent driving. More specifically, it can be applied to the application scenario of autonomous driving based on vehicle cloud computing. Exemplarily, the execution entity of the method provided in the embodiments of the present application may be a server. More specifically, it may be a server of a high-precision map producer. The server is introduced below as the execution entity of the method provided in the embodiments of the present application.

[0086] Figure 1 A schematic diagram of a scenario of a lane centerline extraction method provided in an embodiment of the present application, such as Figure 1 As shown, it includes a terminal device 110, a server 120 and a smart car 130 which are electrically connected in sequence, wherein the terminal device 110 is used to extract the lane centerline in the actual road and transmit the extracted lane centerline to the server 120, the server 120 generates or updates high-precision map data according to the lane centerline, and transmits the high-precision map data to the smart car 130, and the smart car 130 can use the high-precision map data to assist in automatic driving. Optionally, the terminal device 110 can be a computer, a smart phone, a tablet computer, a portable computer, a car computer, a wearable device, a desktop computer, a set-top box, a smart TV, etc.; the server 120 can be an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, and cloud computing.

[0087] Regarding the method of extracting the lane centerline by the terminal device 110 in the above process, in the related technology, in order to solve the problem that the data collection cost is high and the update speed is slow when AI recognizes lane markings, ground edges, etc., resulting in the problem that the lane centerline extraction effect cannot meet the actual needs, trajectory clustering analysis is performed on a subset of trajectories passing through the same road to obtain a trajectory cluster belonging to the same lane, and the trajectory geometric center is extracted from the trajectory subset in the same cluster in the form of a non-parametric curve, and the candidate lane centerline is obtained by fitting, and an intersection relationship graph is established based on the lane change lane line recognition method of maximum clique search, and the lane change lane line is identified and eliminated through maximum clique analysis to obtain a set of non-intersecting lane centerlines as the basic lane centerlines, and the number of lanes at both ends of the road section is verified through a Gaussian mixture model, and candidate lane centerlines are extracted as new lane centerlines based on the newly added Gaussian components and parameters. This method mainly uses trajectory clustering technology. The steps to achieve lane centerline fitting are relatively cumbersome. Since the clustering algorithm is not very robust under different trajectory density conditions, the error generated by clustering may propagate to the next step. For example, in the scenario where there is one lane entering the intersection and two lanes exiting the intersection, this method will fail in the complex overlapping connectivity relationship of the trajectory entering and exiting the intersection. It will also fail in the scenario where the number of lanes at the beginning and end of the road section are inconsistent, which will result in a low accuracy of the extracted lane centerline.

[0088] In view of this, the embodiments of the present application provide a lane centerline extraction method, device, equipment and storage medium, which obtains the road segment trajectory of the target area according to the matching result between the preset trajectory source and the road network, and generates a trajectory feature map according to the attribute information of the preset trajectory feature map, and renders the road segment trajectory. The trajectory feature map includes a density map and / or a heading map, and identifies the first lane centerline of the road segment scene and the second lane centerline of the intersection scene from the trajectory feature map, and obtains the lane centerline of the target area according to the recognition result. In this process, the lane centerline is extracted by replacing the trajectory clustering method in the related technology, by generating a trajectory feature map, and identifying the road centerline of the road segment scene and the road centerline of the intersection scene from the trajectory feature map, and extracting the final road centerline according to the recognition results of different scenes, which can improve the extraction accuracy and efficiency of the lane centerline and provide data support for intelligent driving safety.

[0089] The following is a detailed description of the technical solution of the present application and how the technical solution of the present application solves the above technical problems in conjunction with the accompanying drawings and specific embodiments. It should be noted that these specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0090] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0091] Please refer to Figure 2 , Figure 2 A schematic flow chart of a lane centerline extraction method provided in an embodiment of the present application, the method comprising steps S201-S203:

[0092] Step S201: Obtain the road segment trajectory of the target area according to the matching result between the preset trajectory source and the road network.

[0093] In this embodiment, the preset trajectory source can adopt a real-time kinematic (RTK) trajectory source. As a crowdsourcing data source, the trajectory source can be used to identify the road surface, the vehicle drivable area, etc., wherein the RTK trajectory can achieve a horizontal accuracy of 50cm and an elevation accuracy of 1 meter under a fixed solution of 2σ, and can achieve lane-level positioning, which is of great value for extracting lane-level elements.

[0094] In this embodiment, the trajectory data can be processed by road segment according to the road network structure. Each trajectory segment corresponds to a road segment, and the trajectory data on each road segment is summarized to generate comprehensive trajectory information of each road segment, which can obtain the road segment trajectory by combining the density, average speed, driving direction and other information of the trajectory points, and obtain the road segment trajectory corresponding to the target area. Figure 3 As shown. It can be understood that the target area can be an area determined by the user where the road centerline needs to be extracted, and the corresponding area size can be determined according to geographical boundaries, administrative divisions, points of interest (POI), etc.

