A Lane Line Classification Fitting Method and System

Abstract

This invention provides a lane line classification and fitting method and system. When clustering lane lines, it not only considers the spatial distance of lane line points, but also calculates the angle between the direction vectors of the points, resulting in more accurate classification results. It proposes a shell-based clustering algorithm, which further clusters within each shell to obtain the centroid, and uses the Hungarian algorithm to perform matching from the outside in. It classifies and processes various matching cases, achieving good results in complex scenarios such as divergence and merging.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of lane line processing, and more particularly, to a lane line classification fitting method and system. BACKGROUND

[0002] In the field of automatic driving, through the extensive arrangement of low-cost crowdsourcing collection vehicles, high-frequency collection of high-freshness data, the device cost is low, the collection accuracy is poor, and the collection data volume is large. The existing clustering algorithm has difficulty in processing the large data volume, low efficiency, and complex real scene problems when clustering the lane lines generated by the crowdsourcing data. SUMMARY

[0003] The present application provides a lane line classification fitting method and system to address the difficulty of conventional algorithms in dealing with large data volume, low efficiency, and complex real scene problems.

[0004] According to a first aspect of the present application, a lane line classification fitting method is provided, comprising:

[0005] Based on the crowdsourcing collection data, lane line shape point sequences are obtained, and an improved DBSCAN clustering algorithm is used to cluster the lane line shape points in the lane line shape point sequences to obtain a plurality of clusters after clustering;

[0006] For any cluster, a shell structure is established, the shell structure is composed of a group of uniformly distributed concentric circles, the center of the shell structure is the lane line shape point closest to the center of the any cluster, wherein the adjacent two concentric circles form a circular ring, and all lane line shape points in the any cluster are distributed to the corresponding circular ring;

[0007] According to the matching order from the outer ring to the inner ring, the lane line shape points in the adjacent two circular rings are matched to obtain the matching relationship of the lane line shape points in the any cluster;

[0008] Based on the matching relationship, all lane line shape points in the any cluster are classified;

[0009] According to all lane line shape points of the same classification, fitting is performed to obtain the lane line of each classification.

[0010] On the basis of the above technical solution, the present application can also be improved as follows.

[0011] Optionally, the improved DBSCAN clustering algorithm is used to cluster the lane line shape points in the lane line shape point sequences to obtain a plurality of clusters after clustering, comprising:

[0012] According to the spatial coordinate vector and the direction vector of any two lane line shape points, the weighted distance d between any two lane line shape points p1 and p2 is calculated;

[0013] If the weighted distance d is less than a set threshold, the two lane line points p1 and p2 belong to the same class, otherwise, the two lane line points p1 and p2 are different classes.

[0014] Optionally, the weighted distance d between any two lane line points p1 and p2 is calculated according to the spatial coordinate vector and the direction vector of any two lane line points, comprising:

[0015]

[0016] Wherein, t1, t2, t3 are weight parameters of the transverse, longitudinal and height directions respectively, d1 is the transverse distance between the two lane line points p1 and p2, d2 is the longitudinal distance between the two lane line points p1 and p2, and d3 is the height coordinate difference between the two lane line points p1 and p2, wherein:

[0017] d1 = (e dist *sinα+e dist *sinβ) / 2;

[0018] d2 = (e dist *cosα+e dist *cosβ) / 2;

[0019] Wherein, e dist is the Euclidean distance between the two lane line points p1 and p2, Vector p1p2 = p2-p1, p1 and p2 represent the spatial coordinate vectors of the two lane line points, and v1 and v2 represent the direction vectors of the two lane line points, which are calculated by the gradient calculation formula.

[0020] Optionally, for any cluster, a shell structure is established, comprising:

[0021] For any cluster, based on the spatial coordinate vectors of all lane line points in the cluster, the cluster center coordinates are solved, and the lane line point closest to the cluster center coordinates in the cluster is selected as the center of the shell structure;

[0022] A step size is set to construct a plurality of concentric circles with the center as the center, and the adjacent two concentric circles form a circular ring;

[0023] Based on the distance from each lane line point in the cluster to the center, all lane line points are distributed into the corresponding circular ring.

[0024] Optionally, the lane line points in the adjacent two circular rings are matched in the matching order from the outer ring to the inner ring to obtain the matching relationship of the lane line points in the cluster.

