Methods and devices for determining scene types, storage media, and electronic devices
By creating virtual lane lines and using density clustering technology, the problem of road scene recognition in crowdsourced high-precision maps was solved, achieving accurate recognition and efficient data fusion, and improving the accuracy and coverage of map data.
Patent Information
- Application Number
- CN202410262732.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-07
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2044-03-07
AI Technical Summary
In the process of creating crowdsourced high-precision maps, road scene recognition is difficult, and existing technologies have not been able to effectively solve this problem.
Virtual lane lines are created based on the first trajectory data of the road scene to be identified, a scene recognition area is generated, and the scene type of the road scene is determined by density clustering. Accurate identification is then performed using related information such as entry point, exit point and azimuth angle.
It achieves accurate identification of road scenes, improves the data fusion accuracy and efficiency of crowdsourced high-precision maps, and ensures the accuracy and comprehensiveness of map information.
Smart Images

Figure CN118097617B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of high-precision map technology, and more specifically, to a method and apparatus for determining scene type, a storage medium, and an electronic device. Background Technology
[0002] High-precision maps are a crucial component of modern technological development, with broad application prospects in fields such as autonomous driving, smart cities, logistics, and the Internet of Things. High-precision maps include traditional high-precision maps and crowdsourced high-precision maps. Traditional high-precision maps require significant investment of human, material, and financial resources, while crowdsourced high-precision maps can achieve resource sharing through a sharing economy model, reducing the costs of map creation, updating, and maintenance. Crowdsourced high-precision maps can correct errors and mistakes in real time based on user feedback, continuously improving and updating map data, resulting in more accurate and comprehensive map information. In the process of automated high-precision map creation using crowdsourced data, road scene recognition is crucial; however, real-world road scenes are complex and diverse, making road scene recognition challenging.
[0003] There is still no effective solution to the problem of difficulty in road scene recognition during the construction of crowdsourced high-precision maps in related technologies. Summary of the Invention
[0004] This application provides a method and apparatus for determining scene types, a storage medium, and an electronic device to at least solve the problems of difficulty in road scene recognition during the mapping process of crowdsourced high-precision maps in related technologies.
[0005] According to one aspect of the embodiments of this application, a method for determining a scene type is provided, comprising: creating a virtual lane line for the road scene based on first trajectory data of the road scene to be identified; generating a scene identification region corresponding to the road scene through the virtual lane line, wherein the scene identification region includes the virtual lane line; determining a plurality of second trajectory data located in the scene identification region from the first trajectory data, and determining association information of the plurality of second trajectory data, wherein the association information includes: an entry point of each second trajectory data, an exit point of each second trajectory data, and a first azimuth angle of the entry point and the exit point; and determining the scene type of the road scene based on the association information.
[0006] In an exemplary embodiment, creating a virtual lane line for a road scene based on first trajectory data of the road scene to be identified includes: determining whether a lane line within the road scene is identified; if the lane line is not identified, moving a first data point a target distance along a first direction on the road plane where the first trajectory data is located to obtain a second data point, wherein the first trajectory data includes a plurality of first data points, the first direction is a direction perpendicular to a second direction, and the second direction is the direction indicated by the line connecting a third data point and a fourth data point adjacent to the first data point in the first trajectory data; and determining the trajectory data composed of the plurality of second data points as a virtual lane line.
[0007] In an exemplary embodiment, generating a scene recognition region corresponding to the road scene using the virtual lane lines includes: performing density clustering on the virtual lane lines according to a first clustering condition to obtain a first cluster set, wherein the first clustering condition includes a first distance and a first cluster number, and the cluster set includes multiple data points in the virtual lane lines that satisfy the first clustering condition; calculating the minimum bounding region corresponding to the first cluster set using a preset algorithm, and determining the minimum bounding region as the scene recognition region.
[0008] In an exemplary embodiment, determining the scene type of the road scene based on the associated information includes: performing density clustering on the entry points according to a second clustering condition to obtain a second cluster set, and determining multiple fusion entry points corresponding to the entry points based on the second cluster set, wherein the second clustering condition includes: a second distance, a second number of clusters, and a constraint azimuth angle; performing density clustering on the exit points according to the second clustering condition to obtain a third cluster set, and determining multiple fusion exit points corresponding to the exit points based on the third cluster set; determining a second azimuth angle between the fusion entry point and the fusion exit point; and determining the scene type of the road scene based on the second azimuth angle.
[0009] In an exemplary embodiment, determining the scene type of the road scene based on the second azimuth angle includes: determining a first number of fusion entry points and a second number of fusion exit points; and determining the azimuth angle range corresponding to the second azimuth angle; determining the scene type corresponding to the road scene from multiple scene types based on the first number, the second number, and the azimuth angle range, wherein the type indication information of the multiple scene types is different, and the type indication information includes at least one of the following: the first number, the second number, and the azimuth angle range.
[0010] In an exemplary embodiment, determining the scene type corresponding to the road scene from multiple scene types based on the first quantity, the second quantity, and the azimuth range includes: determining the scene type corresponding to the road scene as an intersection scene when a first condition is met, wherein the first condition includes: both the first quantity and the second quantity are equal to a first value; determining the scene type corresponding to the road scene as a T-junction scene when a second condition is met, wherein the second condition includes one of the following: both the first quantity and the second quantity are less than the first value, and both the first quantity and the second quantity are greater than the second value; one of the first quantity and the second quantity is equal to the second value, and the other quantity is equal to the third value; both the first quantity and the second quantity are equal to the second value, and the multiple second azimuth angles formed between the fusion entry point of the first quantity and the fusion exit point of the second quantity all belong to the first azimuth range. Azimuth range; one of the first quantity and the second quantity is equal to the second value, and the other quantity is equal to the fourth value, and the largest azimuth among the plurality of second azimuths formed between the first quantity of fusion entry points and the second quantity of fusion exit points belongs to the second azimuth range; under the condition of satisfying the third condition, the scene type corresponding to the road scene is determined to be a U-turn intersection scene, wherein the third condition includes one of the following: one of the first quantity and the second quantity is equal to the second value, and the other quantity is equal to the fourth value, and the largest azimuth belongs to the third azimuth range; both the first quantity and the second quantity are equal to the second value, and at least one of the plurality of second azimuths formed between the first quantity of fusion entry points and the second quantity of fusion exit points belongs to the fourth azimuth range, and all other azimuths among the plurality of second azimuths except for the at least one second azimuth belong to the first azimuth range.
[0011] In an exemplary embodiment, determining the scene type corresponding to the road scene from multiple scene types based on the first quantity, the second quantity, and the azimuth range includes: determining the scene type corresponding to the road scene as a two-way turning scene when a fourth condition is met, wherein the fourth condition includes: both the first quantity and the second quantity are equal to a second value, and the azimuth difference between the multiple azimuths belongs to a fifth azimuth range; determining the scene type corresponding to the road scene as a merging scene when a fifth condition is met, wherein the fifth condition includes: the first quantity is equal to a second value, the second quantity is equal to a fourth value, and the largest azimuth among the multiple second azimuths formed between the first quantity of merging entry points and the second quantity of merging exit points belongs to a sixth azimuth range; and determining the road scene when a sixth condition is met. The scene type corresponding to the road scene is a traffic diversion scene. The sixth condition includes: the first quantity equals the fourth value, the second quantity equals the second value, and the largest azimuth angle among the multiple second azimuth angles formed by the first quantity of merging entry points and the second quantity of merging exit points belongs to the sixth azimuth angle range. If the seventh condition is met, the scene type corresponding to the road scene is determined to be a one-way turning scene. The seventh condition includes: both the first quantity and the second quantity are equal to the fourth value, and the azimuth angle differences between the multiple azimuth angles all belong to the fifth azimuth angle range. If the eighth condition is met, the scene type corresponding to the road scene is determined to be a scene to be confirmed. The eighth condition includes: both the first quantity and the second quantity are equal to the fourth value, and the azimuth angle differences between the multiple azimuth angles all belong to the first azimuth angle range.