[0095] In some embodiments, the method for obtaining the matching result between the preset trajectory source and the road network may include the following steps:

[0096] Acquire real-time dynamic measurement RTK trajectory data, and pre-process the RTK trajectory data to obtain a preset trajectory source; wherein the pre-processing includes trajectory stringing processing and trajectory cleaning processing;

[0097] Each track in the preset track source is matched with the road network to obtain the corresponding position of each track in the road network, and obtain the matching result between the preset track source and the road network.

[0098] In this example, the trajectory source is preprocessed and matched with the road network to obtain the section trajectory. Specifically, the crowdsourced trajectory is cleaned and matched with the road network: the RTK trajectory is used as the input trajectory source, where the RTK trajectory has high positioning accuracy and the trajectory within the lane has an obvious hot spot. Trajectory preprocessing includes stringing and cleaning the trajectory, that is, trajectory stringing is to sort the trajectory points of the same device (such as a vehicle) by time to form a continuous trajectory line to ensure the time consistency of the trajectory data. Trajectory cleaning is to remove and correct the outliers of the trajectory data to ensure the spatial continuity and smoothness of the trajectory. Specific operations may include: residence point removal, removing the trajectory points where the vehicle stays for a long time; large angle removal, removing the abnormally large angles in the trajectory, which may be caused by positioning errors or other reasons; speed anomaly removal, removing the trajectory points with abnormal speed, which may be caused by positioning errors or equipment failures; distance anomaly removal, removing the points with abnormally large distances between adjacent trajectory points, which may be caused by signal loss or other reasons.

[0099] Next, after preprocessing the trajectory data, that is, matching the cleaned trajectory with the road network, finding the corresponding relationship between the trajectory and the road network, and breaking the trajectory according to the road network to form a segment trajectory. Optionally, using a map matching algorithm, the cleaned trajectory points are matched with the roads in the road network to determine the road segment corresponding to each trajectory point, and the matched trajectory can be broken according to the road network structure to split the trajectory into multiple segment trajectories, which helps to more accurately analyze the driving behavior of the vehicle on different road sections.

[0100] Specifically, in order to improve the accuracy of obtaining road section trajectories, the matching result includes the corresponding position of each trajectory in the preset trajectory source on the road network. In the above steps, the road section trajectory of the target area is obtained according to the matching result between the preset trajectory source and the road network, which may include the following steps: for the target area, each trajectory is interrupted according to the road network and the corresponding position to obtain the road section trajectory of the target area.

[0101] In this embodiment, by breaking the trajectory according to the road network and its corresponding position in the target area, the section trajectory of each section in the target area can be more accurately obtained. The section may refer to the section between two adjacent intersections.

[0102] Continue to refer to Figure 2 , step S202, converting the road segment trajectory into a trajectory feature map according to the attribute information of the preset trajectory feature map; wherein the trajectory feature map includes a density map and / or a heading map.

[0103] In this embodiment, considering the data noise and error of trajectory data, data sparsity, and the difficulty in distinguishing lanes in complex road conditions, the accuracy of lane line extraction is low when the road section trajectory is directly used (such as by clustering, etc.). This embodiment can aggregate and process a large amount of trajectory data by converting the road section trajectory into a trajectory feature map, effectively smoothing and filtering noise, and improving accuracy; and in some sections, the vehicle trajectory data may not be dense enough, and the trajectory feature map can help identify data sparse areas and fill these gaps through data interpolation or other methods. And in complex road conditions (such as multiple lanes, highway intersections, roundabouts, etc.), the trajectory feature map (such as a density map) can better represent the vehicle's driving direction and density distribution, so as to more clearly identify and distinguish different lanes. Since the trajectory feature map integrates a large amount of vehicle trajectory data, it can provide more comprehensive and reliable road information, help eliminate the influence of individual abnormal trajectories, and obtain a more stable lane centerline. In addition, extracting lane centerlines directly from trajectory data may require complex algorithms and processing steps, but this process can be simplified through heading maps and density maps. Image processing techniques (such as edge detection, cluster analysis, etc.) can more efficiently extract lane centerlines from images.

[0104] By converting road segment trajectories into trajectory feature maps and then extracting lane center lines from these trajectory feature maps, it is easy to identify different scenarios, improve the accuracy and reliability of extraction, simplify the processing process, and better cope with complex road conditions and data problems.

[0105] In this embodiment, the attribute information of the preset trajectory feature map includes height and width, which are determined according to the preset operating range and pixel resolution. Specifically, the user can determine the height and width of the feature map according to actual needs, so that the rendered trajectory feature map meets the user's needs.

[0106] In one example, in the above step S202, converting the road segment trajectory into a trajectory feature map according to the attribute information of a preset trajectory feature map may include the following steps: mapping the road segment trajectory into a rasterized image to convert the trajectory line corresponding to the road segment trajectory into a pixel value, wherein the intersection position of any two intersecting trajectory lines is accumulated with pixel values; and stretching the pixel values ​​according to the height and width to obtain the density map.

