CN116071722A - Lane geometric information extraction method, system, equipment and medium based on road section track - Google Patents
Lane geometric information extraction method, system, equipment and medium based on road section track Download PDFInfo
- Publication number
- CN116071722A CN116071722A CN202310173602.6A CN202310173602A CN116071722A CN 116071722 A CN116071722 A CN 116071722A CN 202310173602 A CN202310173602 A CN 202310173602A CN 116071722 A CN116071722 A CN 116071722A
- Authority
- CN
- China
- Prior art keywords
- lane
- track
- line
- lines
- road
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a road section track-based lane geometric information extraction method, which comprises the steps of carrying out track cluster analysis on track subsets passing through the same road to obtain track clusters belonging to the same lane, extracting track geometric centers from the track subsets in the same cluster in a non-parametric curve mode, and fitting to obtain candidate lane center lines; the lane change lane line identification method based on maximum group search is characterized in that an intersection relation diagram is established, lane change lane lines are identified through maximum group analysis and removed, a lane center line set which is not intersected with each other is obtained to serve as a basic lane center line, the number of lanes at two ends of the road section is rechecked through a Gaussian mixture model, and candidate lane center lines are extracted to serve as newly added lane center lines based on newly added Gaussian components and parameters. The method for acquiring the accurate geometric information of the lane based on the track data effectively extracts the accurate geometric information of the lane, is simple and effective, is easy to realize, has important significance for lane-level mapping application, and provides bottom data support of a high-precision map for automatic driving application.
Description
Technical Field
The invention belongs to the technical field of intelligent traffic information, and particularly relates to a road section track-based lane geometric information extraction method and system.
Background
The high-precision map provides important underlying data support for autopilot. The current road mapping technology for providing a data basis for the high-precision map is gradually changed from the traditional manual measurement to semi-automatic production based on various sensors. The current road mapping technical method is mainly based on satellite remote sensing images, unmanned aerial vehicle orthographic images and a professional mobile vehicle acquisition system for centralized acquisition of roads, and the defects of low updating frequency, high acquisition cost and the like are faced in spite of higher precision of the centralized acquisition mode of the professional equipment.
Because crowd-sourced track data is derived from a traveling vehicle carrying a GPS receiving module, the method has the advantages of wide coverage of urban road areas and high timeliness, and provides a production mode with high time freshness and low cost for road mapping. The existing method for extracting the lane geometric information based on crowdsourcing track data can be divided into a method for extracting the lane geometric information based on track point local distribution and a lane estimation method based on a track clustering algorithm according to a processing object.
The number of the publication CN202010140103, lane geometry information extraction based on local distribution of track points is disclosed: when the vehicle runs along the lane center line, the track points calculated by the vehicle-mounted GPS receiver module follow Gaussian distribution in the direction perpendicular to the lane center line, and the average value center is the lane center line position. Thus, over a road cross-section, the distribution of trajectory points can be regarded as a gaussian mixture distribution of a plurality of gaussian components, while the solution of the lane geometry information can be regarded as a solution of the parameters in the gaussian mixture model.
The specific algorithm flow is as follows: selecting track point data in a sliding window by adopting the sliding window, and projecting the track point data to the direction perpendicular to the lane; introducing known lane widths as known parameters, iterating the number of lanes, and solving the center position, weight and the like of each lane by utilizing a track point distribution fitting Gaussian mixture model; evaluating the confidence coefficient of the solutions under different lane conditions through a risk assessment model; and determining the number and the positions of the lanes according to the confidence.
Publication number CN202111043369, lane estimation method based on trajectory clustering algorithm: compared with the lane geometric information extraction method aiming at the distribution of the local GPS points, the method uses the complete track or sub-track as the input of the cluster, and emphasizes the sequence of the track observation points and the connection relation of the track. And extracting track clusters passing through the same lane through track clustering, wherein each group of clustered track clusters are used for estimating the lane center line. Determining an initial clustering center line according to the distribution state of the target track data; clustering the target track data based on the initial clustering center line to obtain a clustered target clustering center line; and determining a target lane to which the vehicle to be detected belongs based on the target cluster center line. Therefore, the initial clustering center line can be rapidly determined through the distribution state of the acquired target track data with preset quantity; and carrying out clustering processing on the target track data based on the initial clustering center line to obtain an accurate target clustering center line, and further rapidly judging a target lane to which the vehicle to be detected belongs according to the target clustering center line.
The track data can reflect details of the urban vehicle road network, such as the position, width, driving direction, etc. of the lanes. In urban road design, the import and export lanes of a plane intersection, the estuary type parking stations and the collecting and distributing lanes of a expressway are typical real lane number change scenes. The existing method for extracting the geometric information of the lanes based on the track data, such as K-Means clustering, kernel density estimation, gaussian mixture model and the like, essentially models the distribution characteristics of track points in a certain range to acquire the number and the positions of the lanes. However, the method depends on the statistical characteristics of track data, and is sensitive to the distribution of the track data by a Gaussian mixture model in the identification process of the number of lanes and the center position of the lanes, parameters often need to be manually adjusted based on prior information, the accuracy is low, common lane geometry and topology communication information of a divergence and confluence region in an urban road network can not be extracted, and lane change conditions and connection relations in the actual road network are obtained. The lane estimation method based on the track clustering algorithm depends on track data, and when a lane change track generated by driving behavior exists in the track data, the positions and the number of lanes generated by the track are inconsistent with those of actual lanes.
Urban road networks often have more complex lane-level topological structures, such as scenes in which the number of lanes changes in a complex manner, and accurate representation of complete geometric information and topological structures of the lane road networks under a variable-lane scene based on track data is still a very difficult task. Therefore, there is a difficulty in constructing truly complete structured lane geometry information for urban vehicle road networks using the above-described method.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides a method and a system for extracting geometric information of a fine lane based on track data. And (3) through track clustering, co-track fitting of candidate lane lines, and complex traffic scene recognition and processing aiming at the change of the number of lanes, extracting basic lane lines and newly added lane change lines, and realizing complete fine lane center line geometric information extraction.
