Road network data processing method, system, terminal and storage medium

Abstract

The application discloses a road network data processing method and system, a terminal and a storage medium, relates to the technical field of data processing, and comprises the following steps: acquiring to-be-processed road network data corresponding to a road network model, so as to extract road section length and road section shape data of each road section and determine the density of reference points in a unit length; if the reference point density in the unit length does not meet a density control condition, a control curve is determined according to the coordinates of two-dimensional reference points in the road section shape data, target reference points are inserted into the road network model according to the coordinates of the two-dimensional reference points and the control curve, and a reference point sequence is obtained; acquiring a street view image matched with the road network model, so as to determine a target road section with a missing traffic channel in the road network model; for the target road section, supplementary reference points corresponding to the missing traffic channel are generated according to the street view image and the reference point sequence, and the missing traffic channel is completed in the road network model according to the supplementary reference points. In this way, the accuracy of the road network model can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and particularly relates to a road network data processing method and system, a terminal and a storage medium. BACKGROUND

[0002] With the development of science and technology, the application of traffic simulation is more and more widely. In the process of realizing traffic simulation, a corresponding road network model needs to be constructed.

[0003] In the prior art, in the construction process of the simulation road network model, the data in the Extensible Markup Language (XML) format is obtained by converting the OpenStreetMap (OSM), and then the construction of the simulation road network model is directly based on the obtained XML data. The problem of the prior art is that in the process of obtaining the XML data based on the OSM conversion, the problem of the loss of the reference point is prone to occur, thereby causing the simulation road network model to have the problem of the sudden change of the road segment line shape, which is not conducive to improving the accuracy of the simulation road network model.

[0004] Therefore, the related technology needs to be improved and developed. SUMMARY

[0005] The main purpose of the present application is to provide a road network data processing method and system, a terminal and a storage medium, which aims to solve the technical problem that the simulation road network model constructed directly based on the XML data is not conducive to improving the accuracy of the model due to the loss of the reference point in the related technology.

[0006] In order to achieve the above purpose, the first aspect of the present application provides a road network data processing method, wherein the road network data processing method comprises:

[0007] obtaining road network model corresponding to the to-be-processed road network data, extracting the road segment length and road segment shape data corresponding to each road segment from the to-be-processed road network data, wherein the road segment shape data comprises the coordinates of the two-dimensional reference point corresponding to the road segment;

[0008] determining the reference point density in the unit length corresponding to each road segment according to the road segment length and the road segment shape data;

[0009] if the reference point density in the unit length does not meet the preset density control condition, determining a control curve according to the coordinates of the two-dimensional reference point in the road segment shape data, inserting a target reference point in the road network model according to the coordinates of the two-dimensional reference point and the control curve, and obtaining a reference point sequence composed of the two-dimensional reference point and the target reference point;

[0010] obtaining a street view image matched with the road network model;

[0011] determining a target road segment in which a missing traffic lane exists in the road network model according to the street view images;

[0012] generating a supplementary reference point corresponding to the missing traffic lane according to the street view images and the reference point sequence, and completing the missing traffic lane in the road network model according to the supplementary reference point for the target road segment.

[0013] Optionally, the determining of the reference point density per unit length of each road segment according to the road segment length and the road segment shape data comprises:

[0014] determining the number of two-dimensional reference points corresponding to each road segment according to the road segment shape data;

[0015] determining the reference point density per unit length of each road segment according to the ratio of the number of two-dimensional reference points corresponding to the road segment to the length of the road segment.

[0016] Optionally, the control curve is a Bezier curve.

[0017] If the reference point density per unit length does not satisfy the preset density control condition, determining a control curve according to the coordinates of the two-dimensional reference points in the road segment shape data, inserting a target reference point in the road network model according to the coordinates of the two-dimensional reference points and the control curve, and obtaining a reference point sequence composed of the two-dimensional reference points and the target reference point, comprising:

[0018] constructing a reference point sequence according to the coordinates of the two-dimensional reference points in the road segment shape data;

[0019] determining whether the reference point density per unit length is greater than a preset minimum density;

[0020] If the reference point density per unit length is not greater than the minimum density, determining two candidate reference points adjacent to each other and having the maximum distance between adjacent points from the reference point sequence according to the coordinates of the reference points in the reference point sequence;

[0021] determining a Bezier curve corresponding to the candidate reference points according to the coordinates of the candidate reference points;

[0022] inserting a target reference point between the candidate reference points according to the Bezier curve corresponding to the candidate reference points, and updating the reference point sequence;

[0023] updating the density of the fiducial points in the unit length according to the number of the fiducial points in the fiducial point sequence, and returning to the step of judging whether the density of the fiducial points in the unit length is greater than the preset minimum density until the density of the fiducial points in the unit length is greater than the minimum density.

[0024] Optionally, the determining the Bezier curve corresponding to the candidate fiducial point according to the coordinates of the candidate fiducial point comprises:

[0025] If the candidate fiducial point comprises the first fiducial point or the last fiducial point in the fiducial point sequence, a second-order Bezier curve corresponding to the candidate fiducial point is determined according to the coordinates of the candidate fiducial point;

[0026] Otherwise, a third-order Bezier curve corresponding to the candidate fiducial point is determined according to the coordinates of the candidate fiducial point.

[0027] Optionally, the inserting a target fiducial point between the candidate fiducial points according to the Bezier curve corresponding to the candidate fiducial point and updating the fiducial point sequence comprises:

[0028] obtaining a preset position parameter value;

[0029] determining a target fiducial point on the Bezier curve according to the position parameter value, and updating the fiducial point sequence according to the coordinates of the target fiducial point.

[0030] Optionally, the missing traffic channel comprises a missing sidewalk.

[0031] The determining the target road segment in the road network model in which the missing traffic channel exists according to the street view image comprises:

[0032] performing semantic segmentation on the street view image;

[0033] obtaining sidewalk information corresponding to the street view image according to the result of the semantic segmentation;

[0034] judging whether a missing sidewalk exists in each road segment in the road network model according to the sidewalk information corresponding to the street view image and the traffic channel information in the road network model, and taking a road segment in which a missing sidewalk exists as a target road segment.