[0012] In an exemplary embodiment, after determining the scene type of the road scene based on the associated information, the method further includes: determining the number M of driving lines covered by the scene recognition area, wherein the driving lines include one of the following: stop lines, pedestrian crossings, and lane lines, and M is an integer greater than 1; and constructing an inscribed graphical region of the scene recognition area; if M satisfies a preset number specified by the scene type, constructing a target scene area corresponding to the scene type within the inscribed graphical region using the M driving lines, wherein the target scene area is used to indicate the scene area that the scene type is allowed to cover in the road scene.
[0013] In an exemplary embodiment, after determining the scene type of the road scene based on the associated information, the method further includes: if the scene type is determined to be an intersection scene, determining the lane lines in the driving guide lines covered by the scene recognition area, and generating a projection image of the lane lines; analyzing the projection image using a preset segmentation model to determine whether a safety island exists within the scene recognition area, wherein the training set of the preset segmentation model includes: lane line projection images of multiple intersection scenes with safety islands; if the safety island exists within the scene recognition area, determining the intersection scene as an intersection scene with a right-turn scene.
[0014] According to another aspect of the embodiments of this application, a scene type determination device is also provided, comprising: a creation module, configured to create virtual lane lines for the road scene based on first trajectory data of the road scene to be identified; a generation module, configured to generate a scene recognition region corresponding to the road scene through the virtual lane lines, wherein the scene recognition region includes the virtual lane lines; a first determination module, configured to determine a plurality of second trajectory data located in the scene recognition region from the first trajectory data, and determine the association information of the plurality of second trajectory data, wherein the association information includes: an entry point of each second trajectory data, an exit point of each second trajectory data, and a first azimuth angle of the entry point and the exit point; and a second determination module, configured to determine the scene type of the road scene based on the association information.
[0015] According to another aspect of the embodiments of this application, a computer-readable storage medium is also provided, characterized in that the computer-readable storage medium includes a stored program, wherein the program executes the above-described method for determining the scenario type when it runs.
[0016] According to another aspect of the embodiments of this application, an electronic device is also provided, including a memory and a processor, characterized in that the memory stores a computer program, and the processor runs the above-described method for determining the scenario type through the computer program.
[0017] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements the method for determining the above-described scenario type.
[0018] In this embodiment, a virtual lane line is created for the road scene based on first trajectory data of the road scene to be identified; a scene recognition region corresponding to the road scene is generated through the virtual lane line, wherein the scene recognition region includes the virtual lane line; multiple second trajectory data located in the scene recognition region are determined from the first trajectory data, and the association information of the multiple second trajectory data is determined, wherein the association information includes: the entry point of each second trajectory data, the exit point of each second trajectory data, and the first azimuth angle of the entry point and the exit point; the scene type of the road scene is determined based on the association information. This solves the problem of difficulty in road scene recognition during the mapping process of crowdsourced high-precision maps in related technologies. It achieves the technical effect of accurate road scene recognition. Attached Figure Description
[0019] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:
[0020] Figure 1 This is a hardware structure block diagram of a computer terminal for a method of determining a scene type according to an embodiment of this application;
[0021] Figure 2 This is a flowchart of a method for determining a scenario type according to an embodiment of this application;
[0022] Figure 3 This is another flowchart of a method for determining a scene type according to an embodiment of this application;
[0023] Figure 4 This is a schematic diagram of constructing virtual lane lines and approximate scene range according to a method for determining scene type according to an embodiment of this application;
[0024] Figure 5 This is a schematic diagram of road scene type identification according to an embodiment of the present application for determining scene type;
[0025] Figure 6 This is a schematic diagram of a one-way turning road scene construction method according to an embodiment of this application;
[0026] Figure 7 This is a schematic diagram of a bidirectional turning road scene construction method according to an embodiment of this application;
[0027] Figure 8 This is a schematic diagram of a U-turn intersection road scene construction method according to an embodiment of this application;
[0028] Figure 9 This is a schematic diagram of a T-junction road scene construction according to a method for determining scene type according to an embodiment of this application;
[0029] Figure 10 This is a schematic diagram of a crossroads road scene construction method according to an embodiment of this application;
[0030] Figure 11 This is a schematic diagram of a crossroads road scene with a right-hand-leaning scene, according to a method for determining scene type according to an embodiment of this application.
[0031] Figure 12 This is a structural block diagram of a scene type determination device according to an embodiment of this application. Detailed Implementation
[0032] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.
[0033] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.
[0034] The methods and embodiments provided in this application can be run on a computer terminal. Taking running on a computer terminal as an example, Figure 1 This is a hardware structure block diagram of a computer terminal according to a method for determining a scene type based on an embodiment of this application. For example... Figure 1 As shown, a computer terminal may include one or more ( Figure 1Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. In one exemplary embodiment, the computer terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned computer terminal. For example, the camera device may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 Equivalent functions or ratios shown Figure 1 The functions shown have more different configurations.
[0035] The memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the scenario type determination method in this embodiment. The processor 102 runs various functional applications and data processing by executing the computer programs stored in the memory 104, thus implementing the above-described method. The memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage systems, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor 102, and these remote memories can be connected to a secure text network via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.
[0036] The transmission device 106 is used to receive or send data via a network. Specific examples of the network described above may include a wireless network provided by the camera equipment's communication provider. In one example, the transmission system 106 includes a Network Interface Controller (NIC), which can connect to other network devices via a base station to communicate with the Internet.
[0037] This embodiment provides a method for determining the scene type, including but not limited to methods applied to the aforementioned computer terminal. Figure 2 This is a flowchart of a method for determining a scenario type according to an embodiment of this application. The process includes the following steps:
[0038] Step S202: Create virtual lane lines for the road scene based on the first trajectory data of the road scene to be identified;
[0039] It should be noted that the first trajectory data can be crowdsourced trajectory data. Crowdsourced trajectory data is used to construct crowdsourced high-precision maps.
[0040] Crowdsourced high-precision maps, leveraging network technology, can update map data in real time, offering greater data real-time performance and accurately reflecting current traffic conditions, road construction, and other information. Crowdsourced high-precision maps encourage public participation in map data collection and updates; anyone can submit geographic information data via mobile devices, increasing the coverage and depth of map data while also improving its accuracy.
[0041] The process of creating high-precision maps using crowdsourced data mainly includes: task scheduling, vehicle-side perception, cloud-based mapping, map learning, automated mapping, map verification, map compilation, and map publishing. In the automated high-precision map creation process, road scene identification and construction can be performed by processing crowdsourced trajectory data according to different scenarios, improving the accuracy, efficiency, and reliability of data fusion in crowdsourced high-precision map learning.
[0042] Step S204: Generate a scene recognition region corresponding to the road scene using the virtual lane lines, wherein the scene recognition region includes the virtual lane lines;
[0043] Step S206: Determine multiple second trajectory data located in the scene recognition area from the first trajectory data, and determine the association information of the multiple second trajectory data, wherein the association information includes: the entry point of each second trajectory data, the exit point of each second trajectory data, and the first azimuth angle of the entry point and the exit point;
[0044] Step S208: Determine the scene type of the road scene based on the associated information.