In view of this, based on one aspect of the present application, a method for extracting geometric information of a lane based on a track is provided, and a non-parametric curve fitting method based on iterative optimization performs a track cluster analysis on a subset of tracks passing through the same road to obtain a track cluster belonging to the same lane; extracting track geometric centers from track subsets in the same cluster in a non-parametric curve mode, fitting candidate lane center lines, taking the lane center lines as nodes in the graph, and establishing an intersection relation graph among the lane lines; identifying lane change lane lines through the maximum group, and obtaining a lane center line set which is mutually disjoint as a basic lane line; and extracting candidate lane center lines as new lane center lines based on the newly added Gaussian components at the two ends of the road section, determining lane change connection positions according to the basic lane lines and the new lane center lines, and constructing a lane geometric information and topology relation diagram.
Further preferably, the obtaining the basic lane line includes regarding the lane line as a node in the graph, simultaneously, recording geometric features of the lane line by node weights, recording intersection relation features between the lane lines by edge weights, constructing a candidate lane line intersection relation graph by using the lane center lines fitted by each road section, defining each candidate lane line as a node of the graph, and if the two lane lines do not intersect, connecting the edges, constructing an undirected graph, and extracting a set of lane center lines which do not intersect each other as the basic lane line by the undirected graph.
Further preferably, the performing lane center line fitting on the track clusters of the same lane further includes: and fitting the lane center line by adopting an iterative optimization non-parameterized curve fitting method according to the time sequence of the track observation points in the track cluster of the same lane.
Further preferably, the lane centerline fitting includes: randomly selecting a track line as an initial lane line, and calculating a cross section line of each point of the lane line; and calculating the geometric center of the intersection point of the cross section line and the input track cluster as the center point of the lane, and connecting the intersection points of the track clusters as the fitted lane center line.
Further preferably, in the road section with changed lanes, the number of lanes at the lane number increasing end is taken as the total lane number of the road section, the Gaussian components corresponding to the basic lane line and the lane number increasing end are calculated, and the newly added Gaussian components are determined and the newly added lane line is extracted.
Further preferably, the clustering of the tracks in the lane includes: initializing each track segment into a cluster, calculating the Hausdorff distance between each pair of tracks, merging the clusters with the Hausdorff distance smaller than the distance threshold value into one track cluster of the same lane, calculating the inter-cluster distance between all track clusters, merging the clusters with the inter-cluster distance smaller than the distance threshold value into one track cluster of the same lane until all the inter-cluster distances are larger than the distance threshold value, and completing the track clustering of the lanes.
Further preferably, the clustering of the tracks in the lane includes: and obtaining a track cluster belonging to each road section by a density-based clustering algorithm, and taking all track sets belonging to the same road section as a basic unit for extracting lanes.
Further preferably, a basic lane line is obtained based on a maximum group algorithm of the graph, a lane line intersection relation graph of candidate lane lines is constructed by utilizing the lane center lines fitted by each road section, each candidate lane line is defined as a node of the graph, edges between the nodes represent intersection relation between lanes, if two lane lines are disjoint, the edges are connected, an undirected graph is constructed, and a lane center line set which is disjoint each other is extracted through the undirected graph to serve as the basic lane line.
Further preferably, each candidate lane line is defined as a vertex set v= {1, …, n } of the graph, an edge set between nodesRepresenting the intersection relationship between lanes, if two lane lines are not intersected, the edges between the nodes represented by the two lane lines are connected, and G= (V, E) is constructed as a vertex set V= {1, …, n } and an edge set }>If the subset U E V and for any two of the vertices U, V E U, there is (U, V) E, then U is the complete subgraph of G and the largest complete subgraph is the largest clique of the graph.
Further preferably, when there are a plurality of maximum clusters, a set with the largest sum of weights of the selected lane line nodes is used as a basic lane line of the non-lane change, and according to the formula:
and calculating a node weight, namely enabling the lane line weight with smaller track yaw angle average change to be larger by negative logarithmic function transformation, wherein the track weight with larger track yaw angle average change is smaller, and delta yaw represents the track yaw angle average change.
Further preferably, the identifying the number of lanes at two ends of the road segment by using the prior gaussian mixture model comprises: identifying the main direction of the road and the cross-section direction of the road perpendicular to the main direction of the road by the movement direction of the track in the window, and projecting all GPS track points in the window in the cross-section direction of the road to obtain data in the cross-section direction of the road Projection distribution, the position of the center line of each lane is obtained as the peak value mu of the Gaussian component of the lane j The distribution condition of the track observation points on each lane corresponds to the standard deviation sigma of the distribution of each Gaussian component data j Determining the weight of each Gaussian component distribution according to the proportion of the track number in each lane, and constructing a probability density function of the Gaussian mixture model:
wherein k is the number of Gaussian components, j=1, 2, …, k, w j Weights distributed for the jth gaussian component, where w 1 +w 2 +…+w k =1。
Further preferably, the evaluation function selected by the number of lanes is optimally designed,
evaluating the fitting condition of a Gaussian mixture model, preventing overfitting, restraining the lane width value, controlling the distance between peak distribution of each Gaussian component, taking the distance as a punishment item for preventing overfitting when likelihood function is increased along with the mixing score, and selecting the Gaussian mixture model with the minimum evaluation function value as an optimal model, wherein lambda is a regularization parameter and delta h is calculated by the method i Representing the average distance between the centerlines of adjacent lanes.
According to another aspect of the present application, there is provided a track-based lane geometry information extraction system, including: the system comprises a lane track clustering module, a lane center line fitting module, a basic lane line extracting module, a lane change scene detecting module, an additional lane line extracting module, a complete lane line constructing module, a lane track clustering module, a lane center line fitting module, a basic lane line extracting module, a lane change scene detecting module, an additional lane line extracting module, a complete lane line constructing module, and a lane track clustering module, wherein the lane track clustering module is used for carrying out track cluster analysis according to a track subset of the same road section to obtain a track cluster of the same lane and dividing the road section into track clusters of different lanes; the lane center line fitting module extracts the geometric center of the track according to the track subset in the same track cluster to obtain a candidate lane center line; the basic lane line extraction module is used for obtaining basic lane lines which are mutually disjoint according to the candidate lane lines; the lane change scene detection module is used for judging whether the road section belongs to a lane change scene according to the track data of the road section; the newly added lane line extraction module is used for extracting a candidate lane center line based on newly added Gaussian components at two ends of the road section to serve as a newly added lane center line; and the complete lane line construction module is used for constructing the geometric information and the topological relation of the lane under the lane change scene according to the extracted basic lane lines and the newly added lane lines.