[0107] In this example, the density map is generated. The height and width of the feature map are set according to the operating range (which can be the target area range or the range enlarged or reduced based on the target area) and the pixel resolution. The trajectory points are connected into lines and rasterized. If two trajectory lines intersect, the pixel values ​​are accumulated at the intersection, and finally the pixel values ​​are stretched. The formula is as follows:

[0108]

[0109] Where vi is the pixel value of the i-th grid (that is, the trajectory density value), pi is the result of calculating the logarithm of the pixel value, hi is the pixel value after stretching, w is the width, and h is the height.

[0110] It can be understood that the trajectory line corresponding to the road segment trajectory is each trajectory in the road segment trajectory (the trajectory line formed by stringing together the trajectory points).

[0111] Through the above technical solution, the road section trajectory is converted into a density map, which can efficiently display the vehicle's movement trajectory and its density distribution, and provide valuable information for the identification of road sections and intersection scenes, so as to facilitate the extraction of lane center lines.

[0112] In another example, the above-mentioned step S202 converts the road segment trajectory into a trajectory feature map according to the attribute information of the preset trajectory feature map, and may include the following steps: determining the size of the initial heading map according to the height and width of the preset trajectory feature map; calculating and determining the heading value of the trajectory line corresponding to the road segment trajectory, and mapping the heading value to the pixel position of the initial heading map; for each pixel position, sorting the heading values ​​in the pixel position to obtain the sorting result, and obtaining the heading value of each pixel position in the initial heading map according to the sorting result; obtaining the heading map according to the heading value of each pixel position in the initial heading map.

[0113] In this example, the generation of a heading map is involved. According to the operating range and image resolution, the height and width of the heading map are set, and the track points are connected (i.e., track lines, so as to calculate the heading values ​​between adjacent track points). The heading values ​​of the current track point and the next track point are calculated and stored in the corresponding pixel positions. The same pixel position will store the heading values ​​of multiple tracks. Finally, the heading values ​​are sorted, and the median value can be taken according to the sorting result as the heading value of the current pixel. Optionally, in order to continuously express the heading map, the HSV (Hue, Saturation, Value) color space model can be used to express it, where H is the heading value, the value range is [0,359], S=1, V=1. Among them, the hue distribution of the HSV color model is as follows Figure 4 As shown, it can be understood that the HSV color model is a commonly used color representation model, which describes color through three parameters: hue, saturation, and transparency. In this embodiment, S and V take a constant value of 1. The distribution of hue H in the hue distribution diagram of the HSV color model can be represented by a color wheel (0° to 360°), as shown in FIG. Figure 4As shown in the figure, 0° represents red, 60° represents yellow, 120° represents green, 180° represents cyan, 240° represents blue, and 300° represents magenta. This method can intuitively represent different heading values ​​by color, making the data visualization clearer and easier to understand.

[0114] Furthermore, the trajectory line in this example is a trajectory line obtained by connecting adjacent trajectory points; the above steps of calculating and determining the heading value of the trajectory line corresponding to the road section trajectory may include the following steps: calculating the heading value between the adjacent trajectory points based on the longitude difference and latitude difference between the adjacent trajectory points to obtain the heading value of the trajectory line corresponding to the track section.

[0115] For example, the heading values ​​of adjacent track points (i.e., the current track point and the next track point) can be obtained according to the following formula:

[0116] dlon=gps i+1 lon-gps i lon

[0117] y=sin(dlon)*cos(gps i+1 lat)

[0118] x=cos(gps i lat)*sin(gps i+1 lat)-sin(gps i lat)*cos(gps i+1 lat)

[0119] *cos(dlon)

[0120]

[0121] In the formula, dlon represents the longitude difference between the i+1th point and the ith point, GPS i+1 lon represents the longitude of the i+1th point, gps i lon represents the longitude of the i-th point, gps i+1 lat represents the latitude of the i+1th point, x represents the horizontal displacement between the current track point and the next track point, y represents the vertical displacement between the current track point and the next track point, V dir Indicates the heading value between the current track point and the next track point.

[0122] Continue to refer to Figure 2 , step S203, identifying a first lane centerline for a road section scene and a second lane centerline for an intersection scene from the trajectory feature map, and acquiring the lane centerline of the target area according to the identification result.

[0123] In this embodiment, by dividing the road network scenes at the business level, lane geometry extraction of the entire road network is achieved to meet different business needs. Specifically, after converting the trajectory segment into a trajectory feature map, the different scenes in the trajectory feature map are identified to realize the recognition of lane centerlines in intersection and section scenarios (for example, deep learning models can be used for recognition, and the specific recognition process can refer to the embodiments described later; in some embodiments, computer vision image processing methods such as edge detection can also be used to identify lane centerlines in different scenarios), so that the final extracted lane centerlines include the lane centerlines of intersections and the lane centerlines of sections. Among them, an intersection is a part where two or more roads intersect in the same plane. The intersection scene specified in this embodiment includes three states of continuous trajectories entering the intersection, passing the intersection, and exiting the intersection. The section is the area between two intersections. Among them, Figure 5a-5d This is an example of a lane centerline for an intersection scenario: Figure 5a It is the intersection from west to east. Figure 5b It is an intersection entering from east to west. Figure 5c Enter the intersection from south to north, Figure 5d It is the intersection for entering from north to south; Figure 6a and Figure 6b An example of lane centerline for a road segment scenario.