[0025] For lane alignment points within the same ring, clustering is performed based on the improved DBSCAN clustering algorithm to obtain at least one sub-cluster, and the centroid of each sub-cluster is calculated to obtain at least one centroid of each ring.

[0026] Optionally, the step of matching lane alignment points in two adjacent rings according to the matching order from the outer ring to the inner ring to obtain the matching relationship of lane alignment points within any cluster includes:

[0027] Match the lane alignment points in each pair of adjacent rings to form at least one matching chain;

[0028] Accordingly, based on the matching relationship, all lane alignment points within any cluster are classified, including:

[0029] Based on the number of matching chains, all lane alignment points within any given cluster are classified.

[0030] Optionally, the step of fitting the lane lines based on all lane line points of the same category to obtain lane lines for each category includes:

[0031] Based on all lane alignment points of the same category, a cubic polynomial function is used for fitting to obtain the corresponding lane lines. The cubic polynomial function is expressed in the form y = (a0 + a1*x + a2*x). 2 +a3*x 3 ), where a0, a1, a2, a3 are coefficients, a is the independent variable, and y is the dependent variable.

[0032] According to a second aspect of the present invention, a lane line classification fitting system is provided, comprising:

[0033] The clustering module is used to obtain lane alignment point sequences based on crowdsourced data, and to cluster the lane alignment points in the lane alignment point sequences based on the improved DBSCAN clustering algorithm to obtain multiple clusters.

[0034] A construction module is used to construct a shell structure for any cluster. The shell structure consists of a set of uniformly distributed concentric circles. The center of the shell structure is the lane alignment point closest to the center of the cluster. Two adjacent concentric circles form a ring, and all lane alignment points in the cluster are assigned to the corresponding rings.

[0035] The matching module is used to match lane alignment points in two adjacent rings according to the matching order from the outer ring to the inner ring, so as to obtain the matching relationship of lane alignment points in any cluster.

[0036] The classification module is used to classify all lane alignment points within any cluster based on the matching relationship;

[0037] The fitting module is used to fit all lane line points of the same category to obtain the lane lines for each category.

[0038] According to a third aspect of the present invention, an electronic device is provided, including a memory and a processor, the processor being configured to implement a lane line classification fitting method when executing a computer management program stored in the memory.

[0039] According to a fourth aspect of the present invention, a computer-readable storage medium is provided having a computer management class program stored thereon, which, when executed by a processor, implements the steps of a lane line classification fitting method.

[0040] This invention provides a lane line classification and fitting method and system. When clustering lane lines, it not only considers the spatial distance of lane line points, but also calculates the angle between the direction vectors of the points, resulting in more accurate classification results. It proposes a shell-based clustering algorithm, which further clusters within each shell to obtain centroids, and uses the Hungarian algorithm to perform matching from the outside in. It classifies and processes various matching cases, achieving good results in complex scenarios such as divergence and merging. Attached Figure Description

[0041] Figure 1 A flowchart of a lane line classification and fitting method provided by the present invention;

[0042] Figure 2 A schematic diagram showing the partial distribution of the shell structure and lane alignment points within the circular ring;

[0043] Figure 3 A schematic diagram of the center of mass inside the inner shell of a circular ring;

[0044] Figure 4 A diagram illustrating lane marking classification;

[0045] Figure 5 This is a schematic diagram of the structure of a lane line classification and fitting system provided by the present invention;

[0046] Figure 6 A schematic diagram of the hardware structure of a possible electronic device provided by the present invention;

[0047] Figure 7 This is a schematic diagram of the hardware structure of a possible computer-readable storage medium provided by the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. In addition, the technical features of the various embodiments or individual embodiments provided by the present invention can be arbitrarily combined with each other to form feasible technical solutions. Such combinations are not constrained by the order of steps and / or structural composition patterns, but must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

[0049] Figure 1 A flowchart of a lane line classification and fitting method provided by the present invention is shown below. Figure 1 As shown, the method mainly includes the following steps:

[0050] S1. Obtain a lane alignment point sequence based on crowdsourced data collection, and cluster the lane alignment points in the lane alignment point sequence based on the improved DBSCAN clustering algorithm to obtain multiple clusters.