Further preferably, the basic lane line extraction module acquires basic lane lines based on a maximum group algorithm of the graph, a candidate lane line intersection relation graph is constructed by utilizing lane center lines fitted by each road section, each candidate lane line is defined as a node of the graph, edges between the nodes represent intersection relations between lanes, if two lane lines are disjoint, the edges are connected, an undirected graph is constructed, and a lane center line set which is disjoint each other is extracted through the undirected graph to serve as the basic lane lines.
Further preferably, each candidate lane line is defined as a vertex set v= {1, …, n } of the graph, an edge set between nodesRepresenting the intersection relationship between lanes, if two lane lines are not intersected, the edges between the nodes represented by the two lane lines are connected, and G= (V, E) is constructed as a vertex set V= {1, …, n } and an edge set }>If the subset U E V and for any two of the vertices U, V E U, there is (U, V) E, then U is the complete subgraph of G and the largest complete subgraph is the largest clique of the graph.
Further preferably, when there are a plurality of maximum clusters, a set with the largest sum of weights of the selected lane line nodes is used as a basic lane line of the non-lane change, and according to the formula:
and calculating a node weight, namely enabling the lane line weight with smaller track yaw angle average change to be larger by negative logarithmic function transformation, wherein the track weight with larger track yaw angle average change is smaller, and delta yaw represents the track yaw angle average change.
Further preferably, the identifying the number of lanes at two ends of the road segment by using the prior gaussian mixture model comprises: identifying the main direction of a road and the direction of a cross section of the road perpendicular to the main direction of the road by the movement direction of the track in the window, projecting all GPS track points in the window in the direction of the cross section of the road to obtain the projection distribution of data in the direction of the cross section of the road, and obtaining the position of the center line of each lane as the peak value mu of the Gaussian component of the lane j The distribution condition of the track observation points on each lane corresponds to the standard deviation sigma of the distribution of each Gaussian component data j Determining the weight of each Gaussian component distribution according to the proportion of the track number in each lane, and constructing a probability density function of the Gaussian mixture model:
wherein k is the number of Gaussian components, j=1, 2, …, k, w j Weights for the j-th gaussian component distribution, wherein,
w 1 +w 2 +…+W k =1。
further preferably, the evaluation function selected by the number of lanes is optimally designed,
estimating the fitting condition of the Gaussian mixture model, preventing overfitting, restraining the lane width value, controlling the distance between peak distribution of each Gaussian component, and preventing as likelihood function as the mixing score increasesSelecting a Gaussian mixture model with the minimum evaluation function value as an optimal model by using a superfilling penalty term, wherein lambda is a regularization parameter and delta h i Representing the average distance between the centerlines of adjacent lanes.
According to another aspect of the present application, there is provided an electronic device, including: a processor; and a memory storing a program, wherein the program comprises instructions that when executed by the processor cause the processor to perform the trajectory-based lane geometry information extraction method according to the above.
According to another aspect of the present application, a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the trajectory-based lane geometry information extraction method according to the above is provided
The invention provides a method and a system for extracting geometric information of a fine lane based on track data, wherein geometric shape information of a lane center line is extracted through a non-parametric fitting method of track clustering and track iterative optimization, the interference influence of a lane changing track of a vehicle is effectively solved through a lane changing identification method based on a maximum group, scene inspection of changing the number of lanes and extraction of newly added lane information are considered, and the construction of a lane connection relation under the scene of changing the number of lanes is realized. The method reduces the track quantity requirement and parameter solving calculation amount, reduces the dependence on priori information such as lane width, lane curvature and the like, and is more suitable for extracting lane information under a lane change scene, such as a divergence and confluence region. The method reduces the cost of acquiring detailed lane geometric information of the urban road, can effectively extract accurate lane geometric information only based on track data, is simple and effective, is easy to realize, has important significance for lane-level mapping application, and provides bottom data support of a high-precision map for future upper-layer automatic driving application.
Drawings
Fig. 1 illustrates a road section fine lane geometry information extraction system based on trajectory data in the present exemplary embodiment;
fig. 2 is a flowchart of a fine lane geometry information extraction method based on trajectory data in the present exemplary embodiment;
FIG. 3 is a schematic view of a lane clustering effect obtained by the method in the present embodiment;
FIG. 4 is a schematic diagram of an iterative optimization fit of lane centerlines in the present exemplary embodiment;
FIG. 5 is a schematic diagram of a lane change identification and basic lane-line extraction method based on maximum cliques in the present exemplary embodiment;
fig. 6 is a schematic diagram of extraction of new lane lines based on a gaussian mixture model in the present exemplary embodiment;
FIG. 7 is a schematic diagram showing the construction of a complete lane line in a lane change scene obtained by the embodiment;
fig. 8 is a schematic diagram of an electronic device according to an embodiment of the present application.
Wherein, 100-electronic device, 101-processor, 102-bus, 103-memory.
Detailed Description
Embodiments of the present application will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present application are shown in the drawings, it is to be understood that the present application may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and examples of the present application are for illustrative purposes only and are not intended to limit the scope of the present application.
It should be understood that the various steps recited in the method embodiments of the present application may be performed in a different order and/or performed in parallel. Furthermore, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the present application is not limited in this respect.
The term "including" and variations thereof as used herein are intended to be open-ended, i.e., including, but not limited to. The term "based on" is based at least in part on.
According to the invention, candidate lane lines are extracted through lane track clustering and lane center line fitting, and a recognition and processing method is designed aiming at complex traffic scenes with the number of lanes changed, so that basic lane lines and newly added lane changing lines are automatically extracted, and complete fine lane center line geometric information extraction is realized. Comprising the following steps: and acquiring a track data set of each road section. And extracting sub-tracks of the lane sections according to GPS points continuously acquired at regular time intervals on the lane sections, and clustering the starting point-ending point pairs of the sub-tracks to obtain track clusters belonging to each section.