[0124] Through the above technical solution, the lane centerline is extracted by replacing the trajectory clustering method in the related technology. By generating a trajectory feature map and identifying the road centerline of the road section scene and the road centerline of the intersection scene from the trajectory feature map, the final road centerline is extracted according to the recognition results of different scenes. This can improve the extraction accuracy and efficiency of the lane centerline and provide data support for intelligent driving safety.

[0125] Figure 7 It is a flow chart of another lane centerline extraction method provided in an embodiment of the present application. On the basis of the above embodiment, this embodiment combines information such as intersection connectivity relationship groups and road section markings to obtain road section trajectories to improve the accuracy of road section trajectories, so as to facilitate the subsequent lane centerline extraction process. Specifically, in addition to the above steps S201-S203, the method provided in this embodiment may also include the following step S701, and further divide the above step S201 into step S2011.

[0126] Step S701, obtaining the intersection connectivity relationship group and the road section mark of the target area; wherein the intersection connectivity relationship group includes at least one of the following intersection connectivity relationships: entering an intersection, passing an intersection, and exiting an intersection, and the road section mark is used to mark the road section information between every two adjacent intersections.

[0127] In this embodiment, entering an intersection may correspond to the behavior of a vehicle or pedestrian entering a specific intersection from a certain road section in the trajectory data; passing an intersection may correspond to the behavior of a vehicle or pedestrian passing an intersection but not stopping or changing direction at the intersection in the trajectory data; exiting an intersection may correspond to the behavior of a vehicle or pedestrian leaving a specific intersection and entering another road section in the trajectory data.

[0128] In this embodiment, the road segment mark is used to mark the road segment information between every two adjacent intersections. The road segment information may include but is not limited to the length, width, road conditions, traffic light settings, number of lanes, driving direction, speed limit information, and other factors that may affect traffic flow and driving safety (for example, each road segment has a unique identifier ID and may contain other attribute information such as length, direction, number of lanes, etc.). By obtaining these detailed intersection connectivity relationships and road segment mark information, the road segment trajectory can be divided more accurately, thereby laying the foundation for the subsequent lane centerline extraction in different scenarios.

[0129] For example, the intersection connectivity relationship group can be as follows: Figure 8 As shown in the figure, the digital marks on the map are virtual roads connected within the intersection, which can only be used to express traffic relations. First, it is necessary to mark all virtual roads within the intersection and the roads outside the intersection connected to them according to the SD map attributes, and then build a topological relationship in the local area of ​​the intersection, and calculate all the connectivity relationships of each incoming road. Figure 8 In this example, an entry road (shown in red) has three exit roads (shown in blue, green, and yellow). All entry roads are traversed and the entry-exit relationship groups are stored or recorded. Similarly, a road segment is a road between intersections, and the road segments are searched and marked.

[0130] It should be noted that some of the drawings in this embodiment use different colors only to distinguish different objects to facilitate the understanding of the embodiments of the present application, but not to limit the specific content.

[0131] Step S2011: Acquire the road segment trajectory according to the intersection connectivity relationship group, the road segment mark and the matching result.

[0132] In this embodiment, according to the intersection connectivity relationship group and the road segment mark, and the matching relationship between the trajectory and the road network, the trajectory passing through each intersection relationship group and the road segment can be extracted to obtain the road segment trajectory, and combined with the density map and the heading map in the above example, and subsequent recognition can be performed. In an implementable manner, the following process can be included:

[0133] First, the intersection connectivity group and the road segment mark, the trajectory matching result are obtained (the nearest neighbor search algorithm, such as KD tree or R tree, can be used to preliminarily match each trajectory point to the nearest road segment according to the coordinates of the trajectory point, and the map matching algorithm (such as hidden Markov model, shortest path algorithm, etc.) can be used to optimize the matching results of the trajectory point and the road segment to ensure the rationality and continuity of the trajectory in the road network), and the trajectory data is divided into multiple (initial) road segment trajectories according to the matching results. Each road segment trajectory corresponds to the driving situation of the trajectory data on a specific road segment, and each road segment trajectory can be associated with timestamp information for subsequent analysis. Next, the intersection connectivity group is processed to identify each intersection passed by the trajectory, and the sequence of intersections passed by the trajectory is determined according to the information of the intersection connectivity group, and the specific path of the trajectory in the road network is determined according to the sequence of intersections passed by the trajectory, and the continuity and rationality of the trajectory need to be ensured. Finally, the trajectory is extracted, and the complete road segment trajectory can be reconstructed according to the road segment mark and the trajectory matching result. In some implementations, in order to ensure the continuity and consistency of the trajectory. The reconstructed road segment trajectory can also be verified (such as using the topological information of the road network to verify that the trajectory does not violate the road connectivity) to ensure that it conforms to the actual driving path.