[0051] Understandably, when clustering and classifying lane alignment point data collected through crowdsourcing, the first step is to use point clustering to cluster all lane alignment points. Specifically, the point clustering method mainly uses the improved DBSCAN clustering algorithm to cluster lane alignment points in the lane alignment point sequence.

[0052] As an example, the improved DBSCAN clustering algorithm is used to cluster the lane alignment points in the lane alignment point sequence to obtain multiple clusters. This includes: calculating the weighted distance d between any two lane alignment points p1 and p2 based on their spatial coordinate vectors and direction vectors; if the weighted distance d is less than a set threshold, then the two lane alignment points p1 and p2 belong to the same class; otherwise, the two lane alignment points p1 and p2 belong to different classes.

[0053] Understandably, the classification objects, such as lane lines and guardrails, are classified using the current batch of crowdsourced data and the base map data fused from historical crowdsourced data. Each element consists of a time-series label p = {N1, N2, N3, ..., N}. n Each point is represented by (x, y, z) which represents its horizontal coordinate, vertical coordinate, and elevation coordinate. The direction vector v = {v1, v2, v3, ..., v} for each element can be obtained from the gradient calculation formula. n}

[0054] For any two lane alignment points p1 and p2, with direction vectors v1 and v2, calculate the weighted distance d between these two points based on their spatial coordinate vectors and direction vectors. This includes:

[0055]

[0056] Where t1, t2, and t3 are the weight parameters for the lateral, longitudinal, and elevation directions, respectively, and can be taken as 0.6m, 2m, and 4m. d1 is the lateral distance between two lane alignment points p1 and p2, d2 is the longitudinal distance between two lane alignment points p1 and p2, and d3 is the difference in elevation coordinates between two lane alignment points p1 and p2.

[0057] d1=(e dist *sinα+e dist *sinβ) / 2;

[0058] d2=(e dist *cosα+e dist *cosβ) / 2;

[0059] Among them, e dist Let p1 and p2 be the Euclidean distance between two lane alignment points p1 and p2. Vector p1p2 = p2 - p1, where p1 and p2 represent the spatial coordinate vectors of two lane alignment points, and v1 and v2 represent the direction vectors of the two lane alignment points, calculated using the gradient calculation formula.

[0060] The weighted distance d(p1,p2) between shape points p1 and p2 is calculated according to the above distance calculation formula. If d(p1,p2)<1, then shape points p1 and p2 are considered to be of the same type; otherwise, they are considered to be of different types. Based on this classification method, the final clustering result is obtained, resulting in multiple clusters after clustering all lane line shape points. Each cluster includes lane line shape points of the same type.

[0061] S2. For any cluster, establish a shell structure, which consists of a set of uniformly distributed concentric circles. The center of the shell structure is the lane alignment point closest to the center of the cluster. Two adjacent concentric circles form a ring, and all lane alignment points in the cluster are assigned to the corresponding rings.

[0062] Understandably, by clustering all lane alignment points using the above point clustering method, multiple clusters (also called sets) are obtained, c = {C1, C2, C3...C...}. n Each point in the set has coordinate and orientation information.

[0063] For each lane alignment point in a cluster, a shell structure is constructed. The specific construction process is as follows: For any cluster, based on the spatial coordinate vectors of all lane alignment points in the cluster, the coordinates of the cluster center are calculated, and the lane alignment point within the cluster closest to the cluster center coordinates is selected as the center of the shell structure. A step size is set, and several concentric circles are constructed with the center as the center, with two adjacent concentric circles forming an annulus. Based on the distance from each lane alignment point in the cluster to the center of the circle, all lane alignment points are assigned to the corresponding annulus. A schematic diagram of the constructed shell structure and the distribution of lane alignment points can be found in [reference needed]. Figure 2 For example, with a step size of 1m, construct several concentric circles, numbered 0, 1, 2, 3, ..., m from the inside out. A ring is formed between concentric circles numbered 0 and 0, and a ring is also formed between concentric circles numbered (m-1) and m, resulting in a total of (m-1) rings. For instance, the rings from the innermost to the outermost layer are numbered R1, R2, ..., R... m-1 .

[0064] S3. According to the matching order from the outer ring to the inner ring, the lane alignment points in two adjacent rings are matched to obtain the matching relationship of the lane alignment points in any cluster.