And extracting track clusters belonging to the same lane to complete lane track clustering. The method can be concretely realized by adopting the following steps:
initializing each track segment as an original cluster, merging the clusters with the distance between the two original clusters smaller than a distance threshold into a new cluster, further calculating the distance between the new cluster and the original cluster, merging the clusters with the distance between the clusters smaller than the distance threshold until the distances between all the clusters after merging are larger than the distance threshold, and finishing lane track clustering; fitting the lane center line of a track cluster belonging to the same lane, and fitting the lane center line by adopting an iterative optimization non-parametric curve fitting method according to the time sequence of the track observation points in the track cluster of the same lane; constructing a lane line intersection relation diagram for the center lines of the candidate lanes extracted from each road section, constructing an undirected diagram, and extracting a set of mutually disjoint lane center lines so as to realize the extraction of basic lane lines; identifying the number of lanes at two ends of a road section, and determining a lane change scene of the road section; in a road section with changed lanes, taking the number of lanes at the lane number increasing end as the total number of lanes of the road section, corresponding the extracted basic lane line data with Gaussian components at the lane number increasing end, calculating newly increased Gaussian components to determine newly increased lane lines, and extracting the newly increased lane lines; the connection relation between the newly added lane line and the basic lane line is constructed, the public subsequence part between the newly added lane line and the basic lane line is deleted, and the lane change connection position is recorded, so that the construction of the geometric and topological relation of the lane under the lane change scene is realized.
The invention designs a corresponding practical system in a modularized mode, and provides a lane geometric information extraction system based on track data, which comprises the following components:
the system comprises a lane track clustering module, a lane center line fitting module, a basic lane line extracting module, a lane change scene detecting module, an added lane line extracting module and a complete lane line constructing module.
The lane track clustering module is used for dividing track data in the road section into track clusters of different lanes and acquiring the track clusters of the different lanes according to the track data of the same road section; the lane center line fitting module is used for iterative optimization fitting of the lane center lines, and obtaining geometric representation fitting candidate lane lines of the lane center lines according to the track clusters of the same lane; the basic lane line extraction module is used for extracting basic lane lines which are mutually disjoint, and obtaining the basic lane lines which are mutually disjoint according to the candidate lane lines fitted by the lane center line fitting module; the lane change scene detection module is used for judging whether the road section belongs to a scene with changed lane numbers or not, and judging whether the road section belongs to a lane change scene or not according to the track data of the same road section; the newly added lane line extraction module is used for extracting newly added lane lines of the road section and determining the newly added lane lines according to the track data of the same road section and the extracted basic lane lines; the complete lane line construction module is used for constructing a complete lane line structure of the road section, and constructing a final lane line extraction result of the road section according to the extracted basic lane line and the newly-added lane line.
The invention will be further described with reference to the drawings and examples.
As shown in fig. 1, a system for extracting information of a fine lane of a road section based on track data according to an exemplary embodiment of the present application includes: the lane track clustering module 1 is used for dividing track data in a road section into track clusters of different lanes; the lane center line fitting module 2 is used for iterative optimization fitting of the lane center line; the basic lane line extraction module 3 is used for extracting basic lane lines which are mutually disjoint; the lane change scene detection module 4 is used for judging whether the road section belongs to a scene with changed lane numbers; the newly added lane line extraction module 5 is used for extracting newly added lane lines of the road section; and the complete lane line construction module 6 is used for constructing a complete lane line structure of the road section.
Fig. 2 is a schematic flow chart of a fine lane geometric information extraction method based on track data in an exemplary embodiment of the present application, which includes the steps of: and obtaining road section track data, clustering lane tracks, fitting a lane center line, extracting basic lane lines, processing a lane change scene and completing the construction of a lane line structure.
And acquiring track data of each road section.
In the embodiment, a track sequence formed by GPS points continuously collected at fixed time intervals on a lane section is adopted, and a road intersection is taken as an interval to break a long time sequence track into sub-tracks. All sub-tracks can be divided into track sets of different road segments by using an OD (origin-destination) point pair cluster based on the start and end positions of the sub-tracks. The start point and the end point of the sub-track can quickly and effectively determine the spatial position of the track, and track clustering (DBSCAN clustering can be adopted) is performed by calculating the distance sum between the start point and the end point of the sub-track, so that sub-tracks with similar spatial positions and denser distribution are clustered. And finally obtaining track sets belonging to different road sections. The track set belonging to the same road section is used as the input for extracting the geometric information of the subsequent lanes.
Further, in the embodiment, the sub-track sets belonging to the same road section are taken as input, the track hierarchical clustering algorithm is adopted to perform the track clustering, and the sub-track sets of the same road section are further divided into different lanes, so that the track clusters belonging to each lane are extracted.
The method specifically comprises the following steps:
firstly initializing each track segment in a road section into a cluster, calculating Hausdorff distance between each pair of tracks, calculating the distance metric by taking the complete distribution of track sequences into consideration, and calling a formula according to the unidirectional Hausdorff distance d (A, B) from track A to track B and the unidirectional Hausdorff distance d (B, A) from track B to track A:
D h (A,B)=max{d(A,B),d(B,A)}
calculation ofBidirectional Haoskov distance D between track a and track B h (A, B), wherein A and B are tracks comprising a series of ordered GPS observation points, for each point a in A, calculating the nearest point B to the point in track B, calculating the distance between the two points, taking the maximum value in the distance as the value of d (A, B), and similarly, obtaining the unidirectional Haoskov distance d (B, A) from track B to track A. Finally, two-way Haoskov distance D h (A, B) taking the maximum value of d (A, B) and d (B, A).
Tracks with distances less than a distance threshold (the distance threshold may be optimally lane width) are merged into a new cluster of tracks.
The combined clusters are used as new clusters, the distance between the new clusters and other clusters is recalculated, the maximum distance between two tracks is adopted among the clusters until the distance among the clusters is larger than a set distance threshold value, and the lane track clustering is completed.
Fig. 3 is a schematic view showing a lane clustering effect obtained by the method according to the present invention in the present exemplary embodiment. Obtaining a lane clustering result schematic of the graph (b) according to the original track data of the graph (a).
And (5) carrying out iterative optimization fitting on the lane center line. And fitting the lane center line to the track clusters belonging to the same lane.
Aiming at the time sequence of the track observation points, the embodiment improves the traditional main curve algorithm for fitting the point set data, designs a non-parametric curve fitting method suitable for iterative optimization of the track data, and orderly extracts the geometric center of the track to fit the lane center line by using a non-parametric curve form for track subsets in the same cluster. The lane center line fitting module fits and outputs the lane center line as the lane center line according to the track cluster of the same lane.