[0035] Optionally, the generating the supplementary fiducial point corresponding to the missing traffic channel according to the street view image and the fiducial point sequence for the target road segment, and completing the missing traffic channel in the road network model according to the supplementary fiducial point comprises:

[0036] determining the size of the missing sidewalk in the target road segment according to the sidewalk information corresponding to the street view image, wherein the sidewalk information comprises a sidewalk mask.

[0037] According to the reference point sequence and the size of the missing sidewalk, a supplementary reference point corresponding to the missing sidewalk is generated, and the missing sidewalk is completed in the road network model according to the supplementary reference point.

[0038] The second aspect of the present application provides a road network data processing system, wherein the road network data processing system comprises:

[0039] A data acquisition module is configured to acquire road network data to be processed corresponding to a road network model, and extract road segment length and road segment shape data corresponding to each road segment from the road network data to be processed, wherein the road segment shape data comprises coordinates of two-dimensional reference points corresponding to the road segment.

[0040] A reference point density determination module is configured to determine the density of reference points in a unit length corresponding to each road segment according to the road segment length and the road segment shape data.

[0041] A reference point supplement module is configured to, if the density of reference points in the unit length does not meet a preset density control condition, determine a control curve according to the coordinates of the two-dimensional reference points in the road segment shape data, insert a target reference point in the road network model according to the coordinates of the two-dimensional reference points and the control curve, and obtain a reference point sequence composed of the two-dimensional reference points and the target reference point.

[0042] An image acquisition module is configured to acquire a street view image matching the road network model.

[0043] A target road segment determination module is configured to determine a target road segment in which a missing traffic channel exists in the road network model according to the street view image.

[0044] A traffic channel completion module is configured to, for the target road segment, generate a supplementary reference point corresponding to the missing traffic channel according to the street view image and the reference point sequence, and complete the missing traffic channel in the road network model according to the supplementary reference point.

[0045] The third aspect of the present application provides a terminal, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of any one of the road network data processing methods.

[0046] The fourth aspect of the present application provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program, when executed by a processor, implements the steps of any one of the road network data processing methods.

[0047] As can be seen, in the scheme, the road network model corresponding to the to-be-processed road network data is acquired, the road section length and road section shape data corresponding to each road section are extracted from the to-be-processed road network data, wherein the road section shape data includes the coordinates of the two-dimensional reference points corresponding to the road section; the unit length reference point density corresponding to each road section is determined according to the road section length and the road section shape data; if the unit length reference point density does not satisfy the preset density control condition, a control curve is determined according to the coordinates of the two-dimensional reference points in the road section shape data, a target reference point is inserted in the road network model according to the coordinates of the two-dimensional reference points and the control curve, and a reference point sequence composed of the two-dimensional reference points and the target reference point is obtained; a street view image matching the road network model is acquired; a target road section in which a missing traffic channel exists in the road network model is determined according to the street view image; and for the target road section, a supplementary reference point corresponding to the missing traffic channel is generated according to the street view image and the reference point sequence, and the missing traffic channel is completed in the road network model according to the supplementary reference point.

[0048] In this way, for the road network model, when the unit length reference point density corresponding thereto does not satisfy the preset density control condition, a target reference point is inserted to complete the reference points of the road network model. Furthermore, a target road section in which a missing traffic channel exists is also determined according to the street view image, and the missing traffic channel in the road network model is completed. In this way, the completion of the reference points and the completion of the missing traffic channel are performed for the road network model, so as to solve the problem of low model accuracy caused by the loss of reference points, and the accuracy of the road network model is improved. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0050] Figure 1 is a flow diagram of a road network data processing method provided by an embodiment of the present application;

[0051] Figure 2 is a reference point supplement effect diagram provided by an embodiment of the present application;

[0052] Figure 3 is a missing sidewalk supplement effect diagram provided by an embodiment of the present application;

[0053] Figure 4Fig. 1 is a schematic diagram of a component module of a road network data processing system according to an embodiment of the present application;

[0054] Figure 5 Fig. 2 is a schematic diagram of an internal structure of a terminal according to an embodiment of the present application. DETAILED DESCRIPTION

[0055] In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular architectures, techniques, etc., in order to provide a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application can be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known methods, devices, circuits, and

[0056] It is to be understood that the terminology "includes", "has", "holds", "contains" and / or "comprising", when used in this specification and in the following claims, indicates the presence of the described features, integers, steps, operations, elements, and / or components but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0057] It is also to be understood that the terminology used in the description herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used in this specification and the appended claims, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise.

[0058] It will be further understood that the terms "and", "or", as used herein, and in the following claims are used to mean one and / or the other and / or both in the items linked by these terms, in addition to the possible combinations of at least one of the items linked by these terms.

[0059] As used in this specification and the appended claims, the term "if' can be construed to mean "when" or "if" or "once" or "in response to a determination" or "in response to a classification" depending on the context. In other words, the phrase "if [a described condition or event] occurs" can be construed to mean "if [a described condition or event] occurs then, once [a described condition or event] occurs, or in response to a determination that [a described condition or event] has occurred" or "in response to a classification that [a described condition or event] has occurred". Similarly, the phrase "if it is determined [that a described condition or event] occurs" can be construed to mean "if it is determined [that a described condition or event] occurs then, once it is determined [that a described condition or event] occurs, or in response to a determination that [a described condition or event] has occurred" or "in response to a classification that [a described condition or event] has occurred".

[0060] The technical solutions in the embodiments of the present application are clearly and completely described below with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort fall within the scope of the present application.

[0061] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the application. However, the application can be practiced without the specific details. In other instances, well-known methods, procedures, components, and circuits have not been described in detail so as not to unnecessarily obscure aspects of the application.

[0062] At present, the application of traffic simulation is more and more widely. Traffic simulation is a typical traffic data analysis method, so it is necessary to generate a lane-level traffic simulation model, to develop a high-precision simulation road section repair technology in order to meet the needs of geometry and topology optimization of large-scale simulation model while ensuring accurate reflection of the space-time characteristics of traffic flow, and to provide a high-trust and sustainable evolution basic road network base for the vehicle-road cooperative control platform.