[0134] Through the above technical solution, the section trajectory is obtained by combining the information of intersection connectivity relationship group, section marking and so on, which can improve the accuracy of the section trajectory and facilitate the subsequent lane centerline extraction process.

[0135] Fig. 9 This is a flow chart of another lane centerline extraction method provided by an embodiment of the present application. Based on the above embodiment, this embodiment uses a deep learning model to further optimize the recognition efficiency and accuracy of the lane centerline. Specifically, in addition to the above steps S201-S203, this embodiment further divides the above step S203 into the following steps S2031:

[0136] Step S2031: input the trajectory feature map into a pre-trained deep learning model for processing, so as to identify the first lane centerline of the road section scene and the second lane centerline of the intersection scene from the trajectory feature map, and obtain the lane centerline of the target area according to the recognition result, wherein the deep learning model is based on and trained based on the sample density map and / or the sample heading map, and the lane centerlines in the sample density map and / or the sample heading map are labeled.

[0137] In this embodiment, the trajectory feature map is input into a pre-trained deep learning model. The trajectory feature map is generated by the road segment trajectory, which contains various information of the road, such as lane lines, road signs, etc. The input trajectory feature map is processed using a pre-trained deep learning model. The model has been trained with a large amount of sample data and has the ability to recognize and process trajectory feature maps. After being processed by the deep learning model, the output result includes the center line of the first lane and the center line of the second lane. These center lines are an accurate description of the lanes in the trajectory feature map, which facilitates subsequent high-precision map updates, as well as navigation and path planning. Among them, the training of the deep learning model is based on the sample density map and / or the sample heading map. In these sample maps, the lane center lines have been labeled, the sample density map provides the density information of the lane lines, and the sample heading map provides the direction information of the lane lines. In this way, the lane center lines can be efficiently extracted from the trajectory feature map, providing accurate data support for applications such as autonomous driving and navigation systems. Optionally, the deep learning model can use a neural network such as an encoder-decoder structure, and its network structure is as follows: Fig.10a As shown, the encoder part refers to the Encoder network of D-LinkNet, such as Fig.10b As shown in Figure 1, it includes the Encoder layer and the dilated convolution layer. The decoder part is as follows Fig.10c As shown in Figure 1, the one-dimensional convolution in four directions is used to better capture the information in the road direction and is more robust to the input of sparse trajectory points. Fig.10d As shown in (a), (b), and (c) respectively illustrate the sample heading map (different colors represent different heading values), sample density, and annotation map.

[0138] In this embodiment, by labeling the lane centerline, an accurate and learnable target can be provided to the model so that the model can correctly identify and extract the lane centerline from the input data. Optionally, a special labeling tool or software can be used to manually or semi-automatically label the lane centerline on these trajectory feature maps. The labeling tool can provide functions such as drawing lines and adjusting line positions.

[0139] Through the above technical solution, the lane centerline recognition efficiency and accuracy can be further optimized by using a deep learning model.

[0140] To facilitate understanding of the embodiments of the present application, in an exemplary embodiment, Fig.11 As shown in the figure, the lane centerline extraction process for different scenarios is as follows:

[0141] S1. Obtain high-precision (HD) trajectory data such as RTK trajectory data, and use the RTK trajectory data to perform road network matching to obtain a matching result;

[0142] S2, obtaining standard precision (SD) map data, the SD map data including road information, such as intersections, road sections and other information, and extracting intersections and road sections according to the SD map data; specifically, the following steps may be included:

[0143] S2.1. Use digital labels (or other methods) to mark the internal and external roads connected to the intersection, construct the local topological relationship of the intersection based on the intersection marking results, and construct the intersection connectivity relationship group based on the local topological relationship of the intersection;

[0144] S2.2. Use ID (or other methods) to mark the road sections, and merge the sections in the same direction to obtain the road section marking;

[0145] S3, executing the trajectory extraction and rendering process: extracting the segment trajectory according to the matching results, the intersection connectivity relationship group, the segment mark and other information; and generating a density map and a heading map according to the segment trajectory rendering;

[0146] S4, lane perception in all directions of road sections and intersections. This process can use the deep learning model to identify the lane center lines in intersection scenes and road section scenes from the density map and heading map, and vectorize the extracted lane center lines, that is, extract the skeleton line of the model output result and vectorize it to generate the lane center line vector.