Fig. 4 is a schematic diagram illustrating iterative optimization fitting of lane centerlines in an exemplary embodiment of the present application, in which,
(a) Randomly selecting a track line as an initial track line; (b) calculating a cross-sectional line for each point of the initial lane line; (c) Calculating the geometric center of the intersection point of the cross section line and the input track cluster, taking the geometric center as the center point of the lane, connecting the intersection points, and updating the fitted lane center line; (d) And (c) iterating the steps (a) and (b) until the position of the central point is stable and does not become a convergence condition, and obtaining a final fitting result.
And extracting a basic lane line by a lane change identification method based on the maximum group.
Based on the characteristic of the common disjoint between the actual lanes, the embodiment converts the basic lane center line selection problem of non-lane change into the maximum group problem in the solving graph theory.
Fig. 5 is a schematic diagram of a lane change identification and basic lane line extraction method based on the maximum group in the present exemplary embodiment, where:
(a) The method comprises the steps of (1) setting a lane line intersection relation diagram for candidate lane center lines, wherein A, B, C, D, E, F and G in the diagram respectively represent the numbers of the candidate lane center lines, and constructing the lane line intersection relation diagram based on the extracted candidate lane center lines;
(b) As a lane line intersection relation graph, each candidate lane line is defined as a node V= { A, B, C, D, E, F, G } in the graph, edges between the nodes represent the intersection relation between lanes, and if two lane lines do not intersect, the edges are connected, so that an undirected graph is constructed;
(c) Identifying the center line of the basic lane, and taking a set { A, B, C, D } of the center lines of the lanes which are mutually disjoint as an extracted basic lane line result;
(d) And (3) for the corresponding maximum group identification result, the maximum group result of the graph in the box comprises basic lane line nodes A, B, C and D which are mutually disjoint, and the rest nodes E, F and G in the graph are regarded as lane change lines to be removed. The central lines of mutually disjoint basic lanes are extracted through identifying the maximum groups in the undirected graph, and lane change lines are removed efficiently.
In the present exemplary embodiment, an undirected graph is constructed using the following lane change recognition algorithm based on the maximum clique, and a set of lane centerlines that do not intersect each other is extracted based on lane centerline recognition.
Each candidate lane line is defined as the vertex set v= {1, …, n } of the graph, the edge set between nodesIndicating the intersection relationship between lanes if there is no line between two lanesIntersecting, edges between nodes represented by the two lane lines are connected. G= (V, E) is the vertex set v= {1, …, n } and the edge set +.>Is an undirected graph of (1). If the subset U.epsilon.V and for any two of the vertices U, v.epsilon.U, there is (U, V) E, then U is said to be a complete subgraph of G. As shown in fig. 5, the lane line candidates a, B, C, D, E, F, G in (a) are defined as vertex sets v= { a, B, C, D, E, F, G } of the graph, respectively. Taking lane line a as an example, lane line a and lane lines B, C, D, F, G do not intersect, corresponding undirected edges (a, B), (a, C), (a, D), (a, F), (a, G) are respectively constructed. And so on, the lane crossing relation diagram shown in (b) is obtained. The maximum clique of a graph refers to the largest complete subgraph in the graph. Solving the maximum clique to obtain a mutually disjoint set of basic lane lines { A, B, C, D }, which are reserved as basic lane lines of the non-lane change as shown in (C). (d) The maximum group in the graph is in the middle box, and the rest nodes E, F and G in the graph are regarded as lane change lines to be removed.
Considering the more complex case where there are multiple maximum cliques at the same time, this embodiment adds more geometric features of the trajectory to make the determination. The method comprises the steps of measuring the degree of lane change possibly occurring on a lane line based on the average change condition of track yaw angles, introducing node weight attributes, enabling the lane line weight with smaller track yaw angle average change to be larger and the track weight with larger track yaw angle average change to be smaller through negative logarithmic function transformation, and according to the formula:
node weight is calculated. Where n represents the total number of track points of the lane line represented by the node, and Δyw represents the average change in track yaw angle.
The invention converts the node weight attribute into the maximum weight group problem, and when a plurality of maximum groups exist, the set with the maximum sum of the weights of the selected lane line nodes is used as the basic lane line of the non-lane change.
And detecting a lane change scene.
In the embodiment, the prior Gaussian mixture model is adopted to detect the lane change scene (other models can be adopted to identify), the number of lanes at two ends of the road section is respectively identified, and whether the road section belongs to the lane change scene is judged. If the number of lanes at the two ends of the road section is consistent, the condition that the road section has no lane change is indicated, the number of lanes of the road section is equal to the number of the extracted basic lane lines, the extracted basic lane lines are combined, and the final lane line structure integration construction is completed.
And if the detection result shows that the number of lanes at the two ends of the road section changes, determining the road section as a scene of the lane change.