[0045] Through the above steps, a virtual lane line is created for the road scene based on the first trajectory data of the road scene to be identified; a scene recognition region corresponding to the road scene is generated through the virtual lane line, wherein the scene recognition region includes the virtual lane line; multiple second trajectory data located in the scene recognition region are determined from the first trajectory data, and the association information of the multiple second trajectory data is determined, wherein the association information includes: the entry point of each second trajectory data, the exit point of each second trajectory data, and the first azimuth angle of the entry point and the exit point; the scene type of the road scene is determined based on the association information. This solves the problem of difficulty in road scene recognition during the mapping process of crowdsourced high-precision maps in related technologies. It achieves the technical effect of accurate road scene recognition.
[0046] Step S202 above can be implemented as follows: determining whether lane lines within the road scene are identified; if lane lines are not identified, moving the first data point a target distance along a first direction on the road plane where the first trajectory data is located to obtain a second data point, wherein the first trajectory data includes multiple first data points, the first direction is perpendicular to a second direction, and the second direction is indicated by the line connecting the third and fourth data points adjacent to the first data point in the first trajectory data; and determining the trajectory data formed by the multiple second data points as a virtual lane line. The multiple first data points include the third and fourth data points.
[0047] For a road scene, if lane lines are not identified, then for the first data point constituting the first trajectory data, the first data point is moved a target distance in a first direction perpendicular to the movement direction of the first trajectory (equivalent to the second direction) to obtain virtual lane lines. For example, if the second direction corresponding to the first data point points due north, then the first direction should simultaneously include due east and due west. That is, the first trajectory data is simultaneously shifted a target distance to both sides of the road to obtain virtual lane lines on both sides of the road.
[0048] Optionally, the second direction can also be: determining the arc formed by the first data point, the third data point, and the fourth data point, determining the tangent of the first data point on the arc, and determining the direction of the tangent as the second direction.
[0049] Optionally, for the generated virtual lane lines, determine the start and end directions of the virtual lane lines. If the angle between the start and end directions is less than 10°, delete the virtual lane line; otherwise, retain it.
[0050] For step S204 above, there are multiple possible solutions. One possible solution includes: performing density clustering on the virtual lane lines according to a first clustering condition to obtain a first cluster set, wherein the first clustering condition includes: a first distance and a first cluster number, and the cluster set includes: multiple data points in the virtual lane lines that satisfy the first clustering condition; calculating the minimum bounding region corresponding to the first cluster set through a preset algorithm, and determining the minimum bounding region as the scene recognition region.
[0051] That is, given the virtual lane lines included in the road scene, multiple data points in the virtual lane lines that satisfy the first clustering condition are obtained through density clustering, and then the minimum bounding area of the multiple data points is calculated through a preset algorithm to obtain the scene recognition area.
[0052] It should be noted that the first distance in the first clustering condition is generally chosen to be 30 meters, and the first cluster number is 10 points. The virtual lane lines generated within the road scene are generally multiple. The preset algorithm can be selected as the minimum circumcircle algorithm, thus calculating the minimum circumcircle region, which can be directly determined in the calculation process by its center and radius.
[0053] The implementation scheme of step S208 above includes: performing density clustering on the entry points according to the second clustering conditions to obtain a second cluster set, and determining multiple fusion entry points corresponding to the entry points based on the second cluster set, wherein the second clustering conditions include: a second distance, a second number of clusters, and a constraint azimuth angle; performing density clustering on the exit points according to the second clustering conditions to obtain a third cluster set, and determining multiple fusion exit points corresponding to the exit points based on the third cluster set; determining a second azimuth angle between the fusion entry point and the fusion exit point; and determining the scene type of the road scene based on the second azimuth angle.
[0054] Understandably, after determining multiple second trajectory data included in the scene recognition area within the first trajectory data, and determining the entry point, exit point, and first azimuth angle between the entry point and exit point corresponding to the multiple second trajectory data, density clustering based on the second clustering condition is performed on the multiple entry points and exit points respectively to obtain the fused entry point and fused exit point respectively; then, the scene type of the road scene can be determined by calculating the second azimuth angle between the fused entry point and fused exit point.
[0055] It should be noted that the second distance in the second clustering condition can be 30 meters, the second cluster number is 5 points, and the constraint azimuth angle is 90°. In this embodiment, generally, density clustering can be used to determine the entry point clusters and exit point clusters that satisfy the second clustering condition (this can be understood as the second cluster set including multiple entry point clusters and the third cluster set including multiple exit point clusters). Each entry point cluster can determine one merged entry point, and each exit point cluster can determine one merged exit point. The second azimuth angle is calculated to determine the azimuth angle between the merged entry point and the merged exit point. For example, if there are 2 merged entry points and 3 merged exit points, then it is necessary to calculate the second azimuth angle between the first merged entry point and each of the 3 merged exit points, and the second azimuth angle between the second merged entry point and each of the 3 merged exit points, for a total of 6 second azimuth angles.
[0056] In an exemplary embodiment, determining the scene type of the road scene based on the second azimuth angle includes: determining a first number of fusion entry points and a second number of fusion exit points; and determining the azimuth angle range corresponding to the second azimuth angle; determining the scene type corresponding to the road scene from multiple scene types based on the first number, the second number, and the azimuth angle range, wherein the type indication information of the multiple scene types is different, and the type indication information includes at least one of the following: the first number, the second number, and the azimuth angle range.
[0057] That is, the scene type corresponding to the road scene among multiple scene types can be determined based on the first number of fusion entry points, the second number of fusion exit points, and the range of azimuth angles corresponding to the second azimuth angle between the fusion entry points and the fusion exit points.
[0058] Specifically, the scene type corresponding to the road scene is determined from multiple scene types based on the first quantity, the second quantity, and the azimuth range, including:
[0059] If the first condition is met, the scene type corresponding to the road scene is determined to be an intersection scene, wherein the first condition includes: both the first quantity and the second quantity are equal to the first value;
[0060] For example, if both the first and second quantities are equal to 4, that is, the first value is 4, the scene type of the road scene can be determined as an intersection scene.
[0061] Under the condition that the second condition is met, the scene type corresponding to the road scene is determined to be a T-junction scene, wherein the second condition includes one of the following: both the first quantity and the second quantity are less than the first value, and both the first quantity and the second quantity are greater than the second value; one of the first quantity and the second quantity is equal to the second value, and the other quantity is equal to the third value; both the first quantity and the second quantity are equal to the second value, and the plurality of second azimuth angles formed between the first quantity of fusion entry points and the second quantity of fusion exit points all belong to the first azimuth angle range; one of the first quantity and the second quantity is equal to the second value, and the other quantity is equal to the fourth value, and the largest azimuth angle among the plurality of second azimuth angles formed between the first quantity of fusion entry points and the second quantity of fusion exit points belongs to the second azimuth angle range;
[0062] For example, if both the first and second quantities are less than 4 (first value) and greater than 2 (second value), the scene type corresponding to the road scene is determined to be a T-junction scene; if one of the first and second quantities is equal to 2 (second value) and the other is equal to 3 (equivalent to the third value), the scene type corresponding to the road scene is determined to be a T-junction scene; if both the first and second quantities are equal to 2 (second value), and the second azimuth angle between the two fusion entry points and the two fusion exit points is less than 30° (i.e., the first azimuth angle range is 0 to 30°), then the scene type corresponding to the road scene is determined to be a T-junction scene; if one of the first and second quantities is equal to 2 and the other is equal to 1 (fourth value), and the largest azimuth angle among the multiple second azimuth angles formed between the first quantity of fusion entry points and the second quantity of fusion exit points belongs to the second azimuth angle range of 60 to 120° (second azimuth angle range), then the scene type corresponding to the road scene is determined to be a T-junction scene.