[0147] In summary, the technical solution provided by this embodiment is to obtain high-quality road section trajectory clusters by cleaning and matching the RTK trajectory with the road network, and to produce two kinds of density and heading grid maps according to the trajectory density and the direction of travel. At the same time, in order to achieve the goal of efficiently and accurately identifying the center line of the lane, the center line of the lane is marked according to the trajectory density map and the heading map, and a deep convolutional neural network with an Encoder-Decoder structure is designed to segment the center line of the lane. The Encoder structure of D-LinkNet is used as the encoder of the model, and one-dimensional convolutions in four directions are introduced into the decoder structure to capture the discrete information of the trajectory in the road direction and improve the robustness of the model; the road network is divided into intersections and road section scenes, and the trajectory of each scene is extracted and rendered respectively, and the above-mentioned trained model is used for reasoning to obtain the center line of the lane in each scene. It can be seen that compared with the defects of the clustering algorithm used in the related technology that is sensitive to data, complex steps, and difficult to implement, the embodiment of the present application uses a deep learning method to achieve end-to-end learning, and performs more robustly in different scenes and different trajectory density distributions. In addition, in intersection scenarios, the virtual lane connectivity relationship of the intersection can be identified. In road section scenarios, it can not only identify scenarios with a stable number of lanes, but also identify situations where the number of lanes changes when entering an intersection, effectively solving problems in related technologies such as the failure of lane centerline extraction at intersections.

[0148] Fig.12 is a structural diagram of a lane centerline extraction device provided in an embodiment of the present application, such as Fig.12 As shown, the device includes a trajectory acquisition module 121, a conversion generation module 122, and a recognition and extraction module 133, wherein:

[0149] A trajectory acquisition module 121, which is configured to acquire a road segment trajectory of a target area according to a matching result between a preset trajectory source and a road network;

[0150] A conversion generation module 122, which is configured to convert the road segment trajectory into a trajectory feature map according to the attribute information of a preset trajectory feature map; wherein the trajectory feature map includes a density map and / or a heading map;

[0151] The recognition and extraction module 133 is configured to recognize the first lane centerline of the road section scene and the second lane centerline of the intersection scene from the trajectory feature map, and obtain the lane centerline of the target area according to the recognition result.

[0152] In one embodiment, the device further comprises:

[0153] An intersection relationship acquisition module, configured to acquire an intersection connectivity relationship group and a road section mark of the target area; wherein the intersection connectivity relationship group includes at least one of the following intersection connectivity relationships: an entry intersection, a passing intersection, and an exit intersection, and the road section mark is used to mark the road section information between every two adjacent intersections;

[0154] The trajectory acquisition module 121 is specifically configured to acquire the road segment trajectory according to the intersection connectivity relationship group, the road segment mark and the matching result.

[0155] In one embodiment, the matching result includes the corresponding position of each trajectory in the preset trajectory source on the road network;

[0156] The trajectory acquisition module 121 is specifically configured to interrupt each trajectory according to the road network and the corresponding position for the target area to obtain the road segment trajectory of the target area.

[0157] In one embodiment, it further includes:

[0158] An RTK data acquisition module is configured to acquire real-time dynamic measurement RTK trajectory data and preprocess the RTK trajectory data to obtain a preset trajectory source; wherein the preprocessing includes trajectory stringing processing and trajectory cleaning processing;

[0159] A matching module is configured to match each trajectory in the preset trajectory source with the road network to obtain the corresponding position of each trajectory in the road network and obtain a matching result between the preset trajectory source and the road network.

[0160] In one embodiment, the attribute information includes height and width, which are determined according to a preset operating range and pixel resolution;

[0161] The conversion generation module 122 includes:

[0162] a conversion unit, configured to map the road segment trajectory into a rasterized image, so as to convert the trajectory line corresponding to the road segment trajectory into a pixel value, wherein the pixel value is accumulated at the intersection position of any two intersecting trajectory lines;

[0163] a stretching unit, configured to stretch the pixel value according to the height and the width to obtain the density map;

[0164] In one implementation, the conversion generation module 122 includes:

[0165] a size determination unit, configured to determine the size of the initial heading map according to a height and a width of a preset trajectory characteristic map;

[0166] a heading value determination unit, configured to calculate and determine a heading value of a track line corresponding to the road segment track, and map the heading value to a pixel position of the initial heading map;

[0167] a sorting unit, configured to sort the heading values ​​in each pixel position for each pixel position to obtain a sorting result, and obtain the heading value of each pixel position in the initial heading map according to the sorting result;

[0168] An acquisition unit is configured to obtain the heading map according to the heading value of each pixel position in the initial heading map.

[0169] In one embodiment, the trajectory line is a trajectory line obtained by connecting adjacent trajectory points;

[0170] The heading value determination unit is specifically configured to calculate the heading value between the adjacent track points according to the longitude difference and latitude difference between the adjacent track points, so as to obtain the heading value of the track line corresponding to the track section.

[0171] In one embodiment, the recognition and extraction module 123 is specifically configured to input the trajectory feature map into a pre-trained deep learning model for processing to identify the first lane centerline and the second lane centerline from the trajectory feature map, wherein the deep learning model is based on and trained based on a sample density map and / or a sample heading map, and the lane centerlines in the sample density map and / or the sample heading map are labeled.

