Parallel road generation method, system and terminal
By using kernel density estimation and skeletonization techniques based on bus trajectory data, the accuracy problem of parallel road generation in existing technologies has been solved, achieving higher information accuracy and location accuracy, and making it suitable for parallel road generation in intelligent transportation systems.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- SHENZHEN INST OF ADVANCED TECH CHINESE ACAD OF SCI
- Filing Date
- 2023-04-24
- Publication Date
- 2026-06-12
AI Technical Summary
Existing technologies struggle to accurately distinguish and merge parallel road segments when generating parallel roads under GPS noise, leading to redundant edges and navigation errors. This is especially problematic in complex situations such as viaducts and underpasses, where existing methods cannot effectively filter out GPS drift, affecting the accuracy of road generation information and location.
Kernel density estimation is performed using bus trajectory data. The trajectory groups are divided using the K-modes algorithm, and automatic route-finding routes are generated using the Google Directions API. Road centerlines are generated by combining kernel density estimation and skeletonization techniques to filter GPS noise and improve the accuracy of parallel road generation.
By conducting a comprehensive analysis of bus routes, urban GPS drift can be effectively filtered out, improving the accuracy of road generation information and location, reducing false road sections, and enhancing the accuracy of the navigation system.
Smart Images
Figure CN116805463B_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of intelligent transportation technology, and in particular to a method, system and terminal for generating parallel roads. Background Technology
[0002] In recent years, the vast amounts of GPS trajectory data collected by floating cars have been widely applied to road network inference. This is primarily based on two characteristics of trajectories: vehicle movement is constrained by the road network, and the overlay of numerous vehicle trajectories can outline the structure of the road network. Vehicle trajectory data is highly sensitive to changes in real-time road connectivity, meeting the needs of real-time electronic map updates. These research schemes mostly follow a similar pattern, with steps including map matching (aligning trajectories with the map), map inference (inferring new roads from unmatched trajectories), and map merging (merging new roads with the original map). Current map matching algorithms have achieved good results. Therefore, the effectiveness of map updates largely depends on the effectiveness of map inference. Thus, map inference remains a core problem.
[0003] Current parallel road generation methods based on trajectory data can be divided into three categories: (1) Point clustering method, which clusters the original GPS trajectory points according to the geospatial distance and directional similarity of GPS sample points, and adds edges between the center points of consecutive clusters in each trajectory. It regards map inference as a multi-network alignment problem. (2) Kernel density estimation (KDE) method, which divides the road network map into many cells, statistically weights the number of GPS trajectories passing through each cell, and generates a spatial histogram. The histogram is then kernel smoothed (usually using Gaussian distribution), and then morphological thinning or similar skeletonization techniques are used to extract the center line and generate the map. (3) Trajectory merging method utilizes the continuity of the trajectory, inserting GPS trajectory data into the empty road network in sequence to generate the map. However, the existing technology has the following defects:
[0004] Due to GPS noise, GPS points will exhibit a certain degree of offset. When the distance between two parallel roads is small, a large number of GPS points will shift to the middle region of the two parallel road segments. Point cloud clustering methods primarily address GPS noise in parallel road segments by altering the clustering radius parameter. However, in complex situations where two parallel road segments are close together (such as elevated bridges or underpasses parallel to surface roads), clustering methods fail to distinguish between parallel roads, easily connecting non-intersecting roads like elevated bridges and underpasses, and failing to capture detailed topological structures. Similarly, under GPS noise interference, trajectory-based methods generate redundant edges when generating parallel roads; the closer the distance between parallel roads, the more severe the generation of false road segments. These problems pose significant challenges to current road generation methods when handling parallel road segments under GPS noise.
[0005] The GPS trajectory sampling rate is a crucial factor affecting the accuracy of trajectory representation. For example, when a vehicle travels at a speed less than 50 km / h and samples every 30 seconds, the maximum distance between two consecutive samples, p1 and p2, can reach 417 meters. Furthermore, due to sample loss caused by GPS signal loss, the sampling rate may vary along the trajectory, further degrading trajectory quality. This is more pronounced in trajectory-based methods, where lines can no longer represent the shape of the road when the sampling rate is very low. On the other hand, trajectory-based methods require merging current roads with previous paths when generating maps; however, this merging process is challenging: merging connected roads while ensuring that parallel roads such as viaducts, underpasses, and multi-level roads remain separate. Accurately capturing connections is critical, as even a few incorrect connections can lead to numerous navigation errors. Existing point cloud-based clustering and trajectory-based methods consider only information from a single GPS point, such as latitude, longitude, and heading, when merging road segments, resulting in many incorrect connections in challenging areas such as parallel road segments. Summary of the Invention
[0006] Therefore, it is necessary to provide a parallel road generation method, system, and terminal that can improve the information accuracy and location accuracy of road generation, addressing the shortcomings of existing technologies.
[0007] To solve the above problems, this application adopts the following technical solution:
[0008] One objective of this application is to provide a method for generating parallel roads, comprising the following steps:
[0009] Obtain bus trajectory data;
[0010] Kernel density estimation is performed based on the bus trajectory;
[0011] The road centerline is generated based on the kernel density.
[0012] In some embodiments, the step of acquiring bus trajectory data specifically includes:
[0013] Based on the route characteristics of the input bus GPS data, the trajectories of multiple buses on the same route are extracted and regarded as a group of trajectories;
[0014] The bus routes are divided into different groups;
[0015] Count the number of trajectories in each group, and select the group with the most trajectories to represent the trajectory of that bus route;
[0016] Generate automatic pathfinding routes between adjacent trajectories;
[0017] By merging trajectories with the same pattern in this group, we can obtain global bus trajectory data.
[0018] In some embodiments, in the step of extracting the trajectories of multiple buses on the same route based on the route characteristics of the input bus GPS data, the GPS data of each bus includes seven fields: ID, Lon, Lat, Time, Speed, Direction, and Bus Line; where ID is a unique identifier for each bus; Lon is the longitude of the GPS point in the geographic coordinate system; Lat is the latitude of the GPS point in the geographic coordinate system; Time is the time when the location point was received; Speed is the instantaneous speed of the location point; Direction is the direction of travel of the location point; and Bus Line is the route identifier for each bus.
[0019] In some embodiments, the step of dividing the bus trajectory into different groups specifically includes the following step: using the K-modes algorithm to divide the bus trajectory into different groups.
[0020] In some embodiments, the step of generating an automatic pathfinding route between adjacent trajectories specifically includes the following steps:
[0021] The automatic route finding module of the Google Directions API is called to generate automatic route finding lines between adjacent tracks.
[0022] In some embodiments, the step of performing kernel density estimation based on the bus trajectory specifically includes the following steps:
[0023] Divide the area to be generated into 1x1 meter cells;
[0024] Set the pixel of the cell through which the trajectory passes to 1, and set the pixel of the cell without the trajectory to 0, to obtain a binary image in which each pixel has only two possible values.
[0025] Count the number of trajectories in each cell to generate a two-dimensional histogram;
[0026] The two-dimensional histogram is compared with the normal distribution function N(0, σ). 2 Convolution is performed to represent the expected GPS error distribution, and the choice of σ should be based on the expected GPS error and road width.
[0027] In some embodiments, the step of generating the road centerline based on the kernel density specifically includes the following steps:
[0028] Let D(x, y) be the density at position (x, y), for each density layer l ∈ l max Let T l It is a binary image such that when D(x, y) is greater than l, the following holds true:
[0029] T l (x,y)=1
[0030] The algorithm recursively generates the skeleton image S of level l. l :
[0031] S l =skeletonize(T l +S l+1 )
[0032] and:
[0033] S lmax =skeletonize(T lmax )
[0034] This process involves repeatedly refining the image until it converges.
[0035] A second objective of this application is to provide the aforementioned parallel road generation system, comprising:
[0036] The data acquisition unit is used to acquire bus trajectory data;
[0037] An estimation unit is used to perform kernel density estimation based on the bus trajectory;
[0038] A road generation unit is used to generate a road centerline based on the kernel density.
[0039] A third objective of this application is to provide a terminal, the terminal including a processor and a memory coupled to the processor, wherein...
[0040] The memory stores program instructions for implementing the parallel road generation method.
[0041] The processor is used to execute the program instructions stored in the memory to control the generation of parallel roads.
[0042] The present application adopts the above technical solution, and its beneficial effects are as follows:
[0043] The parallel road generation method, system, and terminal provided in this application acquire bus trajectories, perform kernel density estimation based on the bus trajectories, and generate road centerlines based on the kernel density. Because the parallel road generation method, system, and terminal provided in this application utilize the global nature of bus trajectories, the parallel roads with bus routes generated by this method can filter out most urban GPS drift, thereby improving the information accuracy and location accuracy of road generation. Attached Figure Description
[0044] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0045] Figure 1 The flowchart illustrates the steps of the parallel road generation method provided in Embodiment 1 of this application.
[0046] Figure 2 The flowchart of the steps for obtaining bus trajectory provided in Embodiment 1 of this application is shown.
[0047] Figure 3 The flowchart of the steps for obtaining bus trajectory provided in Embodiment 1 of this application is shown.
[0048] Figure 4 This is a flowchart illustrating the steps of kernel density estimation based on the bus trajectory provided in Embodiment 2 of this application.
[0049] Figure 5 This is a schematic diagram of the terminal structure provided in Embodiment 3 of this application. Detailed Implementation
[0050] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.
[0051] In the description of this application, it should be understood that the terms "upper", "lower", "horizontal", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.
[0052] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.
[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments.
[0054] Example 1
[0055] Please combine Figure 1 The following is a flowchart of a parallel road generation method provided in Embodiment 1, including steps S110 to S130. The implementation of each step is described in detail below.
[0056] Step S110: Obtain bus trajectory data.
[0057] Please see Figure 2 The steps for obtaining the bus trajectory include steps S111 to S115, and the implementation of each step is explained in detail below.
[0058] Step S111: Based on the route characteristics of the input bus GPS data, extract the trajectories of multiple buses on the same route as a group of trajectories.
[0059] Specifically, the trajectory data used is collected from Shenzhen buses equipped with GPS devices.
[0060] Each bus GPS data point contains seven fields: ID, Lon, Lat, Time, Speed, Direction, and BusLine. The first six fields are in the same format as taxi data, while the seventh field is the bus route identifier. The GPS data covers an area of approximately 50km x 25km in Shenzhen.
[0061] Step S112: Divide the bus trajectory into different groups.
[0062] In this embodiment, the K-modes algorithm is used to divide the bus trajectory into different groups.
[0063] It is understandable that bus drivers' individual driving habits and road conditions vary, leading to potential differences in bus trajectory patterns even on the same route. To address this issue, this embodiment employs the K-modes algorithm to divide trajectory routes into different groups. Since the K-modes algorithm is an extension of the K-means algorithm, it uses pattern clustering instead of mean clustering and employs a frequency-based method to update the patterns within the clusters, thereby minimizing the clustering cost function.
[0064] Step S113: Count the number of trajectories in each group, and select the group containing the most trajectories to represent the trajectories of the bus route.
[0065] Step S114: Generate an automatic pathfinding route between adjacent trajectories.
[0066] Specifically, the step of generating an automatic pathfinding route between adjacent trajectories includes the following steps:
[0067] The automatic route finding module of the Google Directions API is called to generate automatic route finding lines between adjacent tracks.
[0068] Step S115: Merge trajectories with the same pattern in the group to obtain global bus trajectory data.
[0069] Specifically, some of the technical terms used in this application are explained as follows:
[0070] Latitude and longitude coordinates: The latitude and longitude coordinates of a geographical location on the Earth's surface are coordinates based on the WGS84 geographic coordinate system, denoted as pi = (lon, lat).
[0071] Projected coordinate points: According to certain mathematical rules, the latitude and longitude coordinate points on the Earth's ellipsoid are transformed onto a plane, so that the geographic coordinates (lon, lat) of the ground point are projected onto the plane rectangular coordinates (x, y) of the corresponding point on the map through a functional relationship, realizing the scientific transformation from the Earth's ellipsoid to the map plane.
[0072] Trajectory: A trajectory tr is a spatial sequence of points, denoted as tr: p1→p2→…→pn. Each trajectory tr is sampled from a continuously moving object. Each point i is represented by a two-dimensional coordinate (x, y). i y i )∈R 2 It consists of a time stamp (ti). The points in the trajectory (tr) are arranged in chronological order, and every two consecutive points connected form a trajectory segment.
[0073] Connections: Connections are atomic parts of a road, such as a section between two intersections or a ramp on a highway. Connections can reflect the connectivity and topology between roads.
[0074] Road network: Composed of basic parts commonly referred to as connections. A road network is a directed graph, denoted as G = (V, E), where vertex v = (x, y) ∈ V represents a crossroads or the end of a road, and edge e = (u, v, l) ∈ E is a vertex tuple representing the connectivity from vertex u to vertex v.
[0075] Map matching: Map matching is the process of inferring the road network map sequence of edges e1→e2→…→en for a given trajectory Tr based on a given road network.
[0076] Problem Definition: Given a set of trajectories T = {tr1, tr2, ..., tr...} n}, generate its road network map G.
[0077] Step S120: Perform kernel density estimation based on the bus trajectory.
[0078] Please see Figure 3 The flowchart for kernel density estimation based on the bus trajectory includes steps S121 to S124, and the implementation of each step is described in detail below.
[0079] Step S121: Divide the road area to be generated into 1x1 meter cells.
[0080] Step S122: Set the pixel of the cell with the trajectory to 1 and the pixel of the cell without the trajectory to 0, so that each pixel has only two possible values.
[0081] Step S123: Count the number of trajectories in each cell to generate a two-dimensional histogram.
[0082] Step S124: Compare the two-dimensional histogram with the normal distribution function N(0, σ). 2 A convolution is performed to represent the expected GPS error distribution. The choice of σ should be based on the expected GPS error and the road width. In this invention, σ is set to 8.5 meters.
[0083] Through the above steps, the obtained bus GPS trajectory set is compressed into a two-dimensional density estimation problem.
[0084] Step S130: Generate the road centerline based on the kernel density.
[0085] In this embodiment, the step of generating the road centerline based on the kernel density specifically includes the following steps:
[0086] Let D(x, y) be the density at position (x, y), for each density layer l ∈ l max Let T l It is a binary image such that when D(x, y) is greater than l, the following holds true:
[0087] T l (x,y)=1
[0088] The algorithm recursively generates the skeleton image S of level l. l :
[0089] S l =skeletonize(T l +S l+1 )
[0090] and:
[0091] S lmax =skeletonize(T lmax )
[0092] This process involves repeatedly refining the image until it converges.
[0093] The parallel road generation method provided in this application utilizes the global nature of bus trajectories. This method can filter out most urban GPS drift in the generated parallel roads with bus routes, thereby improving the accuracy of road generation information and location.
[0094] The present invention tested the method on a dataset generated from public transportation in Shenzhen. The experimental results show that the model proposed in this invention can improve the accuracy of road generation information and location accuracy.
[0095] Example 2
[0096] Please see Figure 4 Embodiment 2 of this application provides a schematic diagram of the structure of a parallel road generation system, including: a data acquisition unit 110 for acquiring bus trajectory data; an estimation unit 120 for performing kernel density estimation based on the bus trajectory; and a road generation unit 130 for generating a road centerline based on the kernel density.
[0097] The detailed implementation of the parallel road generation system provided in the above embodiments of this application can be found in Embodiment 1, and will not be repeated here.
[0098] The parallel road generation system provided in this application utilizes the global nature of bus trajectories. This method can filter out most urban GPS drift when generating parallel roads with bus routes, thereby improving the accuracy of road generation information and location.
[0099] Example 3
[0100] Please see Figure 5 This is a schematic diagram of the terminal structure according to an embodiment of this application. The terminal 50 includes a processor 51 and a memory 52 coupled to the processor 51.
[0101] The memory 52 stores program instructions for implementing the parallel road generation method.
[0102] The processor 51 is used to execute the program instructions stored in the memory to control the generation of parallel roads.
[0103] The processor 51 can also be referred to as a CPU (Central Processing Unit). The processor 51 may be an integrated circuit chip with signal processing capabilities. The processor 51 can also be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), an off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor can be a microprocessor or any conventional processor.
[0104] It is understood that the technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0105] The above are merely preferred embodiments of this application, and only specifically describe the technical principles of this application. These descriptions are only for explaining the principles of this application and should not be construed as limiting the scope of protection of this application in any way. Based on this explanation, any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application, as well as other specific embodiments of this application that can be conceived by those skilled in the art without creative effort, should be included within the scope of protection of this application.
Claims
1. A method for generating parallel roads, characterized in that, Includes the following steps: Obtain bus trajectory data; Kernel density estimation is performed based on the bus trajectory; The road centerline is generated based on the kernel density; The steps for obtaining bus trajectory data specifically include: Based on the route characteristics of the input bus GPS data, extract the trajectories of multiple buses on the same route; The bus routes are divided into different groups; Count the number of trajectories in each group, and select the group with the most trajectories to represent the trajectory of that bus route; Generate automatic pathfinding routes between adjacent trajectories; Merge trajectories with the same pattern in the group to obtain global bus trajectory data; The step of kernel density estimation based on the bus trajectory specifically includes the following steps: Divide the area to be generated into 1x1 meter cells; Set the pixel of the cell through which the trajectory passes to 1, and set the pixel of the cell without the trajectory to 0, to obtain a binary image in which each pixel has only two possible values. Count the number of trajectories in each cell to generate a two-dimensional histogram; convolving the two-dimensional histogram with a normal distribution function N(0, σ 2 ) representing an expected GPS error distribution, σ being chosen based on expected GPS error and road width; The step of generating the road centerline based on the kernel density specifically includes the following steps: Let D(x, y) be the density at position (x, y), for each density layer l∈1…l max Let T l It is a binary image such that when D(x, y) is greater than l, the following holds true: The algorithm recursively generates the skeleton image S of level l. l : and: This process involves repeatedly refining the image until it converges.
2. The parallel road generation method as described in claim 1, characterized in that, In the step of extracting the trajectories of multiple buses on the same route based on the route characteristics of the input bus GPS data, the GPS data of each bus contains 7 fields: ID, Lon, Lat, Time, Speed, Direction, and Bus Line. Among them, ID is a unique identifier for each bus; Lon is the longitude positioning of the GPS point in the geographic coordinate system; Lat is the latitude positioning of the GPS point in the geographic coordinate system; Time is the time when the location point is received; Speed is the instantaneous speed of the location point; Direction is the direction of travel of the location point; and Bus Line is the route identifier for each bus.
3. The parallel road generation method as described in claim 1, characterized in that, The step of dividing the bus trajectory into different groups specifically includes the following steps: using the K-modes algorithm to divide the bus trajectory into different groups.
4. The parallel road generation method as described in claim 1, characterized in that, The steps for generating an automatic pathfinding route between adjacent trajectories specifically include the following steps: The automatic route finding module of the Google Directions API is called to generate automatic route finding lines between adjacent tracks.
5. A generation system for the parallel road generation method as described in claims 1 to 4, characterized in that, include: The data acquisition unit is used to acquire bus trajectory data; An estimation unit is used to perform kernel density estimation based on the bus trajectory; A road generation unit is used to generate a road centerline based on the kernel density.
6. A terminal, characterized in that, The terminal includes a processor and a memory coupled to the processor, wherein, The memory stores program instructions for implementing the parallel road generation method according to any one of claims 1-4; The processor is used to execute the program instructions stored in the memory to control the generation of parallel roads.