[0063] Under the condition of satisfying the third condition, the scene type corresponding to the road scene is determined to be a U-turn intersection scene, wherein the third condition includes one of the following: one of the first quantity and the second quantity is equal to the second value, and the other quantity is equal to the fourth value, and the maximum azimuth angle belongs to the third azimuth angle range; both the first quantity and the second quantity are equal to the second value, and at least one of the multiple second azimuth angles formed between the fusion entry point of the first quantity and the fusion exit point of the second quantity belongs to the fourth azimuth angle range, and all other azimuth angles of the multiple second azimuth angles except for the at least one second azimuth angle belong to the first azimuth angle range.
[0064] For example: if one of the first quantity and the second quantity is equal to 2 and the other quantity is equal to 1, and the maximum azimuth angle between the fusion entry point of the first quantity and the fusion exit point of the second quantity is between 120° and 180° (third azimuth angle range), then the scene type corresponding to the road scene is determined to be a U-turn intersection scene; if both the first quantity and the second quantity are equal to 2, and at least one of the multiple second azimuth angles formed between the fusion entry point of the first quantity and the fusion exit point of the second quantity is between 150° and 180° (fourth azimuth angle range), and the remaining azimuth angles of the multiple second azimuth angles, except for the at least one second azimuth angle, are all between 0° and 30°, then the scene type corresponding to the road scene is determined to be a U-turn intersection scene.
[0065] Furthermore, if the fourth condition is met, the scene type corresponding to the road scene is determined to be a two-way turning scene. The fourth condition includes: the first quantity and the second quantity are both equal to the second value, and the azimuth angle difference between the multiple azimuth angles is within the range of the fifth azimuth angle. Optionally, the range of the fifth azimuth angle can be 30 to 180°.
[0066] If the fifth condition is met, the scene type corresponding to the road scene is determined to be a merging scene. The fifth condition includes: the first quantity is equal to the second value, the second quantity is equal to the fourth value, and the largest azimuth angle among the multiple second azimuth angles formed between the first quantity of merging entry points and the second quantity of merging exit points belongs to the sixth azimuth angle range; optionally, the sixth azimuth angle range is 0 to 60°.
[0067] Under the condition of the sixth condition, the scene type corresponding to the road scene is determined to be a diversion scene. The sixth condition includes: the first quantity is equal to the fourth value, the second quantity is equal to the second value, and the largest azimuth angle among the multiple second azimuth angles formed between the first quantity of the fusion entry points and the second quantity of the fusion exit points belongs to the range of the sixth azimuth angle.
[0068] Under the condition of the seventh condition, the scene type corresponding to the road scene is determined to be a one-way turning scene. The seventh condition includes: the first quantity and the second quantity are both equal to the fourth value, and the azimuth angle difference between the multiple azimuth angles is within the range of the fifth azimuth angle.
[0069] If the eighth condition is met, the scene type corresponding to the road scene is determined to be a scene to be confirmed. The eighth condition includes: the first quantity and the second quantity are both equal to the fourth value, and the azimuth angle difference between the multiple azimuth angles is within the range of the first azimuth angle.
[0070] It should be noted that in the above embodiments, the fourth value is less than the second value, the second value is less than the third value, and the third value is less than the first value; the azimuth range includes: the first azimuth range, the second azimuth range, the third azimuth range, the fourth azimuth range, the fifth azimuth range, and the sixth azimuth range;
[0071] Wherein, the second azimuth angle range, the third azimuth angle range, and the sixth azimuth angle range do not overlap, and the sum of the ranges of the second azimuth angle range, the third azimuth angle range, and the sixth azimuth angle range is the target angle; the first azimuth angle range and the fifth azimuth angle range do not overlap, and the sum of the ranges of the first azimuth angle range and the fifth azimuth angle range is the target angle; wherein, the fifth azimuth angle range includes the third azimuth angle range and the second azimuth angle range, the third azimuth angle range includes the fourth azimuth angle range, and the sixth azimuth angle range includes the first azimuth angle range.
[0072] In an exemplary embodiment, after determining the scene type of the road scene based on the associated information, a target scene area corresponding to the scene type can be constructed. The method further includes: determining the number M of driving lines covered by the scene recognition area, wherein the driving lines include one of the following: stop lines, pedestrian crossings, and lane lines, and M is an integer greater than 1; and constructing an inscribed graphic area of the scene recognition area; when M satisfies the preset number specified by the scene type, constructing the target scene area corresponding to the scene type in the inscribed graphic area using the M driving lines, wherein the target scene area is used to indicate the scene area that the scene type is allowed to cover in the road scene.
[0073] That is, the target scene area is constructed based on the number of driving signs covered within the scene recognition area. Optionally, the priority of stop lines, pedestrian crossings, and lane lines is as follows: stop lines have the highest priority, and lane lines have the lowest priority.
[0074] In an exemplary embodiment, after determining the scene type of the road scene based on the associated information, the method further includes: if the scene type is determined to be an intersection scene, determining the lane lines in the driving guide lines covered by the scene recognition area, and generating a projection image of the lane lines; analyzing the projection image using a preset segmentation model to determine whether a safety island exists within the scene recognition area, wherein the training set of the preset segmentation model includes: lane line projection images of multiple intersection scenes with safety islands; if the safety island exists within the scene recognition area, determining the intersection scene as an intersection scene with a right-turn scene.
[0075] Understandably, for intersection scenarios, since there are ordinary intersection scenarios and intersection scenarios with right-turn-prompt scenarios, it is possible to distinguish between these ordinary intersection scenarios and intersection scenarios with right-turn-prompt scenarios by training a deep learning model.
[0076] To better understand the above-mentioned scheme for determining the scenario type, in an optional embodiment, a scheme is also provided for explaining the above scheme.
[0077] With the continuous advancement of artificial intelligence and sensor technology, autonomous driving has become one of the hottest directions in the future transportation field. High-precision maps can provide accurate location information and road conditions for autonomous driving, better assisting vehicles in achieving autonomous navigation and obstacle avoidance. By updating and transmitting map data in real time, vehicles can accurately identify lane lines, traffic lights, traffic signs, etc., and make safe and efficient driving decisions based on map information. In addition, high-precision maps can also share information with other vehicles to achieve or assist in traffic flow optimization and congestion relief, as well as smart city construction. At the same time, high-precision maps can also analyze traffic congestion and optimize route planning, providing the best delivery solutions, reducing logistics costs and improving delivery efficiency.
[0078] Crowdsourced high-precision maps are a type of high-precision map. Utilizing network technology, they can update map data in real time, thus offering greater data timeliness and accurately reflecting current traffic conditions, road construction, and other information. Crowdsourced high-precision maps encourage public participation in data collection and updates; anyone can submit geographic information data via mobile devices, increasing the coverage and depth of map data while also improving its accuracy.
[0079] The process of creating high-precision maps using crowdsourced data mainly includes: task scheduling, vehicle-side perception, cloud-based mapping, map learning, automated mapping, map verification, map compilation, and map publishing. In the automated high-precision map creation process, road scene identification and construction can be performed by processing crowdsourced data according to different scenarios, improving the accuracy, efficiency, and reliability of data fusion for map learning.