[0172] Fig.13 is a computer device provided in an embodiment of the present application, such as Fig.13 As shown, the computer device includes a processor 131, and a memory 132 communicatively connected to the processor 131;

[0173] The memory 132 stores computer-executable instructions;

[0174] The processor 131 executes the computer-executable instructions stored in the memory 132 to implement the lane centerline extraction method, wherein the memory 132 and the processor 131 are connected via a bus 133 .

[0175] The relevant instructions can be understood by referring to the relevant descriptions and effects corresponding to the steps in the method embodiments of the present application, and no further elaboration is made here.

[0176] An embodiment of the present application also provides a computer-readable storage medium, which includes computer-executable instructions. When the computer-executable instructions are executed by a processor, they are used to implement the lane centerline extraction method provided in any one of the first aspects above.

[0177] Among them, the computer-readable storage medium may be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.

[0178] The relevant instructions can be understood by referring to the relevant descriptions and effects corresponding to the steps in the method embodiments of the present application, and no further elaboration is made here.

[0179] An embodiment of the present application also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, it implements the lane centerline extraction method provided in any one of the first aspects above.

[0180] The relevant instructions can be understood by referring to the relevant descriptions and effects corresponding to the steps in the method embodiments of the present application, and no further elaboration is made here.

[0181] The embodiment of the present application also provides a chip, including a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to call and run the computer program from the memory to execute the lane centerline extraction method.

[0182] The relevant instructions can be understood by referring to the relevant descriptions and effects corresponding to the steps in the method embodiments of the present application, and no further elaboration is made here.

[0183] Fig.14 It is a block diagram of a terminal device shown in an exemplary embodiment of the present application. The terminal device 800 can be a mobile phone, a computer, a digital broadcast terminal, a message transceiver device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0184] The terminal device 800 may include one or more of the following components: a processing component 802 , a memory 804 , a power component 806 , a multimedia component 808 , an audio component 810 , an input / output (I / O) interface 812 , a sensor component 814 , and a communication component 816 .

[0185] The processing component 802 generally controls the overall operation of the terminal device 800, such as operations associated with display, phone calls, data communications, camera operations, and recording operations. The processing component 802 may include one or more processors 820 to execute instructions to complete all or part of the steps of the above-mentioned method. In addition, the processing component 802 may include one or more modules to facilitate the interaction between the processing component 802 and other components. For example, the processing component 802 may include a multimedia module to facilitate the interaction between the multimedia component 808 and the processing component 802.

[0186] The memory 804 is configured to store various types of data to support operations on the terminal device 800. Examples of such data include instructions for any application or method operating on the terminal device 800, contact data, phone book data, messages, pictures, videos, etc. The memory 804 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0187] The power supply component 806 provides power to various components of the terminal device 800. The power supply component 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the terminal device 800.

[0188] The multimedia component 808 includes a screen that provides an output interface between the terminal device 800 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundaries of the touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front camera and / or a rear camera. When the terminal device 800 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera may receive external multimedia data. Each front camera and the rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

[0189] The audio component 810 is configured to output and / or input audio signals. For example, the audio component 810 includes a microphone (MIC), and when the terminal device 800 is in an operating mode, such as a call mode, a recording mode, and a speech recognition mode, the microphone is configured to receive an external audio signal. The received audio signal can be further stored in the memory 804 or sent via the communication component 816. In some embodiments, the audio component 810 also includes a speaker for outputting audio signals.

[0190] I / O interface 812 provides an interface between processing component 802 and peripheral interface modules, such as keyboards, click wheels, buttons, etc. These buttons may include but are not limited to: home button, volume button, start button, and lock button.

[0191] The sensor assembly 814 includes one or more sensors for providing various aspects of status assessment for the terminal device 800. For example, the sensor assembly 814 can detect the open / closed state of the terminal device 800, the relative positioning of the components, such as the display and keypad of the terminal device 800, and the sensor assembly 814 can also detect the position change of the terminal device 800 or a component of the terminal device 800, the presence or absence of contact between the user and the terminal device 800, the orientation or acceleration / deceleration of the terminal device 800, and the temperature change of the terminal device 800. The sensor assembly 814 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 814 may also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0192] The communication component 816 is configured to facilitate wired or wireless communication between the terminal device 800 and other devices. The terminal device 800 can access a wireless network based on a communication standard, such as WiFi, 3G, 4G, 5G or other standard communication networks, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 816 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology and other technologies.

[0193] In an exemplary embodiment, the terminal device 800 can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to execute the method provided in any of the above-mentioned method embodiments of the present application.

[0194] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 804 including instructions, and the instructions can be executed by a processor 820 of a terminal device 800 to complete the above method. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0195] An embodiment of the present application also provides a non-temporary computer-readable storage medium. When the instructions in the storage medium are executed by the processor of the terminal device, the terminal device 800 can execute the method provided by any one of the above-mentioned method embodiments of the present application.

[0196] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or modules, which can be electrical, mechanical or other forms.