Further, the identifying the number of lanes at two ends of the road segment by using the prior gaussian mixture model in the present exemplary embodiment specifically includes: the projection distribution of the track points on the road cross section has the characteristics of gathering around the center line of each lane and then gradually becoming sparse towards the two ends of the lane by combining with the GPS track data distribution rule. Starting from the starting point of a road section, constructing a rectangular moving window, taking the maximum coverage width of the track on the road surface as the side length of the window, detecting the track observation point in the window, and gradually moving along the main direction of the road until reaching the end point of the road section. Identifying the main direction of a road and the direction of a cross section of the road perpendicular to the main direction of the road by the movement direction of the track in the window, projecting all GPS track points in the window in the direction of the cross section of the road to obtain the projection distribution of data in the direction of the cross section of the road, modeling based on a Gaussian mixture model, and obtaining the position of the center line of each lane as the peak value mu of each Gaussian component in the Gaussian mixture model j According to the distribution condition of the track observation points on each lane, the standard deviation sigma of the distribution of each Gaussian component data is corresponding to j Determining the weight distributed by each component j in the Gaussian mixture model according to the proportion of the track number in each lane, and constructing a probability density function of the Gaussian mixture model:
the density function p (x) is used to represent the comprehensive overview of the Gaussian mixture modelThe rate. Wherein x represents a sample of locus points projected to the road cross-section direction to be calculated, k is the number of Gaussian mixture components, j=1, 2, …, k, μ j Represents the average value, sigma, of samples belonging to the jth gaussian component j Represents the standard deviation, w, of the samples in the jth Gaussian component j Weights distributed for the jth gaussian component, where w 1 +w 2 +…+w k =1。
Classical gaussian mixture models require manual setting of the number of mixes, i.e. the number of lanes, typically by means of red-pool information criteria and bayesian information criteria to determine the number of components in the mixture model and to determine the best fit model. However, for the application scene of the lane number recognition, the past information criterion judges that the fitting situation can occur, and the fitting effect looks better along with the increase of the number of components. In order to solve the problem of excessive number of mixed components caused by over fitting, the prior condition of variable lane width is introduced on the basis of the traditional Gaussian mixture model, and an evaluation function for selecting the number of lanes is optimally designed so as to improve the applicability of the algorithm in lane number identification. The evaluation function for the lane number selection is as follows:
In the formula, for the track pointn represents the total number of track points of the lane line represented by the node, k is the number of Gaussian mixture components, and p (x i |θ k ) Representing a sample value x i At the Gaussian mixture model parameter θ k Gaussian probability value under conditions, where the model parameter θ k Can be expressed as theta k (w k ,μ k Sigma). Right front item of equal sign +.>For the log-likelihood function, for evaluating the model fitting, the latter term +.>To prevent over-fitting of the regularization term, where λ is the regularization parameter, a default value of 1, Δh i The average distance between the center lines of adjacent lanes is represented, a sigmoid function for constraint of lane width information is introduced into a regular term, and the lane width value is usually in a certain range, for example, 3.25 m-3.75 m under the condition of actual urban roads, the sigmoid function can map real values to intervals of (0, 1), and constraint in a certain range is carried out on the lane width value in the model, so that the distance between peak distribution of each Gaussian component is controlled, and the distance is used as a penalty term for preventing overfitting when the likelihood function is increased along with the increase of the mixing score. And finally, selecting a Gaussian mixture model with the minimum function evaluation value as an optimal model by the model, and determining the number of the mixture components as the number of lanes so as to automatically identify the number of lanes. And extracting newly added lane lines for the lane change scene. Fig. 6 is a schematic diagram showing the extraction of new lane lines based on the gaussian mixture model in the present exemplary embodiment.
And extracting new lane lines except the basic lane lines from the road sections with lane changes. And identifying lane lines of the newly added lane by using the newly added Gaussian components, adding the newly added lane lines on the basic lane lines, and then carrying out track data road cross section projection.
Taking the number of lanes at the lane number increasing end as the total number of lanes after the actual lanes of the road section are increased. And (3) corresponding the extracted basic lane line data with Gaussian components at the lane number increasing end, taking the difference set of all the Gaussian components at the lane number increasing end and the Gaussian components corresponding to the basic lane line as the newly increased Gaussian components, and extracting the lane center line corresponding to the newly increased Gaussian components as the newly increased lane line.
And constructing a complete lane line under the lane change scene. And reserving the obtained disjoint basic lane lines and the new lane lines aiming at the scene that the number of lanes changes. When partial lanes of the newly-added lane line and the basic lane line overlap, deleting a common subsequence part between the newly-added lane line and the basic lane line, reserving a connection relation between the newly-added lane line and the basic lane line, and recording a lane change connection position so as to realize construction of the geometric and topological relation of the lane in a lane change scene.
Fig. 7 shows a complete schematic diagram of a complete lane line construction in a lane change scenario. The finally extracted lane lines are formed by adding new lane line parts reflecting the real change of the road network to the basic lane lines which are not intersected, and the task of constructing the lane lines is completed. Wherein, (a) raw trajectory data, (b) trajectory clustering, (c) lane centerline fitting. (d) Basic lane line extraction, (e) newly added lane line extraction.
As shown in fig. 8, the electronic device 100 includes: a processor 101 and a memory 103. Wherein the processor 101 is coupled to the memory 103, such as via bus 102.
The structure of the electronic device 100 is not limited to the embodiments of the present application.
The processor 101 may be a CPU, general purpose processor, DSP, ASIC, FPGA or other programmable logic device, transistor logic device, hardware component, or any combination thereof. Which may implement or perform the various exemplary logic blocks, modules, and circuits described in connection with this disclosure. The processor 101 may also be a combination that implements computing functionality, e.g., comprising one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.
While the applicant has described and illustrated the embodiments of the present invention in detail in connection with the drawings of the specification, it should be understood by those skilled in the art that the foregoing embodiments are merely preferred embodiments of the present invention, and are provided solely for the purposes of better understanding the principles of the present invention and are not intended to limit the scope of the invention, but one skilled in the art may make various modifications or supplements to the specific embodiments described or substitute in a similar manner without departing from the spirit of the invention or beyond the scope of the appended claims.
Claims (18)
1. A method for extracting geometric information of lanes based on road section track is characterized in that track cluster analysis is carried out on track subsets passing through the same road, sub-track sets of the same road section are divided into different lanes, and track clusters belonging to each lane are extracted;
extracting track geometric centers from track subsets in the same cluster in a non-parametric curve mode, fitting candidate lane center lines, taking the lane center lines as nodes in the graph, and establishing an intersection relation graph among the lane lines; identifying lane change lane lines through the maximum group, and obtaining a lane center line set which is mutually disjoint as a basic lane line; and extracting candidate lane center lines as new lane center lines based on the newly added Gaussian components at the two ends of the road section, determining lane change connection positions according to the basic lane lines and the new lane center lines, and constructing a lane geometric information and topology relation diagram.
2. The method of claim 1, wherein the obtaining the basic lane lines includes regarding the lane lines as nodes in the graph, while the node weights record geometric features of the lane lines, the edge weights record intersection relationship features between the lane lines, constructing a candidate lane line intersection relationship graph using the lane centerlines fitted by each road segment, defining each candidate lane line as a node of the graph, defining edges between the nodes as intersection relationships between the lanes, connecting the edges if the two lane lines do not intersect, constructing an undirected graph, and extracting a set of mutually disjoint lane centerlines as the basic lane lines through the undirected graph.