[0080] In the process of automating high-precision map creation using crowdsourced data, road scene recognition is crucial; however, real-world road scenes are complex and diverse, making road scene recognition difficult. Furthermore, the complexity of road scenes also makes constructing the road scene model's extent challenging; it's difficult to construct the road extent for complex scenes using only a single constraint. In addition, in the process of road scene construction, there are special scenarios with safety islands, making it difficult to accurately construct the road scene extent using traditional methods.
[0081] To address the aforementioned issues, the purpose of this application's optional embodiments is to provide a method and system for road scene recognition and construction based on trajectory and lane lines for crowdsourced high-precision automated mapping. The overall technical flowchart is attached. Figure 3 As shown, the implementation steps include the following:
[0082] Step S31: Data cleaning, including the following steps S311 and S312.
[0083] Step S311: Trajectory data resampling:
[0084] Crowdsourced trajectory data contains significant data redundancy, necessitating resampling at intervals equal to a distance threshold DIS_LIMIT (0.5 meters). Following chronological order, all trajectory data is traversed, checking if the planar distance between adjacent trajectory data points is less than the threshold DIS_LIMIT. If it is, the trajectory point is deleted, and the search continues until the distance exceeds the threshold. The point is then retained, and the traversal continues, checking the distance between that point and the next point until the traversal is complete. (The resampled crowdsourced trajectory data is equivalent to the first trajectory data in the above embodiment).
[0085] Step S312: Lane line data denoising:
[0086] The 3D reconstruction of lane lines in the cloud contains noise, which interferes with scene construction based on lane lines. Therefore, it is necessary to denoise the lane line data. The main process involves determining whether the direction of the lane line exceeds the angle threshold ANGLE_LIMIT. If it does, the lane line data is deleted; otherwise, it is retained. (The denoised lane line data is equivalent to the driving indicator lines in the above embodiment).
[0087] Step S32: Create virtual lane line data.
[0088] Given that urban roads often lack lane markings at intersections (except for turning areas at major intersections), while lane markings frequently exist between intersections, the road scene is divided into intersection scenarios (including crossroads, T-junctions, U-turn intersections, etc.) and non-intersection scenarios. Where there are no lane markings, the trajectory data (equivalent to the first trajectory data) is offset 2 meters to the left and right (equivalent to the target distance in the above embodiment) with adjacent points as the direction, generating virtual lane markings, such as... Figure 4 As shown.
[0089] Determine the start and end directions of the virtual lane line. If they are less than the angle threshold ANGLE_LIMIT(10°), delete the virtual lane line; otherwise, retain it.
[0090] Step S33: Construct the approximate range of the intersection scene through virtual lane line clustering.
[0091] The virtual lane line data is clustered by density clustering with a distance of 30 meters and a cluster size of 10 points. The virtual lane line data in an intersection scene is added to a set (equivalent to the first cluster set in the above embodiment).
[0092] Step S34: Construct the approximate scope of the intersection scene.
[0093] By using the minimum circumcircle algorithm, the center and radius of the minimum circumcircle of the virtual lane line data within an intersection scene are calculated based on the virtual lane line data being added to a set, thereby constructing the approximate range of the intersection scene.
[0094] Step S35: Construct node data for the entry and exit locations of the intersection scene.
[0095] The starting and ending points of multiple trajectory data points across the intersection scene are subjected to density clustering with azimuth constraints. Clustering is performed according to a 30-meter distance, 5-point clusters, and a 90-degree angle constraint. The coordinates of each point set (equivalent to the multiple entry point clusters included in the second cluster set and the multiple exit point clusters included in the third cluster set in the above embodiment) and the azimuth are averaged to obtain the fused position and average azimuth of the entry and exit points at the intersection (equivalent to the second azimuth). Then, using the trajectory data within the intersection scene, the association relationship (i.e., the second azimuth) of all entry and exit points is established.
[0096] Step S36: Scene recognition.
[0097] like Figure 5 As shown, scene identification is performed based on the number and relationship of entry and exit points at intersections. The range between intersection scenes is identified as a general road scene.
[0098] If there is an entry point, an exit point, or both, and the azimuth difference between the entry and exit points is less than the angle threshold ANGLE_LIMIT (30°), the scene is identified as a scene to be confirmed. This is mainly due to the limited crowdsourced data for this scene. If there is an entry point and an exit point, and the azimuth difference is greater than the angle threshold ANGLE_LIMIT, then the scene is determined to be a one-way turning scene.
[0099] If there is one entering node and two exiting nodes, the maximum value of the azimuth angle between all entering and exiting nodes, MAX_ANGLE, needs to be calculated. If MAX_ANGLE is less than ANGLE_LIMIT1 (60°), the scene type is determined to be a traffic splitting scene; if MAX_ANGLE is greater than ANGLE_LIMIT1 (60°) and less than ANGLE_LIMIT2 (120°), the scene type is determined to be a T-junction scene; if MAX_ANGLE is greater than ANGLE_LIMIT1 (120°), the scene type is determined to be a U-turn intersection.
[0100] If there are two entering nodes and one exiting node, the maximum value of the azimuth angle between all entering and exiting nodes, MAX_ANGLE, needs to be calculated. If MAX_ANGLE is less than ANGLE_LIMIT1 (60°), the scene type is determined to be a merging scene; if MAX_ANGLE is greater than ANGLE_LIMIT1 (60°) and less than ANGLE_LIMIT2 (120°), the scene type is determined to be a T-junction scene; if MAX_ANGLE is greater than ANGLE_LIMIT1 (120°), the scene type is determined to be a U-turn intersection.
[0101] If there are two entering nodes and two exiting nodes, determine if the difference in azimuth angles between all associated entering and exiting nodes is greater than the angle threshold ANGLE_LIMIT(30°). If all are greater than the threshold, the scenario is classified as a two-way turn. If an angle is greater than the angle threshold ANGLE_LIMIT1(150°) and all other angles are less than the angle threshold ANGLE_LIMIT(30°), the scenario is classified as a U-turn. If all angles are less than the angle threshold ANGLE_LIMIT(30°), the scenario is classified as pending. All other cases are classified as T-junction scenarios.
[0102] If there are two entry nodes and three exit nodes, or three entry nodes and two exit nodes, then the scenario is determined to be a T-junction.
[0103] If there are three entry nodes and three exit nodes, the scenario is determined to be a T-junction.
[0104] If there are four entry nodes and four exit nodes, the scenario is determined to be an intersection.
[0105] The remaining scenarios are classified as other scenarios.
[0106] Optionally, considering that intersections are further divided into ordinary intersections and intersections with right-turn-prompt scenarios, it is necessary to further subdivide the intersection scene. Lane line data within a large number of intersections with safety islands are projected onto 512×512 images. A deep learning model is trained using manually labeled safety island ranges within the intersection to construct a safety island segmentation model. The segmentation model is then used to infer the presence of a safety island from the projected images of the intersection. If a safety island exists, it is classified as an intersection scene with a right-turn-prompt scenario; otherwise, it is classified as an ordinary intersection scene.
[0107] Step S37: Scene scope construction.
[0108] The scene range is constructed by using elements such as stop lines, pedestrian crossings, and lane lines reconstructed from perception to accurately define the scene range. The priority of element utilization is stop lines > pedestrian crossings > lane lines.
[0109] First, a stop-line spatial index is built to support spatial range queries.
[0110] Based on the circular scene range constructed in step 3, an inscribed square range is created. The direction of any pair of scene entry and exit points with an angle less than the angle threshold ANGLE_LIMIT (3°) is selected as the reference side direction of the inscribed square.
[0111] For general road scenarios, the original road scenario recognition range is maintained.