[0065] As an example, the process of matching lane alignment points in two adjacent rings according to the matching order from the outer ring to the inner ring to obtain the matching relationship of lane alignment points in any cluster further includes: for lane alignment points in the same ring, clustering based on the improved DBSCAN clustering algorithm to obtain at least one sub-cluster, and calculating the centroid point within the shell of each sub-cluster to obtain at least one centroid point within the shell of each ring.

[0066] Understandably, after assigning all lane alignment points within a cluster to the corresponding rings in step S2, for lane alignment points within the same ring, the point clustering method in step S1 is used for clustering to form at least one sub-cluster within each ring. The average value is calculated based on the spatial coordinate vector and direction vector of the lane alignment points within each sub-cluster, and this average is used as the centroid of each sub-cluster, resulting in multiple centroids within each ring. (See also...) Figure 3 .

[0067] As an example, the step of matching lane alignment points in two adjacent rings according to the matching order from the outer ring to the inner ring to obtain the matching relationship of lane alignment points in any cluster includes: matching lane alignment points in every two adjacent rings to form at least one matching chain; correspondingly, classifying all lane alignment points in any cluster based on the matching relationship includes: classifying all lane alignment points in any cluster based on the number of matching chains.

[0068] Understandably, when clustering lane line points within the shell structure, the centroids within adjacent rings are matched according to the matching order from the outer ring to the inner ring. For example, the aforementioned rings from the innermost to the outermost layer are numbered R1, R2, ..., R... m-1 So first match R m-1 Multiple intrashell centroids within the annulus and R m-2 Multiple centroids within the shell of the annulus, then matched with R m-2 Multiple intrashell centroids within the annulus and R m-3 Multiple intrashell centroids within the annulus are used in sequence, and finally the intrashell centroids within the R2 annulus and the intrashell centroids within the R1 annulus are matched.

[0069] In the process of matching the centroids within the shells of two adjacent rings, the centroid of the outer ring is defined as the source point, and the centroid of the inner ring as the target point. The problem is transformed into an optimal allocation problem among multiple source points and multiple target points. First, the Hungarian algorithm is used to obtain a one-to-one solution. The loss function is calculated at this point, and the calculation of the loss function borrows from the weighted distance calculation method used in point clustering in step 1, setting the loss to 1 when the source and target points are not assigned. After using the Hungarian algorithm, the source and target points that are not matched are reassigned. Each unassigned source point is traversed, and the minimum loss matching point among the target points is found. If the loss after assignment is less than that before assignment, the matching pair is added to the set. The matching solution is continuously adjusted to minimize the loss function. This is the optimal matching solution, and the matching pair of source and target points is added to the set.

[0070] The matching results between the source point and the target point are divided into five types, as shown in Table 1 below:

[0071] Table 1. Matching Results of Source and Target Points

[0072]

[0073] The following is an explanation of Table 1:

[0074] For "No predecessor" connections, it means that a new punctuation mark needs to be added.

[0075] For a "one-to-one connection", the existing line needs to be extended to add a new set of points.

[0076] For a "split" connection, the corresponding line segment needs to be terminated, and a new node and N new markings need to be added. This corresponds to the handling of lane line "branching" scenarios.

[0077] For a "merge" connection, create a node, terminate N existing lane markings, and generate a new lane marking. This corresponds to the handling of lane marking "merging" scenarios.

[0078] A "No successor" connection indicates that the current iteration needs to be terminated and the next iteration needs to be started. This situation is caused by noise or other reasons that result in a missing cluster center in the shell.

[0079] The matching methods described above yield the matching relationships of the centroids within the shell. These relationships may be mismatched, or one source point may correspond to multiple target points, multiple source points to one target point, or one source point to one target point. Based on these matching relationships and the ordered pairs of matching points, at least one matching chain can be formed. Specifically, if one source point corresponds to multiple target points, the corresponding scenario is a road divergence; if multiple source points correspond to one target point, the corresponding scenario is a road merging. In this case, multiple matching chains may be formed.

[0080] S4. Based on the matching relationship, classify all lane alignment points within any cluster.

[0081] Understandably, see Figure 4 Through matching, two matching chains are ultimately formed, so the lane alignment points can be divided into two categories, such as main roads and branch roads.

[0082] S5: Fit the lane lines of each category to all lane line points of the same category.