[0063] The existing simulation road network model relies on XML format data converted based on OSM, and the problem of missing reference points may occur in the conversion process, resulting in road section line mutation, which is difficult to meet the motion and visual requirements of microscopic simulation. At the same time, affected by factors such as street view blind area, later expansion and reconstruction, and artificial modeling omissions, the simulation road network often has the problem of missing sidewalks, which affects the accuracy of the simulation road network model. The traditional artificial inspection and street view and simulation data comparison by section method is time-consuming, high-cost, and slow to update, and has been unable to meet the needs of large-scale road network high-frequency iteration. Therefore, a low-cost and high-precision road section vector automatic repair scheme is urgently needed.

[0064] In order to solve at least one of the above technical problems, the application provides a road network data processing method. In the application scheme, the road network model corresponding to the to-be-processed road network data is obtained, and the road section length and road section shape data corresponding to each road section are extracted from the to-be-processed road network data, wherein the road section shape data includes the coordinates of the two-dimensional reference points corresponding to the road section; the reference point density in the unit length of each road section is determined according to the road section length and the road section shape data; if the reference point density in the unit length does not meet the preset density control condition, a control curve is determined according to the coordinates of the two-dimensional reference points in the road section shape data, a target reference point is inserted into the road network model according to the coordinates of the two-dimensional reference points and the control curve, and a reference point sequence composed of the two-dimensional reference points and the target reference point is obtained; a street view image matching the road network model is obtained; a target road section with a missing traffic channel in the road network model is determined according to the street view image; for the target road section, a supplementary reference point corresponding to the missing traffic channel is generated according to the street view image and the reference point sequence, and the missing traffic channel is completed in the road network model according to the supplementary reference point.

[0065] Thus, for the road network model, when the density of the reference points in the corresponding unit length thereof does not satisfy the preset density control condition, target reference points are inserted to complete the reference points of the road network model. Further, the target road segment with a missing traffic channel is determined according to the street view image, and the missing traffic channel in the road network model is completed. Thus, the completion of the reference points and the completion of the missing traffic channel are performed for the road network model, so as to solve the problem of low model accuracy caused by the loss of reference points, and to improve the accuracy of the road network model.

[0066] As shown in Figure 1 The embodiment of the present application provides a road network data processing method, and specifically, the above method comprises the following steps:

[0067] In step S100, the road network model corresponding to the to-be-processed road network data is obtained, and the road segment length and road segment shape data corresponding to each road segment are extracted from the to-be-processed road network data, wherein the road segment shape data comprises the coordinates of the two-dimensional reference points corresponding to the road segment.

[0068] In step S200, the density of the reference points in the unit length corresponding to each road segment is determined according to the road segment length and the road segment shape data.

[0069] In step S300, if the density of the reference points in the unit length does not satisfy the preset density control condition, a control curve is determined according to the coordinates of the two-dimensional reference points in the road segment shape data, target reference points are inserted in the road network model according to the coordinates of the two-dimensional reference points and the control curve, and a reference point sequence composed of the two-dimensional reference points and the target reference points is obtained.

[0070] In step S400, a street view image matching the road network model is obtained.

[0071] In step S500, a target road segment with a missing traffic channel in the road network model is determined according to the street view image.

[0072] In step S600, for the target road segment, the supplementary reference points corresponding to the missing traffic channel are generated according to the street view image and the reference point sequence, and the missing traffic channel is completed in the road network model according to the supplementary reference points.

[0073] The road network model is a simulation road network model generated in advance based on XML data. In the embodiment of the present application, the road network model is a corresponding simulation road network model in the city traffic simulation software (SUMO, Simulation of Urban MObility).

[0074] Specifically, XML format data of a road network model is acquired as to-be-processed road network data, which contains four types of elements of intersections, road segments, lanes and connecting lines. All road segments are extracted from the to-be-processed road network data through Python, and the road segment length and road segment shape data corresponding to each road segment are determined.

[0075] In some specific application scenarios, the attributes and values of the road segment identifier (id), the road segment length (length) and the road segment shape data (shape) corresponding to each road segment can be recorded. The road segment shape data includes the coordinates of all two-dimensional reference points describing the road segment, which is represented as , wherein, represents the total number of two-dimensional reference points corresponding to the road segment, represents the coordinates of the first two-dimensional reference point corresponding to the road segment.

[0076] Specifically, the determination of the reference point density per unit length corresponding to each road segment according to the road segment length and the road segment shape data includes:

[0077] determining the number of two-dimensional reference points corresponding to each road segment according to the road segment shape data;

[0078] For each road segment, the reference point density per unit length corresponding to the road segment is determined according to the ratio of the number of two-dimensional reference points corresponding to the road segment to the road segment length.

[0079] For a road segment, the ratio of the number of all two-dimensional reference points to the road segment length is calculated to obtain the reference point density per unit length. If the requirement is not met, a new reference point is added.

[0080] Specifically, the road segment length is , and the road segment is composed of two-dimensional reference points, then the reference point density per unit length corresponding to the road segment is as shown in the following formula:

[0081] ;

[0082] Based on the above formula, the reference point density per unit length corresponding to each road segment can be calculated and obtained, so as to determine whether the reference point density per unit length meets the preset density control condition. If not, a new reference point needs to be added for the road segment. In some application scenarios, the target number of reference points can be obtained by multiplying the road segment length by the minimum reference point density.

[0083] It should be noted that the control curve is a type of three-dimensional curve that is preset, which is equivalent to linear basic elements such as straight lines, circular curves, clothoids and parabolas in two-dimensional design. Specifically, the three-dimensional curve is set to have good continuity, so that the selection of a new reference point can be performed on the curve.

[0084] In particular, the above curve satisfies the following predetermined requirements: the selected line shape ensures G1 continuity at any point in the range, i.e., slope continuity; the curve that can be expressed by the selected mathematical model of the curve has a unique and definite geometric shape in space; the selected line shape has geometric invariance and does not change in geometric shape due to translation or rotation of the reference coordinate system, but only depends on the change in the relative positions of the control points; the selected line shape can be intuitively adjusted and changed using the control method, and the degree of adjustment can be easily understood by the designer through parameters to achieve the expected adjustment result; the mathematical model or equation expression of the selected line shape should reduce the calculation difficulty and workload as much as possible, reduce the number of unknown parameters, avoid the appearance of high-order equations, and enable the computer to process data in a short time.

[0085] In some application scenarios, a Hermite spline curve, a Catmull-Rom spline curve, or a T spline curve can be selected as the control curve.