[0197] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the application disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0198] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A lane centerline extraction method, characterized in that: include: According to the matching result between the preset trajectory source and the road network, the road section trajectory of the target area is obtained; According to the attribute information of the preset trajectory feature map, the road segment trajectory is converted into a trajectory feature map; wherein the trajectory feature map includes a density map and / or a heading map; A first lane centerline for a road section scene and a second lane centerline for an intersection scene are identified from the trajectory feature map, and the lane centerline of the target area is acquired according to the identification result.

2. The method according to claim 1, characterized in that Also includes: Obtaining a road intersection connectivity relationship group and a road section mark of the target area; wherein the road intersection connectivity relationship group includes at least one of the following road intersection connectivity relationships: an entry road intersection, a passing road intersection, and an exit road intersection, and the road section mark is used to mark the road section information between every two adjacent road intersections; The step of obtaining the road segment trajectory of the target area according to the matching result between the preset trajectory source and the road network includes: The road segment trajectory is obtained according to the intersection connectivity relationship group, the road segment mark and the matching result.

3. The method according to claim 1 or 2, characterized in that: The matching result includes the corresponding position of each trajectory in the preset trajectory source on the road network; The step of obtaining the road segment trajectory of the target area according to the matching result between the preset trajectory source and the road network includes: For the target area, each trajectory is interrupted according to the road network and the corresponding position to obtain a road segment trajectory of the target area.

4. The method according to claim 3, characterized in that The method for obtaining the matching result between the preset trajectory source and the road network includes: Acquire real-time dynamic measurement RTK trajectory data, and pre-process the RTK trajectory data to obtain a preset trajectory source; wherein the pre-processing includes trajectory stringing processing and trajectory cleaning processing; Each track in the preset track source is matched with the road network to obtain the corresponding position of each track in the road network, and obtain the matching result between the preset track source and the road network.

5. The method according to claim 1, characterized in that The attribute information includes height and width, which are determined according to a preset operating range and pixel resolution; The step of converting the road segment trajectory into a trajectory feature map according to the attribute information of the preset trajectory feature map comprises: Mapping the road segment trajectory into a rasterized image to convert the trajectory line corresponding to the road segment trajectory into a pixel value, wherein the pixel values ​​are accumulated at the intersection position of any two intersecting trajectory lines; The pixel value is stretched according to the height and width to obtain the density map.

6. The method according to claim 1 or 5, characterized in that: The step of converting the road segment trajectory into a trajectory feature map according to the attribute information of the preset trajectory feature map comprises: Determine the size of the initial heading map according to the height and width of the preset trajectory feature map; Calculate and determine the heading value of the track line corresponding to the road segment track, and map the heading value to the pixel position of the initial heading map; For each pixel position, sorting the heading values ​​in the pixel position to obtain a sorting result, and obtaining the heading value of each pixel position in the initial heading map according to the sorting result; The heading map is obtained according to the heading value of each pixel position in the initial heading map.

7. The method according to claim 5, characterized in that The trajectory line is a trajectory line obtained by connecting adjacent trajectory points; The calculating and determining the heading value of the trajectory line corresponding to the road segment trajectory includes: The heading value between the adjacent track points is calculated according to the longitude difference and the latitude difference between the adjacent track points to obtain the heading value of the track line corresponding to the track section.

8. The method according to any one of claims 1 to 5, characterized in that: The step of identifying the center line of a first lane in a road section scene and the center line of a second lane in an intersection scene from the trajectory feature graph includes: The trajectory feature map is input into a pre-trained deep learning model for processing to identify the first lane centerline and the second lane centerline from the trajectory feature map, wherein the deep learning model is based on and trained based on a sample density map and / or a sample heading map, and the lane centerlines in the sample density map and / or the sample heading map are labeled.

9. A lane centerline extraction device, characterized in that: include: A trajectory acquisition module, which is configured to acquire a road segment trajectory of a target area according to a matching result between a preset trajectory source and a road network; A conversion generation module, which is configured to convert the road segment trajectory into a trajectory feature map according to the attribute information of a preset trajectory feature map; wherein the trajectory feature map includes a density map and / or a heading map; The recognition and extraction module is configured to recognize a first lane centerline of a road section scene and a second lane centerline of an intersection scene from the trajectory feature map, and obtain the lane centerline of the target area according to the recognition result.

10. A computer device / computer readable storage medium / computer program product, characterized in that: The computer device comprises a memory and a processor; the memory stores computer-executable instructions; the processor executes the computer-executable instructions stored in the memory, so that the computer device executes the lane centerline extraction method according to any one of claims 1 to 8; and / or, The computer-readable storage medium includes computer-executable instructions, and when the computer-executable instructions are executed by a processor, they are used to implement the lane centerline extraction method according to any one of claims 1 to 8; and / or, The computer program product comprises a computer program, which, when executed by a processor, implements the lane centerline extraction method according to any one of claims 1 to 8.

Citation Information

Cited By

  • Intersection identification method, apparatus and device, and storage medium

    CN116563708A

  • Method and device for optimizing high-precision map data in flying area and electronic equipment

    CN119884100A

  • City and county traffic data analysis method, device and equipment and storage medium

    CN122637597A