[0083] Understandably, based on the matching point pairs of each category, several ordered point sets can be obtained, each point set represented as g = {G1, G2, G3, ..., G...} n The function is fitted using a cubic polynomial, and its form is y = (a0 + a1x + a2x). 2 +a3*x 3 The root mean square error is used to evaluate the results of the fitting function. The unknown parameters a0, a1, a2, a3 are obtained to minimize the sum of the loss between the predicted and target values ​​of the sample points, and the corresponding lane lines are fitted.

[0084] Figure 5A structural diagram of a lane line classification fitting system provided in an embodiment of the present invention is shown below. Figure 5 As shown, a lane line classification and fitting system includes a clustering module 501, a construction module 502, a matching module 503, a classification module 504, and a fitting module 505, wherein:

[0085] Clustering module 501 is used to obtain lane alignment point sequences based on crowdsourced data, and to cluster the lane alignment points in the lane alignment point sequences based on the improved DBSCAN clustering algorithm to obtain multiple clusters.

[0086] The construction module 502 is used to construct a shell structure for any cluster. The shell structure is composed of a set of uniformly distributed concentric circles. The center of the shell structure is the lane line point closest to the center of the cluster. Two adjacent concentric circles form a ring, and all lane line points in the cluster are assigned to the corresponding rings.

[0087] The matching module 503 is used to match lane alignment points in two adjacent rings according to the matching order from the outer ring to the inner ring, so as to obtain the matching relationship of lane alignment points in any cluster.

[0088] The classification module 504 is used to classify all lane line points within any cluster based on the matching relationship;

[0089] The fitting module 505 is used to fit all lane line points of the same category to obtain the lane lines for each category.

[0090] It is understood that the lane line classification and fitting system provided by the present invention corresponds to the lane line classification and fitting methods provided in the foregoing embodiments. The relevant technical features of the lane line classification and fitting system can be referred to the relevant technical features of the lane line classification and fitting methods, and will not be repeated here.

[0091] Please see Figure 6 , Figure 6 A schematic diagram illustrating an embodiment of the electronic device provided in this invention. For example... Figure 6 As shown, an embodiment of the present invention provides an electronic device 600, including a memory 610, a processor 620, and a computer program 611 stored in the memory 610 and executable on the processor 620. When the processor 620 executes the computer program 611, it implements the steps of a lane line classification fitting method.

[0092] Please see Figure 7 , Figure 7 This is a schematic diagram illustrating an embodiment of a computer-readable storage medium provided by the present invention. (See diagram below.) Figure 7As shown, this embodiment provides a computer-readable storage medium 700, on which a computer program 711 is stored. When the computer program 711 is executed by a processor, it implements the steps of a lane line classification fitting method.

[0093] This invention provides a lane line classification and fitting method and system. When clustering lane lines, it considers not only the spatial distance between lane line points but also calculates the angle between the direction vectors of these points, resulting in more accurate classification. A shell-based clustering algorithm is proposed, where centroids are obtained by further clustering within each shell, and the Hungarian algorithm is used for outward-to-inward matching. Various matching scenarios are classified and processed, achieving good results in complex scenarios such as divergence and merging.

[0094] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0095] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0097] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0098] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0099] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0100] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A lane line classification fitting method, characterized by, The method comprises the following steps: Based on the crowd-sourced data acquisition, a lane line point sequence is obtained, and an improved DBSCAN clustering algorithm is used to cluster lane line points in the lane line point sequence to obtain a plurality of clusters after clustering; For any cluster, a shell structure is established, the shell structure is composed of a group of uniformly distributed concentric circles, and the center of the shell structure is the lane line point closest to the center of the cluster, wherein two adjacent concentric circles form a circular ring, and all lane line points in the cluster are distributed into the corresponding circular ring; According to the matching order from the outer ring to the inner ring, the lane line points in the adjacent two circular rings are matched to obtain the matching relationship of the lane line points in the cluster. Based on the matching relationship, all lane line points in the cluster are classified; According to all lane line points of the same classification, the lane line of each classification is obtained by fitting.

2. The lane line classification fitting method of claim 1, wherein, The improved DBSCAN clustering algorithm is used to cluster lane line points in the lane line point sequence to obtain a plurality of clusters after clustering, which comprises the following steps: According to the spatial coordinate vector and the direction vector of any two lane line points, the weighted distance d between any two lane line points p1 and p2 is calculated; If the weighted distance d is less than a set threshold, the two lane line points p1 and p2 belong to the same class, otherwise, the two lane line points p1 and p2 belong to different classes.