[0086] In the embodiments of the present application, the control curve is a Bezier curve; if the density of the reference points in the unit length does not meet the predetermined density control condition, the coordinates of the two-dimensional reference points in the road segment shape data are used to determine the control curve, a target reference point is inserted between the two-dimensional reference points in the road network model according to the coordinates of the two-dimensional reference points and the control curve, and a reference point sequence composed of the two-dimensional reference points and the target reference point is obtained, including:

[0087] The coordinates of the two-dimensional reference points in the road segment shape data are used to construct a reference point sequence;

[0088] It is determined whether the density of the reference points in the unit length is greater than a predetermined minimum density;

[0089] If the density of the reference points in the unit length is not greater than the minimum density, two candidate reference points with the maximum distance between adjacent points are determined from the reference point sequence according to the coordinates of the reference points in the reference point sequence;

[0090] The coordinates of the candidate reference points are used to determine the Bezier curve corresponding to the candidate reference points;

[0091] A target reference point is inserted between the candidate reference points according to the Bezier curve corresponding to the candidate reference points, and the reference point sequence is updated;

[0092] The density of the fiducial points in the unit length is updated according to the number of the fiducial points in the fiducial point sequence, and the step of judging whether the density of the fiducial points in the unit length is greater than the preset minimum density is returned to be executed again until the density of the fiducial points in the unit length is greater than the minimum density.

[0093] It should be noted that the number of fiducial points in the fiducial point sequence reaching the target number of fiducial points can also be used as a judgment condition for ending the loop, which is not limited here.

[0094] Specifically, the determination of the Bezier curve corresponding to the candidate fiducial point according to the coordinates of the candidate fiducial point comprises:

[0095] If the candidate fiducial point includes the first fiducial point or the last fiducial point in the fiducial point sequence, a second-order Bezier curve corresponding to the candidate fiducial point is determined according to the coordinates of the candidate fiducial point.

[0096] Otherwise, a third-order Bezier curve corresponding to the candidate fiducial point is determined according to the coordinates of the candidate fiducial point.

[0097] Further, the insertion of a target fiducial point between the candidate fiducial points according to the Bezier curve corresponding to the candidate fiducial point and the updating of the fiducial point sequence comprise:

[0098] Obtaining a preset position parameter value;

[0099] Determining a target fiducial point on the Bezier curve according to the position parameter value, and updating the fiducial point sequence according to the coordinates of the target fiducial point.

[0100] In actual application, in order to ensure the fluency of the connection of the fiducial points, a Bezier curve is selected as the control curve. Considering that the osculating plane of the fitting curve of the road section should always change in three-dimensional space, at least four control points are needed to form a control polygon with three edges to ensure that a spatial curve can be formed, so a Bezier curve of three orders or more can be selected.

[0101] The expression of the second-order Bezier curve is as follows:

[0102] ;

[0103] Among them, to indicate three sequentially arranged control points required to construct a second-order Bezier curve. indicates the position of a point on the curve, and the value range is [0, 1].

[0104] The expression of the third-order Bezier curve is as follows:

[0105] ;

[0106] wherein, to denote four sequentially arranged control points required to construct a cubic Bezier curve. denote the position of a point on the curve, and the value range is [0, 1].

[0107] Each order Bezier curve function can be described as follows:

[0108] ;

[0109] wherein, denote the control points required to construct a Bezier curve, denote the polynomial coefficients corresponding to the control points.

[0110] The coefficients of the control points in each order Bezier curve function are , The calculation is as follows:

[0111] ;

[0112] wherein, denote the order of the Bezier curve. denote the control point index, which is an integer from 0 to .

[0113] When using a cubic Bezier curve, four control points are required. For the two reference points and ( ) given in the reference point sequence, they are taken as the starting control point and the ending control point of the Bezier curve, respectively, wherein, represents the number of reference points in the reference point sequence, and also represents the number of reference points corresponding to the road segment, and both have the same meaning.

[0114] For the intermediate control point , an auxiliary straight line AB is drawn through the point parallel to the vector , and the position of the control point is determined according to a preset proportion in the area between the point on the straight line AB and the point . The above-mentioned preset proportion can be pre-set according to actual needs, or can be determined based on the ratio of the head-to-tail length of the road segment to the length of the road segment. The head-to-tail length Specifically, it refers to the straight-line distance between the first and last benchmark points in the benchmark point sequence, and the road segment length. Specifically, it is the sum of the straight-line distances between two consecutive reference points in the reference point sequence. Control points Satisfy the following formula:

[0115] ;

[0116] in, The above-mentioned preset ratio is a customizable parameter between 0 and 1. The choice of its value affects the shape and smoothness of the Bézier curve, and its specific value can be set and adjusted according to actual needs.

[0117] For control points past the point Construct parallel vectors An auxiliary line CD, and a point on line CD. With point In the area between them, control points are obtained according to the aforementioned preset ratio. Location of control points. Satisfy the following formula:

[0118] ;

[0119] In this embodiment of the application, the length of the first and last ends of the road segment is utilized. With road segment length The ratio is used as a determining parameter The basis is shown in the following formula:

[0120] ;

[0121] ;

[0122] ;

[0123] in, and These represent the x and y coordinates of the first reference point in the reference point sequence, respectively. and These represent the first and second points in the reference point sequence, respectively. The x and y coordinates of each reference point are used, and other parameters are similarly calculated, which will not be elaborated here.

[0124] After determining the location of the control points, the corresponding control curves can be determined based on the location of the control points, and then the target reference points can be generated based on the control curves.

[0125] Specifically, two candidate reference points with the largest distance between adjacent points in the current reference point sequence are calculated and found out, and based on the Bezier curve parameter equation corresponding to the two candidate reference points, the coordinates of the point corresponding to the position parameter value of 0.5 (the specific value can be set and adjusted according to actual requirements) are determined as the coordinates of the new reference point (i.e., the target reference point), and the reference point is inserted into the existing reference point sequence.

[0126] The generation process of the target reference point is repeated until a sufficient number of reference points are obtained.

[0127] In the embodiments of the present application, the missing traffic channel includes a missing sidewalk; and determining the target road segment in which the missing traffic channel exists in the road network model according to the street view image includes:

[0128] performing semantic segmentation on the street view image;

[0129] obtaining sidewalk information corresponding to the street view image according to the result of semantic segmentation;

[0130] judging whether a missing sidewalk exists in each road segment in the road network model according to the sidewalk information corresponding to the street view image and the traffic channel information in the road network model, and taking the road segment in which the missing sidewalk exists as the target road segment.

[0131] Specifically, for the target road segment, the supplementary reference point corresponding to the missing traffic channel is generated according to the street view image and the reference point sequence, and the missing traffic channel is completed in the road network model according to the supplementary reference point, including:

[0132] determining the size of the missing sidewalk in the target road segment according to the sidewalk information corresponding to the street view image, wherein the sidewalk information includes a sidewalk mask;

[0133] generating the supplementary reference point corresponding to the missing sidewalk according to the reference point sequence and the size of the missing sidewalk, and completing the missing sidewalk in the road network model according to the supplementary reference point.

[0134] The street view image can be obtained in batches based on a crawler technology, or can be obtained by other means, which is not limited here.

[0135] In some application scenarios, all street view images are geographically registered, and the street view images are geographically registered to realize registration of the street view images and a road network model. Pixel-level semantic segmentation is performed on the street view images to locate a region of interest (ROI) corresponding to a road region, and a region mask and a local feature are output, which provide input samples for subsequent fine-grained recognition. On the basis of segmentation, a fine-grained classifier is trained to recognize a sidewalk category of the ROI. Specifically, a road image dataset related to a target region is constructed, unified preprocessing and data enhancement are performed, and training / validation / testing sets are proportionally divided to evaluate Top-1 accuracy and robustness. The fine-grained classifier is based on a transfer learning strategy, uses a deep backbone network pre-trained on a large-scale dataset as a basis, combines an attention mechanism, and is fine-tuned to the local road image dataset to enhance sensitivity to texture and morphological features, improve cross-domain feature alignment, and improve recognition accuracy under small samples. The classifier outputs a semantic label, a local confidence score, and a corresponding local visual feature vector of each recognized object. The sidewalk information in the simulation model is compared, and it is determined whether the sidewalk of the road section is missing. For the road section with missing sidewalks, new sidewalk elements are added, and the street view image and the simulation model lane are geometrically registered to determine the size of the sidewalk. Finally, according to the road section reference point and the size of the sidewalk, the sidewalk reference point is calculated along the normal direction of the sub-road section where the road section reference point is located, and the sidewalk reference point is converted into a geometric shape expression conforming to the simulation road network data specification.

[0136] Specifically, in the embodiments of the present application, the corresponding street view image is found according to the longitude and latitude of each reference point of the road section, as the street view RGB image to be processed. For the street view RGB image to be processed, a preset preprocessing operation is performed for preprocessing. Specifically, center / equal scaling is first performed according to the aspect ratio, and then normalized to the ImageNet mean and standard deviation; the image for pixel-level semantic segmentation is uniformly adjusted to 1024x2048 to balance segmentation accuracy and computational overhead; and the ROI cropped image for fine-grained classification is uniformly scaled to 224x224 pixels in subsequent steps to adapt to the input requirements of the classifier. The preprocessing records the geographical reference of the original image to facilitate subsequent coordinate mapping.

[0137] On the basis of analyzing the construction experience of public datasets such as ImageNet, a high-quality road image dataset of the region where the road network model is located and the surrounding area is supplemented. The road dataset contains a total of 5641 images, and the number of images of each type of road is between 120 and 250. The enhanced images are divided into training set and test set according to the ratio of 4:1. The images in the training set are used for full training and iterative optimization of the network, and the test set is used to evaluate the Top-1 accuracy of the final model on the 20-class local road recognition task. In this way, the dataset is constructed based on the above steps. Further, based on the above training set and test set, the training and testing of the semantic segmentation model can be realized until the semantic segmentation model is trained.

[0138] In the embodiments of the application, the semantic segmentation model is a model based on the DeepLabV3+ native architecture. The pixel-level semantic segmentation is performed for key elements in the street view picture, and feature fusion is performed for the semantic segmentation process of the street view elements in the complex scene. The ROI of each type of element is accurately defined to provide input for the downstream recognition and detection task. The Backbone selects the ResNet-101 pre-trained on ImageNet. Specifically, ResNet-101 is selected as the backbone network, and ResNet-101 specifically refers to a residual network (ResNet) with 101 layers. ResNet-101 effectively solves the problems of gradient disappearance and gradient explosion in the training of deep neural networks by introducing a residual learning mechanism and a skip connection, so that the network can be more effectively trained to a deeper layer, thereby learning more complex feature representations. In addition, the pre-trained model of ResNet-101 on a large-scale dataset can be migrated to a specific task, accelerating the training process and improving the performance on a small dataset.

[0139] The network corresponding to the semantic segmentation model first performs a series of downsampling and dilated convolution operations on the input RGB street view image to obtain rich high-level semantic features. Then, five parallel features (dilation rates = 1, 6, 2, 18, 24) are generated through the Atrous Spatial Pyramid Pooling (ASPP) module, and then reduced and fused through 1x1 convolution. After being fused with shallow features in the decoder (Decoder), it is upsampled to the original image size through bilinear interpolation, and the probability distribution of each pixel is output.

[0140] The loss function is pixel-level cross-entropy (Cross-Entropy), and the learning rate decay strategy is adopted. The optimizer adopts SGD+Momentum (momentum = 0.9, weight_decay = 1e-4).

[0141] When training and testing the semantic segmentation model, the DeepLabV3+ pre-training weight is loaded, the single image full resolution forward inference is performed, the classification pixel probability map is obtained, and post-processing such as DenseCRF and connected domain filtering is performed to refine the edge, eliminate noise, and improve the coherence of the segmentation boundary. The segmentation effect is monitored by mIoU and pixel accuracy (PA). In this way, the post-processing and evaluation of the semantic segmentation model are realized.

[0142] After completing the DeepLabV3+ based semantic segmentation and locating the ROI of the road elements, these regions are segmented, and the semantic labels, region masks, and local features are output to provide input samples for subsequent fine-grained recognition.

[0143] Further, a lightweight lane category detection model is trained to output the attribute labels and spatial positions corresponding to each element. The identification base network of the model is determined, different architectures of deep convolutional neural networks DenseNet, Inception, ResNeXt, and MobileNet are selected, the full network is trained on the image training dataset, and the trained model is used to infer the test dataset. The above model effects are evaluated by the recognition accuracy Top1-ACC, and the pre-trained model is fine-tuned and predicted using the self-built dataset to compare the recognition effects of the network model migrated on ImageNet. The full network is trained on the road training set, and after iteration, the trained lane category detection model is obtained. The lane category detection model is used to infer the test set, and the ResNet50 with the highest Top1-ACC accuracy is obtained as the backbone network of lane recognition. In this way, the training of the fine-grained classifier is realized. Further, in the use process, the sidewalk information in the street view image can be recognized based on the trained model.

[0144] Compared with the ImageNet million image dataset, the training effect of the network parameters of the local dataset may be limited by the size of the sample amount, so the transfer learning fine-tuning is performed.

[0145] The structure and parameters of the pre-trained network model on the large dataset are migrated to the local, and the local dataset is used for fine-tuning. In the embodiment of the application, the fine-tuning method is to freeze all parameters except the last layer, only learn the characteristics of the local dataset on the parameters of the fully connected layer at the last layer, and fine-tune according to the back propagation. In this way, the transfer learning and fine-tuning of the model are realized, and the accuracy of the model is further improved.

[0146] Building upon the basic network model, an attention mechanism is added to improve recognition accuracy. Training and testing are performed on a dataset using a "channel-first, spatial-second" convolutional block attention module (CBAM), which fully considers the combination of channel and spatial domain attention, and incorporates average pooling and convolution operations. The optimized network is then used to select the most suitable candidate materials.

[0147] For each lane ROI, two parallel feature vectors are extracted: one from the intermediate layer of the semantic segmentation network; and the other from the penultimate layer of the fine-grained classifier. The extracted raw floating-point vectors are L2 normalized and subjected to Principal Component Analysis (PCA) to reduce storage and retrieval overhead; simultaneously, a set of simple color / texture statistics are generated as auxiliary features. All features are written in a uniform format to the corresponding JSON metadata field of the ROI and simultaneously written to the vector database for efficient approximate nearest neighbor retrieval.

[0148] Furthermore, the features related to sidewalks mentioned above are used as the sidewalk information corresponding to the street view image. Lane registration is performed between the street view image and the road network model, and the missing sidewalks are obtained by comparing the street view image with the road network model. The size of the missing sidewalks is then estimated based on the ratio between the driving lanes in the road network model and the lanes in the street view image.

[0149] Specifically, within the simulated lanes of the simulated road segment in the road network model, a geometrically complete and attribute-reliable simulated lane is selected as the baseline lane from the registered road segment. A pixel-level mask of this baseline lane obtained based on semantic segmentation is acquired in the street view image as the pedestrian walkway mask, and its pixel width is measured on the mask's cross-section. (Unit: pixels). The XML data of the road network model already records the actual design width of the baseline lane. (Unit: meters) This allows us to establish the ground sampling distance coefficients that influence the detection process. Subsequently, for missing sidewalks marked as missing during the detection phase, their corresponding pixel-level masks are extracted from the same frame image, and the pixel width is measured. (Unit: pixels). The actual width of the missing sidewalk. It can be obtained through the following proportional mapping:

[0150] ;

[0151] In this way, the true width of the missing sidewalk can be calculated. , as the size of the missing sidewalk. Further, according to the existing reference points in the reference point sequence, the corresponding supplementary reference point of the sidewalk is calculated in combination with the size of the missing sidewalk, so as to realize the completion of the sidewalk.

[0152] Specifically, for each reference point in the reference point sequence , the direction vector of the next adjacent reference point is calculated, as shown in the following formula:

[0153] ;

[0154] wherein, represents the direction vector formed by the reference point and the adjacent reference point , and and represent the horizontal coordinate and the vertical coordinate of the reference point , respectively. and represent the horizontal coordinate and the vertical coordinate of the reference point , respectively.

[0155] For each reference point , the normal vector perpendicular to (i.e. the vector after rotating 90° clockwise) is calculated, as shown in the following formula:

[0156] ;

[0157] wherein, represents the normal vector perpendicular to in the clockwise direction; represents the component of in the direction of , and represents the component of in the direction of .

[0158] For each reference point , the corresponding sidewalk supplementary reference point is calculated according to the following formula:

[0159] ;

[0160] wherein, represents the total width of the motorway of the current target road section, represents the actual width of the missing sidewalk.

[0161] In this way, based on the original reference points and the newly generated supplementary reference points, the missing sidewalk can be completed, thereby improving the accuracy of the road network model.

[0162] Figure 2 is a reference point supplement effect schematic diagram provided by an embodiment of the present application, Figure 2 In the red area, the intersection is represented by the black area, the motor vehicle lane is represented by the gray area, and the pedestrian lane is represented by the motor vehicle lane and the pedestrian lane together. Specifically, Figure 2 In the right-turn ramp is identified as a road segment that needs to supplement the reference point, Figure 2 In the supplement effect is reflected in the number of reference points of the road segment from 6 to 14. Figure 3 is a missing pedestrian lane supplement effect schematic diagram provided by an embodiment of the present application, Figure 3 In the supplement effect is reflected in the increase of the gray pedestrian lane.

[0163] In the present embodiment, the reference point completion of the curved road segment and the completion of the missing pedestrian lane are realized based on the above road network data processing method. Specifically, for the road segment whose reference point density in unit length does not meet the preset density control condition, new reference points are inserted according to the Bezier curve between its continuous reference points until the reference point density in unit length meets the preset density control condition. When completing the missing pedestrian lane, the image semantic segmentation model is used to extract the pedestrian lane information in the street view picture corresponding to the road segment, and the pedestrian lane information in the simulation model is compared to determine whether the road segment has a missing pedestrian lane. For the road segment with a missing pedestrian lane, the street view picture and the simulation model lane are geometrically registered to determine the size of the pedestrian lane. Finally, the pedestrian lane reference point is calculated according to the road segment reference point combined with the size of the pedestrian lane, thereby realizing the completion of the pedestrian lane.

[0164] In this way, for the road network model, when the reference point density in the corresponding unit length thereof does not meet the preset density control condition, the target reference point is inserted to realize the completion of the reference points of the road network model. Further, the target road segment with a missing traffic channel is determined according to the street view image, and the missing traffic channel in the road network model is completed. In this way, the completion of the reference points and the completion of the missing traffic channel are performed for the road network model, so as to solve the problem of low model accuracy caused by the loss of reference points, and to improve the accuracy of the road network model.

[0165] Specifically, taking the road segments of the SUMO simulation road network as the research object, it is judged whether it needs to be processed by calculating the benchmark point density; after judging that it needs to be processed, the slope of the Bezier curve on both sides of the benchmark point is equal, and the direction of the line connecting the front and rear benchmark points is taken as the benchmark, an auxiliary straight line is made through the benchmark point and parallel to the direction, control points are selected on the auxiliary straight line in proportion, and a cubic Bezier curve is constructed between the continuous benchmark points of the curve segment, and new benchmark points are repeatedly added between the benchmark points with the largest interval according to the constructed Bezier curve; further, according to the latitude and longitude of the existing benchmark points, the corresponding street view pictures are crawled, the sidewalk information is extracted by using the image semantic segmentation model, and whether the sidewalk of the road segment is missing is judged by comparing the sidewalk information of the simulation model with the sidewalk information of the road segment. For the road segment with missing, new sidewalk elements are added, and the sidewalk size is determined by geometric registration between the street view picture and the simulation model lane; finally, according to the road segment benchmark point combined with the sidewalk size, the sidewalk benchmark point is calculated along the normal direction of the sub-road segment where the road segment benchmark point is located, and the sidewalk benchmark point is converted into a geometric shape expression conforming to the simulation road network data specification.

[0166] In view of the problems of missing road segment benchmark points and missing sidewalks in the SUMO simulation model in the conversion process, the Bezier curve interpolation strategy and the street view image semantic segmentation technology are used to realize the completeness and continuity of the road segment in geometry and topology, solve the problem of large manual intervention, and improve the accuracy and efficiency of automatic repair and update of the road segment model.

[0167] As shown in Figure 4 Corresponding to the road network data processing method described above, the embodiments of the present application also provide a road network data processing system, which comprises:

[0168] A data acquisition module 410 is configured to acquire road network model corresponding to the to-be-processed road network data, and extract the road segment length and road segment shape data corresponding to each road segment from the to-be-processed road network data, wherein the road segment shape data comprises the coordinates of the two-dimensional benchmark points corresponding to the road segment.

[0169] A benchmark point density determination module 420 is configured to determine the benchmark point density per unit length corresponding to each road segment according to the road segment length and the road segment shape data.

[0170] A benchmark point supplement module 430 is configured to, if the benchmark point density per unit length does not satisfy the preset density control condition, determine a control curve according to the coordinates of the two-dimensional benchmark points in the road segment shape data, insert a target benchmark point in the road network model according to the coordinates of the two-dimensional benchmark points and the control curve, and obtain a benchmark point sequence composed of the two-dimensional benchmark points and the target benchmark point.

[0171] An image acquisition module 440 is configured to acquire a street view image matched with the road network model.

[0172] The target section determination module 450 is configured to determine a target section with a missing traffic lane in the road network model according to the street view image.

[0173] The traffic lane completion module 460 is configured to, for the target section, generate a supplementary reference point corresponding to the missing traffic lane according to the street view image and the reference point sequence, and complete the missing traffic lane in the road network model according to the supplementary reference point.

[0174] In this way, for the road network model, when the reference point density in the corresponding unit length thereof does not satisfy the preset density control condition, the target reference point is inserted to complete the reference points of the road network model. Further, the target section with the missing traffic lane is determined according to the street view image, and the missing traffic lane in the road network model is completed. In this way, the completion of the reference points and the completion of the missing traffic lane are performed for the road network model, so as to solve the problem of low model accuracy caused by the loss of reference points, and the accuracy of the road network model is improved.

[0175] It should be noted that the specific structure and implementation manner of the road network data processing system and each module or unit thereof can be referred to the corresponding description in the above method embodiments, which will not be described here.

[0176] It should be noted that the division manner of each module of the road network data processing system is not unique, and is not specifically limited here.

[0177] Based on the above embodiments, the application further provides a terminal, and a principle block diagram thereof can be shown as Figure 5 The terminal includes a processor, a memory, a network interface, and a display screen connected through a system bus. The processor of the terminal is configured to provide calculation and control capabilities. The memory of the terminal includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The network interface of the terminal is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement the steps of any one of the road network data processing methods. The display screen of the terminal can be a liquid crystal display screen or an electronic ink display screen.

[0178] Those skilled in the art can understand, Figure 5 that the principle block diagram shown in the above

[0179] In one embodiment, a terminal is provided, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, which, when executed by the processor, implements the steps of any of the road network data processing methods provided in the embodiments.

[0180] The embodiments of the present application further provide a computer readable storage medium, which stores a computer program, and the computer program, when executed by a processor, implements the steps of any of the road network data processing methods provided in the embodiments.

[0181] It should be understood that the sequence of the steps in the above embodiments does not mean the order of execution, and the execution order of the processes should be determined according to their functions and inherent logic, and should not constitute any limitation on the implementation process of the embodiments.

[0182] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is taken as an example, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit, and the integrated unit can be realized in the form of hardware or software. In addition, the specific names of the functional units and modules are only for easy distinction, and do not limit the protection scope of the present application. The specific working process of the units and modules in the device can be referred to the corresponding process in the foregoing method embodiments, which will not be described here.

[0183] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the related description of other embodiments.

[0184] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in connection with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different ways to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0185] In the embodiments provided by the present application, it should be understood that the disclosed system / terminal device and method can be implemented in other manners. For example, the embodiments of the system / terminal device described above are merely schematic; for example, the division of the modules or units is merely logical function division; an actual mapping relationship between a physical hardware and a function can be different, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings between different units, or the coordinated actions between different units can not be actual couplings or actions, but can be replaced with other couplings or actions that can achieve the same purpose.

[0186] The integrated modules / units, if implemented in the form of software function units and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such an understanding, the present application implements all or part of the processes in the above-described embodiments by means of a computer program, and the computer program can be stored in a computer readable storage medium. The computer program, when executed by a processor, can implement the steps of the above-described various method embodiments. The computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form. The computer readable medium can include any entity or device capable of carrying the above computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier wave signal and software distribution medium, etc. It should be noted that the computer readable storage medium contained in the above can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0187] The above-described embodiments are merely used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to some technical features; and these modifications or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.