3. The method of claim 1, wherein the lane centerline fitting comprises: fitting a lane center line by adopting an iterative optimization non-parameterized curve fitting method according to the time sequence of the track observation points in the track cluster of the same lane; randomly selecting a track line as an initial lane line, calculating a cross section line of each point of the lane line, calculating the geometric center of the cross section line and the intersection point of the input track clusters as the center point of the lane, and connecting the intersection points of the track clusters as the fitted lane center line.
4. The method according to claim 1, wherein the number of lanes at the lane change increasing end is taken as the total number of lanes of the road section, the gaussian component corresponding to the basic lane line and the lane number increasing end is calculated, and the newly added gaussian component is determined and extracted.
5. The method of one of claims 1 to 4, wherein the clustering of the trajectories in the lane comprises: initializing each track segment to a cluster, calculating the Hastell distance between each pair of tracks, merging the clusters with the Hastell distance smaller than the distance threshold into a new track cluster, and calculating the inter-cluster distance between the new track cluster and other track clusters, merging the clusters with the inter-cluster distance smaller than the distance threshold value into track clusters of the same lane until all the inter-cluster distances are larger than the distance threshold value, and completing the track clustering of the lane.
6. The method of one of claims 1 to 4, wherein the clustering of the trajectories in the lane comprises: and obtaining a track cluster of the lanes belonging to each road section based on a density clustering algorithm, and taking all track sets belonging to the same road section as a basic unit for extracting the lanes.
7. The method according to claim 2, which comprisesCharacterized in that each candidate lane line is defined as a vertex set V {1, …, n } of the graph, an edge set between nodesRepresenting the intersection relationship between lanes, if two lane lines are not intersected, the edges between the nodes represented by the two lane lines are connected, and G (V, E) is constructed as a vertex set V {1, …, n } and an edge setIf the subset U E V and for any two of the vertices U, V E U, there is (U, V) E, then U is the complete subgraph of G and the largest complete subgraph is the largest clique of the graph.
8. The method of claim 7, wherein when there are a plurality of maximum cliques, a set of the maximum sum of weights of the selected lane-line nodes is taken as a basic lane-line of the non-lane-change, according to the formula:
and calculating a node weight, namely enabling the lane line weight with smaller track yaw angle average change to be larger by negative logarithmic function transformation, wherein the track weight with larger track yaw angle average change is smaller, and delta yaw represents the track yaw angle average change.
9. The method of any one of claims 1-4, 7, 8, wherein identifying the number of lanes at both ends of the road segment comprises: detecting track observation points in the window by taking the maximum coverage width of the track in the road surface as the side length of the window, gradually moving along the main direction of the road from the starting point of the road section until the end point of the road section, identifying the main direction of the road and the cross section direction of the road perpendicular to the main direction of the road by the movement direction of the track in the window, and projecting all GPS track points in the window in the cross section direction of the road to obtain a numberAccording to the projection distribution in the road cross section direction, the position of the center line of each lane is obtained as the peak value mu of the Gaussian component of the lane j The distribution condition of the track observation points on each lane corresponds to the standard deviation sigma of the distribution of each Gaussian component data j Determining the weight of each Gaussian component distribution according to the proportion of the track number in each lane, and constructing a probability density function of the Gaussian mixture model:
wherein k is the number of Gaussian components, j=1, 2, …, k, w j Weights for the j-th gaussian component distribution, wherein,
w 1 +w 2 +…+w k =1。
10. the method of claim 9, wherein the evaluation function of lane number selection is optimized according to the formula: Evaluating the fitting condition of a Gaussian mixture model, preventing overfitting, restraining the lane width value, controlling the distance between peak distribution of each Gaussian component, taking the distance as a punishment item for preventing overfitting when likelihood function is increased along with the mixing score, and selecting the Gaussian mixture model with the minimum evaluation function value as an optimal model, wherein lambda is a regularization parameter and delta h is calculated by the method i Representing the average distance between the centerlines of adjacent lanes.
11. A road segment trajectory-based lane geometry information extraction system, comprising: the system comprises a lane track clustering module, a lane center line fitting module, a basic lane line extraction module, a lane change scene detection module, an added lane line extraction module and a complete lane line construction module, wherein the lane track clustering module is used for carrying out track cluster analysis according to a track subset of the same road section, dividing a sub-track set of the same road section into different lanes and extracting track clusters belonging to each lane; the lane center line fitting module is used for extracting the geometric center of the track from the track subsets in the same cluster in a non-parametric curve mode and fitting the candidate lane center line; the basic lane line extraction module is used for obtaining basic lane lines which are mutually disjoint according to the candidate lane lines; the lane change scene detection module is used for judging whether the road section belongs to a lane change scene according to the track data of the road section; the newly added lane line extraction module is used for extracting a candidate lane center line based on newly added Gaussian components at two ends of the road section to serve as a newly added lane center line; and the complete lane line construction module is used for constructing the geometric information and the topological relation of the lane under the lane change scene according to the extracted basic lane lines and the newly added lane lines.
12. The system of claim 11, wherein the basic lane line extraction module obtains basic lane lines based on a maximum clique algorithm of the graph, regards lane center lines as nodes in the graph, simultaneously, the node weights record geometric features of the lane lines, the edge weights record intersection relationship features among the lane lines, a candidate lane line intersection relationship graph is constructed by using the lane center lines fitted by each road section, each candidate lane line is defined as a node of the graph, edges between the nodes represent intersection relationships among lanes, edges are connected if two lane lines do not intersect, an undirected graph is constructed, and a set of lane center lines which do not intersect each other is extracted through the undirected graph as the basic lane lines.
13. The system of claim 12, wherein each candidate lane line is defined as a vertex set V {1, …, n } of the graph, an edge set between nodesRepresenting the intersection relationship between lanes, if two lane lines are not intersected, the edges between the nodes represented by the two lane lines are connected, and G (V, E) is constructed as a vertex set V {1, …, n } and an edge setIf the subset U E V and for any two of the vertices U, V E U, there is (U, V) E, then U is the complete subgraph of G and the largest complete subgraph is the largest clique of the graph.
14. The system of claim 13, wherein when there are a plurality of maximum cliques, the set of the largest sum of weights of the selected lane-line nodes is taken as the basic lane-line of the non-lane-change, according to the formula:
and calculating a node weight, namely enabling the lane line weight with smaller track yaw angle average change to be larger by negative logarithmic function transformation, wherein the track weight with larger track yaw angle average change is smaller, and delta yaw represents the track yaw angle average change.