3. The lane line classification fitting method of claim 2, wherein, According to the spatial coordinate vector and the direction vector of any two lane line points, the weighted distance d between any two lane line points p1 and p2 is calculated, which comprises the following steps: Wherein, t1, t2, t3 are weight parameters in the transverse, longitudinal and height directions respectively, d1 is the transverse distance between two lane line points p1 and p2, d2 is the longitudinal distance between two lane line points p1 and p2, and d3 is the difference between the elevation coordinates of two lane line points p1 and p2. d1 = (e dist *sinα+e dist *sinβ) / 2; d2 = (e dist *cosα+e dist *cosβ) / 2; where e dist is the Euclidean distance between two lane shape points p1 and p2, Vector p1p2 = p2 - p1, p1 and p2 represent the spatial coordinate vectors of two lane shape points, v1 and v2 represent the direction vectors of two lane shape points, which are calculated by the gradient calculation formula.

4. The lane line classification fitting method of claim 1, wherein, For any cluster, a shell structure is established, which comprises the following steps: For any cluster, the center coordinates of the cluster are solved based on the spatial coordinate vector of all lane line points in the cluster, and the lane line point closest to the center coordinates of the cluster in the cluster is selected as the center of the shell structure; Set the step size, and construct a plurality of concentric circles with the center as the center, and the adjacent two concentric circles form a circular ring; Based on the distance from each lane line point in the cluster to the center, all lane line points are distributed into the corresponding circular ring.

5. The lane line classification fitting method of claim 1, wherein, The matching relationship of the lane line points in the cluster is obtained by matching the lane line points in the adjacent two circular rings in the matching order from the outer ring to the inner ring, which further comprises the following steps: For the lane line points in the same circular ring, the improved DBSCAN clustering algorithm is used for clustering to obtain at least one sub-cluster, and the in-shell centroid point of each sub-cluster is calculated to obtain at least one in-shell centroid point in each circular ring.

6. The lane line classification fitting method of claim 5, wherein, The matching relationship of the lane line points in the cluster is obtained by matching the lane line points in the adjacent two circular rings in the matching order from the outer ring to the inner ring, which comprises the following steps: matching lane linear points in each of the two adjacent annuli to form at least one matching chain; correspondingly, based on the matching relationship, classifying all lane linear points in the any cluster, including: based on the number of matching chains, classifying all lane linear points in the any cluster.

7. The lane line classification fitting method of claim 1, wherein, fitting all lane linear points of the same classification to obtain a lane line of each classification, including: According to all lane line shape points of the same classification, a cubic polynomial function is used for fitting to obtain a corresponding lane line, wherein the cubic polynomial function is in the form of y=(a0+a1*x+a2*x 2 +a3*x 3 ), a0, a1, a2, and a3 are coefficients, a is an independent variable, and y is a dependent variable.

8. A lane line classification fitting system, comprising: including: a clustering module, configured to obtain a lane linear point sequence based on crowd-sourced collection data, and cluster lane linear points in the lane linear point sequence based on an improved DBSCAN clustering algorithm to obtain a plurality of clusters after clustering; a construction module, configured to, for any cluster, construct a shell structure composed of a group of uniformly distributed concentric circles, a center of the shell structure being a lane linear point closest to a center of the any cluster, wherein two adjacent concentric circles form an annulus, and all lane linear points in the any cluster are distributed into corresponding annuli; a matching module, configured to match lane linear points in two adjacent annuli in a matching order from an outer annulus to an inner annulus to obtain a matching relationship of lane linear points in the any cluster; a classification module, configured to classify all lane linear points in the any cluster based on the matching relationship; a fitting module, configured to fit all lane linear points of the same classification to obtain a lane line of each classification.

9. An electronic device, comprising: including a memory and a processor, the processor being configured to implement steps of the lane line classification and fitting method according to any one of claims 1-7 when executing a computer management program stored in the memory.

10. A computer-readable storage medium, characterized in that, having a computer management program stored thereon, the computer management program being configured to implement steps of the lane line classification and fitting method according to any one of claims 1-7 when executed by a processor.

Images