[0112] like Figure 6 , Figure 7 As shown, for both one-way and two-way turning scenarios, stop lines within the approximate scene range constructed by step 3 are obtained, and it is determined whether the number of stop lines is 2. If there are more than 2 stop lines, the original lane scene range is maintained; if there are 2 stop lines, the intersection point of the two stop lines is first calculated, and then, using the intersection point as the starting point and the other endpoint of the two stop lines as the direction point, the distance is calculated according to the radius of the approximate scene range. Calculate the other two scene range nodes by multiplying the result, and then calculate the last node according to the parallel line principle. Construct the final diamond-shaped scene range.
[0113] like Figure 8 As shown, for a U-turn scenario, the stop lines within the inscribed square of the scene are obtained, and it is determined whether the number of stop lines is 2. If there are not 2 stop lines, the original lane scene range is maintained; if there are 2 stop lines, the four intersection points of the lane lines and the inscribed square are calculated, and the lane scene range is reconstructed.
[0114] like Figure 9 As shown, for a T-junction scenario, the stop lines within the inscribed square area of the scene are obtained, and it is determined whether the number of stop lines is 3. If there are fewer than 3 stop lines, the original recognized road scene range is retained; if there are 3 stop lines, the intersection points between each pair of lane lines are found, and the distance between the three intersection points is calculated, with the shortest side being used as the reference side. The points on the other two stop lines that are furthest perpendicular to the reference side are found, and parallel to the reference side, straight lines are drawn to find the intersection points with the other two longer sides. The endpoints of the shorter sides and the intersection points of the straight lines with the longer sides are connected sequentially to construct the road scene range.
[0115] like Figure 10As shown, for a typical intersection scenario, the stop lines within the inscribed square area of the scene are obtained, and it is determined whether the number of stop lines is 4. If there are not 4, the original recognized road scene range is retained; if there are 4, the center points of the left and right stop lines are calculated, the azimuth angle between the center point and the starting point of each stop line is calculated, and the stop lines are sorted clockwise. The intersection points of adjacent stop lines are calculated in turn, and the intersection points are connected to construct the road scene range.
[0116] like Figure 11 As shown, for a crossroads scene with a right-hand drive, a safety island segmentation model is used to construct the triangular safety island range. Assuming the stop line is a straight line and the two shorter sides of the safety island are line segments, all intersections of the stop line and the safety island are calculated, and the intersections of these intersections with the two intersections of the safety island are used to construct the final road scene range.
[0117] Step S38: Through the above scheme, the final scene range was constructed, and the work of recognizing and constructing road scenes based on trajectories and lane lines was completed.
[0118] In summary, the optional embodiments of this application propose a new method for determining the approximate range of a road scene. The approximate range of the scene can be quickly and accurately constructed using virtual lane lines. The road scene is identified by using the number of entering and exiting nodes of the intersection scene and the azimuth angle of the matching nodes. Furthermore, a method for constructing a road scene based on fused stop lines is proposed, and a road scene with right-hand-leaning scenes is constructed using deep learning methods.
[0119] This enables accurate and efficient identification and construction of urban routine road scenes, and achieves automated processing, improving the accuracy and reliability of data fusion in map learning, thereby improving the efficiency and quality of crowdsourced high-precision map production.
[0120] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to run the methods of the various embodiments of this application.
[0121] This embodiment also provides a scene type determination device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.
[0122] Figure 12 This is a structural block diagram of a scene type determination device according to an embodiment of this application; as shown below. Figure 12 As shown, it includes:
[0123] A creation module 1202 is used to create virtual lane lines for the road scene based on the first trajectory data of the road scene to be identified.
[0124] The generation module 1204 is used to generate a scene recognition region corresponding to the road scene through the virtual lane lines, wherein the scene recognition region includes: the virtual lane lines;
[0125] The first determining module 1206 is used to determine multiple second trajectory data located in the scene recognition area from the first trajectory data, and to determine the association information of the multiple second trajectory data, wherein the association information includes: the entry point of each second trajectory data, the exit point of each second trajectory data, and the first azimuth angle of the entry point and the exit point;
[0126] The second determining module 1208 is used to determine the scene type of the road scene based on the associated information.
[0127] The apparatus described in this application creates virtual lane lines for a road scene based on first trajectory data of the road scene to be identified; generates a scene recognition region corresponding to the road scene using the virtual lane lines, wherein the scene recognition region includes the virtual lane lines; determines multiple second trajectory data located within the scene recognition region from the first trajectory data, and determines the association information of the multiple second trajectory data, wherein the association information includes: the entry point of each second trajectory data, the exit point of each second trajectory data, and the first azimuth angle of the entry point and the exit point; and determines the scene type of the road scene based on the association information. This solves the problem of difficulty in road scene recognition during the mapping process of crowdsourced high-precision maps in related technologies. It achieves the technical effect of accurate road scene recognition.
[0128] In an exemplary embodiment, the creation module 1202 is further configured to determine whether lane lines within the road scene are identified; if lane lines are not identified, the first data point is moved a target distance along a first direction on the road plane where the first trajectory data is located to obtain a second data point, wherein the first trajectory data includes multiple first data points, the first direction is a direction perpendicular to a second direction, and the second direction is the direction indicated by the line connecting the third and fourth data points adjacent to the first data point in the first trajectory data; the trajectory data composed of multiple second data points is determined as a virtual lane line.
[0129] In an exemplary embodiment, the generation module 1204 is further configured to perform density clustering on the virtual lane lines according to a first clustering condition to obtain a first cluster set, wherein the first clustering condition includes: a first distance and a first cluster number, and the cluster set includes: multiple data points in the virtual lane lines that satisfy the first clustering condition; calculate the minimum bounding region corresponding to the first cluster set through a preset algorithm, and determine the minimum bounding region as the scene recognition region.
[0130] In an exemplary embodiment, the second determining module 1208 is further configured to perform density clustering on the entry points according to a second clustering condition to obtain a second cluster set, and determine multiple fusion entry points corresponding to the entry points based on the second cluster set, wherein the second clustering condition includes: a second distance, a second number of clusters, and a constraint azimuth angle; and to perform density clustering on the exit points according to the second clustering condition to obtain a third cluster set, and determine multiple fusion exit points corresponding to the exit points based on the third cluster set; determine a second azimuth angle between the fusion entry point and the fusion exit point; and determine the scene type of the road scene based on the second azimuth angle.
[0131] In an exemplary embodiment, the second determining module 1208 is further configured to determine a first number of fusion entry points and a second number of fusion exit points; and to determine the azimuth range corresponding to the second azimuth; and to determine the scene type corresponding to the road scene from multiple scene types based on the first number, the second number, and the azimuth range, wherein the type indication information of the multiple scene types is different, and the type indication information includes at least one of the following: the first number, the second number, and the azimuth range.
[0132] In an exemplary embodiment, the second determining module 1208 is further configured to determine, under a first condition, that the scene type corresponding to the road scene is a crossroads scene, wherein the first condition includes: both the first quantity and the second quantity are equal to a first value; and under a second condition, determine that the scene type corresponding to the road scene is a T-junction scene, wherein the second condition includes one of the following: both the first quantity and the second quantity are less than the first value, and both the first quantity and the second quantity are greater than the second value; one of the first quantity and the second quantity is equal to the second value, and the other quantity is equal to a third value; both the first quantity and the second quantity are equal to the second value, and the plurality of second azimuth angles formed between the fusion entry points of the first quantity and the fusion exit points of the second quantity all belong to the range of the first azimuth angle; and among the first quantity and the second quantity... The maximum azimuth angle among the plurality of second azimuth angles formed between the first number of fusion entry points and the second number of fusion exit points belongs to the range of second azimuth angles. Under the condition of satisfying the third condition, the scene type corresponding to the road scene is determined to be a U-turn intersection scene. The third condition includes one of the following: one of the first number and the second number is equal to the second value, and the other number is equal to the fourth value, and the maximum azimuth angle belongs to the range of the third azimuth angle; both the first number and the second number are equal to the second value, and at least one of the plurality of second azimuth angles formed between the first number of fusion entry points and the second number of fusion exit points belongs to the range of the fourth azimuth angle, and all other azimuth angles among the plurality of second azimuth angles except for the at least one second azimuth angle belong to the range of the first azimuth angle.
[0133] In an exemplary embodiment, the second determining module 1208 is further configured to, when a fourth condition is met, determine that the scene type corresponding to the road scene is a two-way turning scene, wherein the fourth condition includes: both the first quantity and the second quantity are equal to the second value, and the azimuth angle difference between the plurality of azimuth angles belongs to the fifth azimuth angle range; when a fifth condition is met, determine that the scene type corresponding to the road scene is a merging scene, wherein the fifth condition includes: the first quantity is equal to the second value, the second quantity is equal to the fourth value, and the largest azimuth angle among the plurality of second azimuth angles formed between the first quantity of merging entry points and the second quantity of merging exit points belongs to the sixth azimuth angle range; when a sixth condition is met, determine that the scene type corresponding to the road scene is a diversion scene, wherein the... The sixth condition includes: the first quantity equals the fourth value, the second quantity equals the second value, and the largest azimuth angle among the plurality of second azimuth angles formed between the first quantity of fusion entry points and the second quantity of fusion exit points belongs to the sixth azimuth angle range; if the seventh condition is met, the scene type corresponding to the road scene is determined to be a one-way turning scene, wherein the seventh condition includes: both the first quantity and the second quantity are equal to the fourth value, and the azimuth angle difference between the plurality of azimuth angles belongs to the fifth azimuth angle range; if the eighth condition is met, the scene type corresponding to the road scene is determined to be a scene to be confirmed, wherein the eighth condition includes: both the first quantity and the second quantity are equal to the fourth value, and the azimuth angle difference between the plurality of azimuth angles belongs to the first azimuth angle range.
[0134] In one exemplary embodiment, the apparatus further includes a construction module for determining the number M of driving lines covered by the scene recognition region, wherein the driving lines include one of the following: stop lines, pedestrian crossings, and lane lines, and M is an integer greater than 1; and constructing an inscribed graphical region of the scene recognition region; and, if M satisfies a preset number specified by the scene type, constructing a target scene region corresponding to the scene type within the inscribed graphical region using the M driving lines, wherein the target scene region is used to indicate the scene region that the scene type is allowed to cover in the road scene.
[0135] In an exemplary embodiment, the apparatus further includes a third determining module, configured to, when the scene type is determined to be an intersection scene, determine the lane lines in the driving guide lines covered by the scene recognition area and generate a projection image of the lane lines; analyze the projection image using a preset segmentation model to determine whether a safety island exists within the scene recognition area, wherein the training set of the preset segmentation model includes: lane line projection images of multiple intersection scenes with safety islands; and, if the safety island exists within the scene recognition area, determine the intersection scene as an intersection scene with a right-turn scene.
[0136] Embodiments of this application also provide a storage medium including a stored program, wherein the program, when executed, performs any of the methods described above.
[0137] Optionally, in this embodiment, the storage medium may be configured to store program code for performing the following steps:
[0138] S1, Create virtual lane lines for the road scene based on the first trajectory data of the road scene to be identified;
[0139] S2, generate a scene recognition region corresponding to the road scene through the virtual lane lines, wherein the scene recognition region includes: the virtual lane lines;
[0140] S3, determine multiple second trajectory data located in the scene recognition area from the first trajectory data, and determine the association information of the multiple second trajectory data, wherein the association information includes: the entry point of each second trajectory data, the exit point of each second trajectory data, and the first azimuth angle of the entry point and the exit point;
[0141] S4, determine the scene type of the road scene based on the associated information.
[0142] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.
[0143] Specific examples in this embodiment can be found in the examples described in the above embodiments and exemplary implementations, and will not be repeated here.
[0144] Embodiments of this application also provide an electronic device including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.
[0145] Optionally, in this embodiment, the processor can be configured to perform the following steps via a computer program:
[0146] S1, Create virtual lane lines for the road scene based on the first trajectory data of the road scene to be identified;
[0147] S2, generate a scene recognition region corresponding to the road scene through the virtual lane lines, wherein the scene recognition region includes: the virtual lane lines;
[0148] S3, determine multiple second trajectory data located in the scene recognition area from the first trajectory data, and determine the association information of the multiple second trajectory data, wherein the association information includes: the entry point of each second trajectory data, the exit point of each second trajectory data, and the first azimuth angle of the entry point and the exit point;
[0149] S4, determine the scene type of the road scene based on the associated information.
[0150] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor and the input / output device is connected to the processor.
[0151] Embodiments of this application also provide a computer program product, including a computer program that, when executed by a processor, implements the steps of the methods described in various embodiments of this application.
[0152] Embodiments of this application also provide a computer program product, including a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the methods described in various embodiments of this application.
[0153] Optionally, specific examples in this embodiment can refer to the examples described in the above embodiments and optional implementations, and will not be repeated here.
[0154] Obviously, those skilled in the art should understand that the modules or steps of this application described above can be implemented using general-purpose computing systems. They can be centralized on a single computing system or distributed across a network of multiple computing systems. Optionally, they can be implemented using program code executable by a computing system, thereby storing them in a storage system for execution by the computing system. In some cases, the steps shown or described can be executed in a different order than presented here, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.
[0155] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the principles of this application should be included within the protection scope of this application.
Claims
1. A method for determining a scene type, characterized in that, include: Virtual lane lines are created for the road scene based on the first trajectory data of the road scene to be identified; The scene recognition region corresponding to the road scene is generated by the virtual lane lines, wherein the scene recognition region includes: the virtual lane lines; The process of generating a scene recognition region corresponding to the road scene using the virtual lane lines includes: performing density clustering on the virtual lane lines according to a first clustering condition to obtain a first cluster set, wherein the first clustering condition includes a first distance and a first cluster number, and the first cluster set includes multiple data points in the virtual lane lines that satisfy the first clustering condition; calculating the minimum bounding region corresponding to the first cluster set using a preset algorithm, and determining the minimum bounding region as the scene recognition region; Multiple second trajectory data located in the scene recognition area are determined from the first trajectory data, and the association information of the multiple second trajectory data is determined, wherein the association information includes: the entry point of each second trajectory data, the exit point of each second trajectory data, and the first azimuth angle of the entry point and the exit point; The scene type of the road scene is determined based on the associated information; The process of determining the scene type of the road scene based on the associated information includes: Density clustering is performed on the entry points according to the second clustering condition to obtain a second cluster set. Based on the second cluster set, multiple fusion entry points corresponding to the entry points are determined. The second clustering condition includes: a second distance, a second number of clusters, and a constraint azimuth angle; and Density clustering is performed on the departure points according to the second clustering condition to obtain a third cluster set, and multiple fusion departure points corresponding to the departure points are determined according to the third cluster set. Determine the second azimuth angle between the fusion entry point and the fusion exit point; The scene type of the road scene is determined based on the second azimuth angle.
2. The method for determining the scene type according to claim 1, characterized in that, Based on the first trajectory data of the road scene to be identified, virtual lane lines are created for the road scene, including: Determine whether lane lines within the road scene have been identified; If the lane line is not identified, the first data point is moved a target distance in the first direction on the road plane where the first trajectory data is located to obtain the second data point. The first trajectory data includes multiple first data points. The first direction is a direction perpendicular to the second direction. The second direction is the direction indicated by the line connecting the third and fourth data points adjacent to the first data point in the first trajectory data. The trajectory data composed of multiple second data points is determined as virtual lane lines.
3. The method for determining the scene type according to claim 1, characterized in that, Determining the scene type of the road scene based on the second azimuth angle includes: Determine the first number of fusion entry points and the second number of fusion exit points; and Determine the azimuth range corresponding to the second azimuth; The scene type corresponding to the road scene is determined from multiple scene types based on the first quantity, the second quantity, and the azimuth range, wherein the type indication information of the multiple scene types is different, and the type indication information includes at least one of the following: the first quantity, the second quantity, and the azimuth range.
4. The method for determining the scene type according to claim 3, characterized in that, The scene type corresponding to the road scene is determined from multiple scene types based on the first quantity, the second quantity, and the azimuth range, including: If the first condition is met, the scene type corresponding to the road scene is determined to be an intersection scene, wherein the first condition includes: both the first quantity and the second quantity are equal to the first value; Under the condition that the second condition is met, the scene type corresponding to the road scene is determined to be a T-junction scene, wherein the second condition includes one of the following: both the first quantity and the second quantity are less than the first value, and both the first quantity and the second quantity are greater than the second value; one of the first quantity and the second quantity is equal to the second value, and the other quantity is equal to the third value; both the first quantity and the second quantity are equal to the second value, and the plurality of second azimuth angles formed between the first quantity of fusion entry points and the second quantity of fusion exit points all belong to the first azimuth angle range; one of the first quantity and the second quantity is equal to the second value, and the other quantity is equal to the fourth value, and the largest azimuth angle among the plurality of second azimuth angles formed between the first quantity of fusion entry points and the second quantity of fusion exit points belongs to the second azimuth angle range; Under the condition of satisfying the third condition, the scene type corresponding to the road scene is determined to be a U-turn intersection scene, wherein the third condition includes one of the following: one of the first quantity and the second quantity is equal to the second value, and the other quantity is equal to the fourth value, and the maximum azimuth angle belongs to the third azimuth angle range; both the first quantity and the second quantity are equal to the second value, and at least one of the multiple second azimuth angles formed between the fusion entry point of the first quantity and the fusion exit point of the second quantity belongs to the fourth azimuth angle range, and all other azimuth angles of the multiple second azimuth angles except for the at least one second azimuth angle belong to the first azimuth angle range.
5. The method for determining the scene type according to claim 3, characterized in that, The scene type corresponding to the road scene is determined from multiple scene types based on the first quantity, the second quantity, and the azimuth range, including: Under the condition of satisfying the fourth condition, the scene type corresponding to the road scene is determined to be a two-way turning scene, wherein the fourth condition includes: the first quantity and the second quantity are both equal to the second value, and the azimuth angle difference between the multiple azimuth angles is within the range of the fifth azimuth angle; If the fifth condition is met, the scene type corresponding to the road scene is determined to be a merging scene. The fifth condition includes: the first quantity is equal to the second value, the second quantity is equal to the fourth value, and the largest azimuth angle among the multiple second azimuth angles formed between the first quantity of merging entry points and the second quantity of merging exit points belongs to the range of the sixth azimuth angle. Under the condition of the sixth condition, the scene type corresponding to the road scene is determined to be a diversion scene. The sixth condition includes: the first quantity is equal to the fourth value, the second quantity is equal to the second value, and the largest azimuth angle among the multiple second azimuth angles formed between the first quantity of the fusion entry points and the second quantity of the fusion exit points belongs to the range of the sixth azimuth angle. Under the condition of the seventh condition, the scene type corresponding to the road scene is determined to be a one-way turning scene. The seventh condition includes: the first quantity and the second quantity are both equal to the fourth value, and the azimuth angle difference between the multiple azimuth angles is within the range of the fifth azimuth angle. If the eighth condition is met, the scene type corresponding to the road scene is determined to be a scene to be confirmed. The eighth condition includes: the first quantity and the second quantity are both equal to the fourth value, and the azimuth angle difference between the multiple azimuth angles is within the range of the first azimuth angle.
6. The method for determining scene type according to claim 1, characterized in that, After determining the scene type of the road scene based on the associated information, the method further includes: Determine the number M of driving guidelines covered by the scene recognition area, wherein the driving guidelines include one of the following: stop lines, pedestrian crossings, and lane lines, and M is an integer greater than 1; and Construct the inscribed graphical region of the scene recognition region; When M meets the preset number specified by the scene type, the target scene area corresponding to the scene type is constructed in the inscribed graphic area by M driving indicator lines, wherein the target scene area is used to indicate the scene area that the scene type is allowed to cover in the road scene.
7. The method for determining scene type according to claim 1, characterized in that, After determining the scene type of the road scene based on the associated information, the method further includes: If the scene type is determined to be an intersection scene, the lane lines in the driving guide lines covered by the scene recognition area are determined, and the projected image of the lane lines is generated. The projected images are analyzed using a preset segmentation model to determine whether a safety island exists within the scene recognition area. The training set of the preset segmentation model includes lane line projected images of multiple intersection scenes with safety islands. If the safety island exists within the scene recognition area, the intersection scene is determined to be an intersection scene with a right-hand drive.
8. A device for determining a scene type, characterized in that, include: A module is created to generate virtual lane lines for the road scene based on the first trajectory data of the road scene to be identified. A generation module is used to generate a scene recognition region corresponding to the road scene through the virtual lane lines, wherein the scene recognition region includes: the virtual lane lines; The generation module is further configured to perform density clustering on the virtual lane lines according to a first clustering condition to obtain a first cluster set, wherein the first clustering condition includes a first distance and a first cluster number, and the first cluster set includes multiple data points in the virtual lane lines that satisfy the first clustering condition; calculate the minimum bounding region corresponding to the first cluster set through a preset algorithm, and determine the minimum bounding region as the scene recognition region; The first determining module is used to determine multiple second trajectory data located in the scene recognition area from the first trajectory data, and to determine the association information of the multiple second trajectory data, wherein the association information includes: the entry point of each second trajectory data, the exit point of each second trajectory data, and the first azimuth angle of the entry point and the exit point; The second determining module is used to determine the scene type of the road scene based on the associated information; The second determining module is further configured to perform density clustering on the entry points according to the second clustering conditions to obtain a second cluster set, and determine multiple fusion entry points corresponding to the entry points based on the second cluster set. The second clustering conditions include: a second distance, a second number of clusters, and a constraint azimuth angle. Density clustering is performed on the departure points according to the second clustering condition to obtain a third cluster set, and multiple fusion departure points corresponding to the departure points are determined according to the third cluster set. Determine the second azimuth angle between the fusion entry point and the fusion exit point; The scene type of the road scene is determined based on the second azimuth angle.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein the program, when executed, performs the method described in any one of claims 1 to 7.
10. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and the processor is configured to run the method described in any one of claims 1 to 7 through the computer program.
Citation Information
Patent Citations
Personalized guide line extraction method and system based on a crowdsourcing trajectory
CN114022627A
Road traffic congestion level evaluation method and device, electronic equipment and storage medium
CN116128360A