Claims

1. A road network data processing method, characterized by, The method includes: Obtain the road network data to be processed corresponding to the road network model, and extract the road segment length and road segment shape data corresponding to each road segment from the road network data to be processed, wherein the road segment shape data includes the coordinates of the two-dimensional reference point corresponding to the road segment; Based on the road segment length and road segment shape data, determine the density of reference points per unit length for each road segment; If the density of reference points within the unit length does not meet the preset density control conditions, then a control curve is determined based on the coordinates of the two-dimensional reference points in the road segment shape data. Based on the coordinates of the two-dimensional reference points and the control curve, a target reference point is inserted into the road network model, and a reference point sequence composed of the two-dimensional reference points and the target reference point is obtained. Obtain street view images that match the road network model; Based on the street view image, identify the target road segments in the road network model that have missing traffic channels; For the target road segment, supplementary reference points are generated based on the street view image and the reference point sequence, and the missing traffic channels are completed in the road network model based on the supplementary reference points.

2. The road network data processing method according to claim 1, characterized in that, The step of determining the density of reference points per unit length for each road segment based on the road segment length and road segment shape data includes: The number of two-dimensional reference points corresponding to each road segment is determined based on the road segment shape data; For each road segment, the density of reference points per unit length of the road segment is determined based on the ratio of the number of two-dimensional reference points corresponding to the road segment to the length of the road segment.

3. The road network data processing method according to claim 1, characterized in that, The control curve is a Bézier curve; If the density of reference points within the unit length does not meet the preset density control conditions, a control curve is determined based on the coordinates of the two-dimensional reference points in the road segment shape data. Based on the coordinates of the two-dimensional reference points and the control curve, target reference points are inserted into the road network model, and a reference point sequence consisting of the two-dimensional reference points and the target reference points is obtained, including: Based on the coordinates of the two-dimensional reference points in the road segment shape data, a reference point sequence is constructed; Determine whether the density of reference points within the unit length is greater than a preset minimum density; If the density of reference points within the unit length is not greater than the minimum density, then based on the coordinates of each reference point in the reference point sequence, determine two candidate reference points that are adjacent to each other and have the largest distance between adjacent points from the reference point sequence. Determine the Bézier curve corresponding to the candidate reference point based on the coordinates of the candidate reference point; Based on the Bézier curves corresponding to the candidate reference points, a target reference point is inserted between the candidate reference points, and the reference point sequence is updated; The reference point density within a unit length is updated based on the number of reference points in the reference point sequence, and the step of determining whether the reference point density within a unit length is greater than the preset minimum density is re-executed until the reference point density within a unit length is greater than the minimum density.

4. The road network data processing method according to claim 3, characterized in that, Determining the Bézier curve corresponding to the candidate reference point based on the coordinates of the candidate reference point includes: If the candidate reference point comprises a first reference point or a last reference point in the sequence of reference points, a second order Bezier curve corresponding to the candidate reference point is determined according to the coordinates of the candidate reference point; Otherwise, a third order Bezier curve corresponding to the candidate reference point is determined according to the coordinates of the candidate reference point.

5. The road network data processing method according to claim 3, characterized in that, The inserting a target reference point between the candidate reference points according to the Bezier curve corresponding to the candidate reference point and updating the sequence of reference points comprises: obtaining a preset position parameter value; determining a target reference point on the Bezier curve according to the position parameter value, and updating the sequence of reference points according to the coordinates of the target reference point.

6. The road network data processing method according to claim 1, wherein The missing traffic channel comprises a missing sidewalk; The determining a target road segment in which a missing traffic channel exists in the road network model according to the street view image comprises: performing semantic segmentation on the street view image; obtaining sidewalk information corresponding to the street view image according to the result of semantic segmentation; judging whether a missing sidewalk exists in each road segment in the road network model according to the sidewalk information corresponding to the street view image and traffic channel information in the road network model, and taking a road segment in which a missing sidewalk exists as a target road segment.

7. The road network data processing method according to claim 6, characterized in that, The generating a supplementary reference point corresponding to the missing traffic channel according to the street view image and the sequence of reference points for the target road segment, and completing the missing traffic channel in the road network model according to the supplementary reference point comprises: determining the size of a missing sidewalk in a target road segment according to sidewalk information corresponding to the street view image, wherein the sidewalk information comprises a sidewalk mask; generating a supplementary reference point corresponding to the missing sidewalk according to the sequence of reference points and the size of the missing sidewalk, and completing the missing sidewalk in the road network model according to the supplementary reference point.

8. A road network data processing system, characterized by, The system comprises: a data acquisition module configured to acquire road network model corresponding to the to-be-processed road network data, and extract road segment length and road segment shape data corresponding to each road segment from the to-be-processed road network data, wherein the road segment shape data comprises the coordinates of two-dimensional reference points corresponding to the road segment; a reference point density determination module configured to determine the reference point density per unit length corresponding to each road segment according to the road segment length and the road segment shape data; a reference point supplementing module configured to, if the reference point density per unit length does not meet a preset density control condition, determine a control curve according to the coordinates of the two-dimensional reference points in the road segment shape data, insert a target reference point in the road network model according to the coordinates of the two-dimensional reference points and the control curve, and obtain a sequence of reference points composed of the two-dimensional reference points and the target reference point; an image acquisition module configured to acquire a street view image matched with the road network model; a target road segment determination module configured to determine a target road segment in which a missing traffic channel exists in the road network model according to the street view image; and a reference point supplementing module configured to, if the reference point density per unit length does not meet a preset density control condition, determine a control curve according to the coordinates of the two-dimensional reference points in the road segment shape data, insert a target reference point in the road network model according to the coordinates of the two-dimensional reference points and the control curve, and obtain a sequence of reference points composed of the two-dimensional reference points and the target reference point. The traffic channel completion module is configured to generate a supplementary reference point corresponding to the missing traffic channel according to the street view image and the reference point sequence, and complete the missing traffic channel in the road network model according to the supplementary reference point.

9. A terminal, characterized by comprising: The terminal comprises a memory, a processor, and a computer program stored on the memory and executable on the processor, and the computer program, when executed by the processor, implements the steps of the road network data processing method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program, when executed by the processor, implements the steps of the road network data processing method according to any one of claims 1 to 7.

Images