15. The system according to any one of claims 11-14, wherein a priori gaussian mixture model is used to identify the number of lanes at both ends of a road segment, the maximum coverage width of the track on the road surface is taken as the side length of the window, the start point of the road segment is taken as the start point of the road segment, the track observation point in the window is detected, the track observation point is gradually moved along the main direction of the road until the end point of the road segment, the main direction of the road and the cross-sectional direction of the road perpendicular to the main direction of the road are identified through the movement direction of the track in the window, all the GPS track points in the window are projected in the cross-sectional direction of the road, the projection distribution of the data in the cross-sectional direction of the road is obtained, and the position of the center line of each lane is obtained as the peak μ of the gaussian component of the lane j The distribution condition of the track observation points on each lane corresponds to the standard deviation sigma of the distribution of each Gaussian component data j Determining the weight of each Gaussian component distribution according to the proportion of the track number in each lane, and constructing a probability density function of the Gaussian mixture model:where k is the number of gaussian components, j=1, 2, …, k,w j weights distributed for the jth gaussian component, where w 1 +w 2 +…+w k =1。
16. The system of claim 15, wherein the evaluation function selected by the number of lanes is optimized,evaluating the fitting condition of a Gaussian mixture model, preventing overfitting, restraining the lane width value, controlling the distance between peak distribution of each Gaussian component, taking the distance as a punishment item for preventing overfitting when likelihood function is increased along with the mixing score, and selecting the Gaussian mixture model with the minimum evaluation function value as an optimal model, wherein lambda is a regularization parameter and delta h is calculated by the method i Representing the average distance between the centerlines of adjacent lanes.
17. An electronic device, comprising: a processor; and a memory storing a program, characterized in that, wherein the program comprises instructions for, the instructions, when executed by the processor, cause the processor to perform the trajectory-based lane geometry information extraction method according to any one of claims 1-10.
18. A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are for causing the computer to perform the track-based lane geometry information extraction method according to any one of claims 1-10.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310173602.6A CN116071722A (en) | 2023-02-28 | 2023-02-28 | Lane geometric information extraction method, system, equipment and medium based on road section track |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310173602.6A CN116071722A (en) | 2023-02-28 | 2023-02-28 | Lane geometric information extraction method, system, equipment and medium based on road section track |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116071722A true CN116071722A (en) | 2023-05-05 |
Family
ID=86183687
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310173602.6A Pending CN116071722A (en) | 2023-02-28 | 2023-02-28 | Lane geometric information extraction method, system, equipment and medium based on road section track |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116071722A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116504068A (en) * | 2023-06-26 | 2023-07-28 | 创辉达设计股份有限公司江苏分公司 | Statistical method, device, computer equipment and storage medium for lane-level traffic flow |
CN116805015A (en) * | 2023-08-25 | 2023-09-26 | 山东黄河三角洲国家级自然保护区管理委员会 | Bird migration route graph theory modeling method based on GPS tracking data |
CN118094476A (en) * | 2024-04-24 | 2024-05-28 | 吉林大学 | Method for judging driving vehicle lane by fusing satellite remote sensing image and digital road network |
-
2023
- 2023-02-28 CN CN202310173602.6A patent/CN116071722A/en active Pending
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116504068A (en) * | 2023-06-26 | 2023-07-28 | 创辉达设计股份有限公司江苏分公司 | Statistical method, device, computer equipment and storage medium for lane-level traffic flow |
CN116805015A (en) * | 2023-08-25 | 2023-09-26 | 山东黄河三角洲国家级自然保护区管理委员会 | Bird migration route graph theory modeling method based on GPS tracking data |
CN116805015B (en) * | 2023-08-25 | 2023-11-28 | 中国林业科学研究院森林生态环境与自然保护研究所(国家林业和草原局世界自然遗产保护研究中心) | Bird migration route graph theory modeling method based on GPS tracking data |
CN118094476A (en) * | 2024-04-24 | 2024-05-28 | 吉林大学 | Method for judging driving vehicle lane by fusing satellite remote sensing image and digital road network |
CN118094476B (en) * | 2024-04-24 | 2024-06-28 | 吉林大学 | Method for judging driving vehicle lane by fusing satellite remote sensing image and digital road network |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109241069B (en) | Road network rapid updating method and system based on track adaptive clustering | |
CN116071722A (en) | Lane geometric information extraction method, system, equipment and medium based on road section track | |
CN109256028B (en) | Method for automatically generating unmanned high-precision road network | |
CN106227726B (en) | Path extraction method based on vehicle track data | |
CN108961758B (en) | Road junction widening lane detection method based on gradient lifting decision tree | |
Uduwaragoda et al. | Generating lane level road data from vehicle trajectories using kernel density estimation | |
CN112184736B (en) | Multi-plane extraction method based on European clustering | |
CN109101743B (en) | Method for constructing high-precision road network model | |
CN105957342A (en) | Lane-level road mapping method and system based on crowdsourcing space-time big data | |
CN111539454A (en) | Vehicle track clustering method and system based on meta-learning | |
Qiu et al. | Automatic extraction of road networks from GPS traces | |
Hashemi | Automatic inference of road and pedestrian networks from spatial-temporal trajectories | |
Ezzat et al. | Building road segments and detecting turns from gps tracks | |
Revilloud et al. | A lane marker estimation method for improving lane detection | |
CN111123333B (en) | Vehicle track positioning method fusing bayonet and GPS data | |
CN115081505A (en) | Pedestrian network incremental generation method based on walking track data | |
CN115346183A (en) | Lane line detection method, terminal and storage medium | |
Tang et al. | A novel method for road intersection construction from vehicle trajectory data | |
Fu et al. | Density adaptive approach for generating road network from GPS trajectories | |
Dørum | Deriving double-digitized road network geometry from probe data | |
CN116363856A (en) | Method and device for generating road intersection topological structure | |
CN112347983B (en) | Lane line detection processing method, lane line detection processing device, computer equipment and storage medium | |
CN117664168A (en) | Park vehicle navigation path planning method, medium and system | |
CN116628531A (en) | Crowd-sourced map road object element clustering method, system and storage medium | |
Arman et al. | Road centreline and lane reconstruction from pervasive GPS tracking on motorways |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |