CN116310189A

CN116310189A - Map model construction method and terminal

Info

Publication number: CN116310189A
Application number: CN202310578856.6A
Authority: CN
Inventors: 崔婵婕; 刘林海; 张品品; 任宇鹏; 黄积晟; 李乾坤
Original assignee: Zhejiang Dahua Technology Co Ltd
Current assignee: Zhejiang Dahua Technology Co Ltd
Priority date: 2023-05-22
Filing date: 2023-05-22
Publication date: 2023-06-23
Anticipated expiration: 2043-05-22
Also published as: CN116310189B

Abstract

The invention provides a map model construction method and a terminal, wherein the map model construction method comprises the following steps: acquiring an orthographic image and a digital surface model corresponding to a target area; performing category detection on each pixel in the orthophoto map to obtain a detection category corresponding to each pixel; determining a connected domain corresponding to each detection category based on the pixel category corresponding to each pixel; mapping the position of each connected domain in the orthographic image to a digital surface model, and determining the height information of each connected domain; and constructing a map model of the target area based on the height information of the connected domains and the positions of the connected domains in the orthophoto map. According to the method and the device, the position of the ground object contained in the orthographic image of the target area is detected, the height information of the ground object corresponding to the connected domain is determined based on the area data of the connected domain mapped in the digital surface model, and then the map model of the target area is constructed, so that the accuracy of the map model is improved.

Description

Map model construction method and terminal

Technical Field

The present invention relates to the field of image recognition technologies, and in particular, to a map model building method and a terminal.

Background

Urban road intersections are traffic scenes with the most complex environments, the most participants and the most frequent problem conditions in urban roads. The road traffic system is a node and a junction of the road traffic system, and is used for bearing a large amount of traffic flow, and the smoothness degree of the intersection directly influences traffic capacity. The holographic road junction deployed based on the high-precision map, the intelligent camera and the radar fusion road surface sensing system can calculate various data such as the vehicle position, the speed, the track, the driving gesture, the vehicle attribute, the lane occupancy, the average speed, the vehicle queuing length, the zebra crossing state and the like of the road junction in real time, and perform intelligent analysis, and form a traffic flow thermodynamic diagram, so that the work of controlling and adjusting the signal duration of the traffic signal lamp can be performed in time, the traffic running efficiency is improved, and the traffic jam is relieved. The high-precision map is used as a base map of the holographic intersection, is a key place for opening the holographic visual angle of the real-scene intersection, and can more clearly describe a kilometer model on one hand and more accurately reflect real-time road conditions of each lane on the other hand.

The current high-precision data acquisition comprises three methods of mobile vehicle measurement acquisition, 1:500 topographic map mapping and unmanned aerial vehicle aerial survey. The mobile surveying and mapping vehicle needs to be provided with sensor equipment such as a laser radar, inertial navigation (RTK), a panoramic camera and the like, and has high integrated cost and high acquisition cost; 1:500 topographic map mapping, a great deal of work such as network layout, field mapping, internal construction mapping and the like needs to be controlled, and the working period is long; the unmanned aerial vehicle aerial survey is a method with lower acquisition cost and higher processing speed, but the accuracy of the holographic intersection map model obtained by the unmanned aerial vehicle aerial survey is poor at present.

Disclosure of Invention

The invention mainly solves the technical problem of providing a map model construction method and a terminal, and solves the problem of lower accuracy of a map model constructed in the prior art.

In order to solve the technical problems, the first technical scheme adopted by the invention is as follows: the map model building method comprises the following steps:

acquiring an orthographic image and a digital surface model corresponding to a target area;

performing category detection on each pixel in the orthophoto map of the target area to obtain a detection category corresponding to each pixel;

determining a connected domain corresponding to each detection category in the target area based on the pixel category corresponding to each pixel in the orthographic image;

mapping the position of each connected domain in the orthographic image to a digital surface model, and determining the height information of each connected domain;

and constructing a map model of the target area based on the height information of the connected domains and the positions of the connected domains in the orthophoto map.

The method for acquiring the orthographic image and the digital surface model corresponding to the target area comprises the following steps:

acquiring an image sequence, wherein the image sequence comprises a plurality of video frames which are acquired aiming at a target area, and the video frames contain targets;

Performing splicing treatment on all video frames in the image sequence to obtain an orthographic image and a digital surface model corresponding to the target area; the orthophoto map and the digital surface model are the same size.

The method comprises the steps of obtaining an image sequence, wherein the image sequence comprises a plurality of video frames acquired aiming at a target area, and the method comprises the following steps:

the calibration position is distributed and controlled in the target area;

image acquisition is carried out on the target area, so that a plurality of video frames aiming at the target area are obtained;

and after the step of performing stitching processing on all video frames in the image sequence to obtain the orthographic image and the digital surface model of the target area, the method further comprises the following steps:

the orthographic image and the digital surface model are corrected separately based on the position of the same calibration position in each video frame.

Wherein, carry on the class detection to every pixel in the orthographic image of the goal area, get the correspondent detection class of every pixel, including:

clipping the orthographic image to obtain a plurality of orthographic image subgraphs with preset sizes;

performing category detection on each pixel in the orthophoto image subgraph through the ground feature element identification model to obtain a detection category corresponding to each pixel; the ground feature element identification model comprises a coding module, a feature enhancement module and a decoding module which are sequentially cascaded;

The detection type of each pixel in each orthophoto image sub-image is used as the detection type of the pixel at the corresponding position in the orthophoto image.

The method for detecting the categories of the pixels in the orthophoto image subgraph through the ground feature element recognition model comprises the following steps:

carrying out feature extraction on each pixel in the orthophoto image subgraph through an encoding module to obtain a pixel feature map;

extracting the characteristics of the pixel characteristic map through a characteristic enhancement module to obtain a target characteristic map;

performing feature fusion on the target feature images corresponding to the pixels through a decoding module to obtain fusion feature images;

and determining the detection category of the pixel based on the fusion feature map corresponding to the pixel.

The coding module comprises a plurality of first feature extraction units which are sequentially cascaded, wherein each first feature extraction unit comprises a linear coding layer and a first feature extraction layer which are sequentially cascaded, and the other first feature extraction units comprise a block diagram splicing layer and a second feature extraction layer which are sequentially cascaded;

performing feature extraction on each pixel in the orthophoto image subgraph through an encoding module to obtain a pixel feature map, wherein the feature extraction comprises the following steps:

converting each pixel in the orthophoto image subgraph into one-dimensional data through a linear coding layer to obtain one-dimensional characteristics corresponding to each pixel;

Carrying out feature extraction on one-dimensional features of pixels through a first feature extraction layer to obtain a first feature map;

the first feature images of the pixels are spliced and fused through a block image splicing layer, so that a combined feature image is obtained;

and carrying out feature extraction on the combined feature images of the corresponding pixels of the second feature extraction layer to obtain a pixel feature image, wherein the size of the pixel feature image is smaller than that of the first feature image.

Wherein the feature enhancement module comprises a plurality of second feature extraction units which are cascaded in turn,

feature extraction is carried out on the pixel feature map through a feature enhancement module to obtain a target feature map, and the method comprises the following steps:

performing feature extraction and up-sampling on the pixel feature map through a first and second feature extraction unit to obtain a third feature map;

performing feature fusion on the third feature map and the first feature map through a second feature extraction unit to obtain a fourth feature map; the third feature map is the same size as the first feature map;

and carrying out feature extraction on the fourth feature map through a second feature extraction unit to obtain a target feature map.

The method for obtaining the fusion feature map comprises the steps of:

And interpolating the third feature map, the fourth feature map and the target feature map to the same size through a decoding module, and carrying out feature fusion to obtain a fusion feature map.

Wherein, based on the pixel category corresponding to each pixel in the orthographic image, determining the connected domain corresponding to each detection category in the target area comprises the following steps:

traversing all detection categories, and selecting one detection category as a target category;

assigning all pixels corresponding to the target category in the orthophotomap as a first identifier;

and determining connected domain information corresponding to the target category based on all pixels corresponding to the first identifier, wherein the connected domain information comprises connected domains.

The detection category at least comprises a road line category and a category to be detected; the road line category is a detection category of the road line, and the category to be detected is a detection category of the target to be detected;

mapping the position of each connected domain in the orthographic image to a digital surface model, determining height information of each connected domain, comprising:

mapping the position of the connected domain corresponding to the road line category in the orthographic image to a digital surface model, and determining a first height value corresponding to the road line;

mapping the position of the connected domain corresponding to the category to be detected in the orthophoto map to a digital surface model, and determining a second height value corresponding to the target to be detected;

And determining the height information of the object to be detected based on the difference value between the first height value and the second height value.

Wherein mapping the position of each connected domain in the orthographic image to the digital surface model, determining the height information of each connected domain, further comprising:

and associating the height information of the object to be detected with all pixels contained in the corresponding connected domain.

The connected domain information further comprises a connected domain number, an external rectangular frame and a total number of connected domains corresponding to the detection category;

mapping the position of the connected domain corresponding to the category to be detected in the orthographic image to a digital surface model, and determining a second height value corresponding to the target to be detected, wherein the method comprises the following steps:

in response to the number of the connected domains being not greater than the total number of the connected domains of the detection category corresponding to the connected domains, assigning 1 to the corresponding pixels of the connected domains in the circumscribed rectangular frame of the connected domains, and 0 to the other pixels except the connected domains in the circumscribed rectangular frame;

multiplying the assignment of each pixel in the external rectangular frame by the numerical value of the corresponding position in the digital surface model to obtain the corresponding height value of each pixel;

and determining a second height value of the target to be detected corresponding to the connected domain based on the height values of all pixels corresponding to the connected domain in the external rectangular frame.

The determining a second height value of the object to be detected corresponding to the connected domain based on the height values of all pixels corresponding to the connected domain in the external rectangular frame comprises the following steps:

and taking the average value of the height values of all pixels corresponding to the connected domain in the external rectangular frame as a second height value of the target to be detected corresponding to the connected domain.

Wherein, based on the height information of the connected domain and the position of each connected domain in the orthophoto map, a map model of the target area is constructed, comprising:

vectorizing detection categories corresponding to pixels in the orthographic image and height information corresponding to the connected domains to generate contour lines of the connected domains respectively; the contour lines corresponding to the connected domains are provided with data types, wherein the data types comprise point data, line data and surface data; the coordinate positions of the contour points contained in the contour lines belong to a spherical coordinate system;

responding to the data type corresponding to the connected domain as surface data, and converting the coordinate position of each contour point in the spherical coordinate system in the contour line corresponding to the connected domain into a plane coordinate position in the plane coordinate system;

and constructing a map model of the target area based on the plane coordinate positions of the contour points in the connected domain and the height information of the connected domain.

The method for constructing the map model of the target area based on the plane coordinate positions of the contour points in the connected domain and the height information of the connected domain comprises the following steps:

performing bottom surface modeling based on the plane coordinate positions of all contour points in the connected domain, and generating a bottom surface model;

determining the plane coordinate position of the vertex contour point based on the plane coordinate position of the contour point and the height information corresponding to the contour point;

performing top surface modeling based on the plane coordinate position of the vertex contour point to generate a top surface model;

performing side modeling based on the contour points and the vertex contour points to generate a side model;

and combining the bottom surface model, the top surface model and the side surface model to obtain a map model of the target area.

In order to solve the technical problems, a second technical scheme adopted by the invention is as follows: there is provided a terminal comprising a memory, a processor and a computer program stored in the memory and running on the processor, the processor being adapted to execute program data to carry out the steps of the above-described map model building method.

The beneficial effects of the invention are as follows: different from the situation of the prior art, the map model building method and the terminal provided by the invention comprise the following steps: acquiring an orthographic image and a digital surface model corresponding to a target area; performing category detection on each pixel in the orthophoto map of the target area to obtain a detection category corresponding to each pixel; determining a connected domain corresponding to each detection category in the target area based on the pixel category corresponding to each pixel in the orthographic image; mapping the position of each connected domain in the orthographic image to a digital surface model, and determining the height information of each connected domain; and constructing a map model of the target area based on the height information of the connected domains and the positions of the connected domains in the orthophoto map. According to the method and the device, the connected domain in the target area is obtained through category detection of each pixel in the orthographic image of the target area, the position of the ground object contained in the target area is determined, the height information of the ground object corresponding to the connected domain is determined based on the area data of the connected domain mapped in the digital surface model, and the map model of the target area is built through the position information and the height information of the ground object contained in the target area, so that the accuracy of the built map model is improved.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a schematic flow chart of a map model construction method provided by the invention;

FIG. 2 is a schematic diagram of an embodiment of a map model construction method provided by the present invention;

FIG. 3 is a flowchart illustrating an embodiment of step S1 in the map model construction method provided in FIG. 1;

FIG. 4 is an image of a target area provided by the present invention;

FIG. 5 is a flowchart illustrating an embodiment of step S2 in the map model construction method provided in FIG. 1;

FIG. 6 is a schematic structural diagram of an embodiment of a model for identifying feature elements provided by the present invention;

FIG. 7 is a flowchart illustrating an embodiment of step S4 in the map model construction method provided in FIG. 1;

FIG. 8 is a flowchart illustrating a step S4 of the map model construction method of FIG. 1 according to an embodiment;

FIG. 9 is a flowchart of a step S5 of the map model construction method provided in FIG. 1;

FIG. 10 is a flowchart illustrating a step S53 of the map model construction method of FIG. 9 according to an embodiment;

FIG. 11 is a schematic diagram of a map modeling apparatus according to an embodiment of the present invention;

FIG. 12 is a schematic diagram of a frame of an embodiment of a terminal provided by the present invention;

fig. 13 is a schematic diagram of a frame of an embodiment of a computer readable storage medium according to the present invention.

Detailed Description

The following describes the embodiments of the present application in detail with reference to the drawings.

In the following description, for purposes of explanation and not limitation, specific details are set forth such as the particular system architecture, interfaces, techniques, etc., in order to provide a thorough understanding of the present application.

The term "and/or" is herein merely an association relationship describing an associated object, meaning that there may be three relationships, e.g., a and/or B, may represent: a exists alone, A and B exist together, and B exists alone. In addition, the character "/" herein generally indicates that the front and rear associated objects are an "or" relationship. Further, "a plurality" herein means two or more than two.

In order to enable those skilled in the art to better understand the technical scheme of the present invention, the map model construction method provided by the present invention is described in further detail below with reference to the accompanying drawings and the detailed description.

Referring to fig. 1 and fig. 2, fig. 1 is a schematic flow chart of a map model construction method provided by the present invention; fig. 2 is a schematic diagram of an embodiment of a map model construction method provided by the present invention.

In this embodiment, a map model construction method is provided, and includes the following steps.

S1: an orthographic image and a digital surface model corresponding to the target region are acquired.

S2: and performing category detection on each pixel in the orthophoto map of the target area to obtain a detection category corresponding to each pixel.

S3: and determining the connected domain corresponding to each detection type in the target area based on the pixel type corresponding to each pixel in the orthographic image.

S4: mapping the position of each connected domain in the orthographic image to a digital surface model, and determining the height information of each connected domain.

S5: and constructing a map model of the target area based on the height information of the connected domains and the positions of the connected domains in the orthophoto map.

Referring to fig. 3, fig. 3 is a flowchart illustrating an embodiment of step S1 in the map model construction method provided in fig. 1.

Specifically, the step of acquiring the orthophoto map and the digital surface model corresponding to the target region in step S1 includes the following steps.

S11: and acquiring an image sequence, wherein the image sequence comprises a plurality of video frames which are acquired aiming at a target area, and the video frames contain targets.

Specifically, the calibration position is distributed and controlled in the target area; and acquiring images of the target area to obtain a plurality of video frames aiming at the target area.

In this embodiment, the image sequence may be a plurality of continuous video frames collected for the target area, or may be a plurality of discontinuous video frames collected for the target area.

Referring to fig. 4, fig. 4 is an image of a target area provided by the present invention.

In one embodiment, the target area is a road intersection area, and the intersection is taken as the target area and extends outwards by 200 meters from the center of the target area, for example, the target area is a rectangular frame. And (3) distributing and controlling the calibration positions in the target area, mainly distributing and controlling the calibration positions at the road edge positions and the road intersections in order not to influence traffic, and recording actual coordinates of the calibration positions. The calibration position is the position of the five-pointed star. The actual coordinates are coordinate positions in a spherical coordinate system. A route, such as a line within a rectangular box, is planned within the target area. And flying the target area along the planned route by the unmanned aerial vehicle, and continuously collecting images in the target area. And taking the acquired multiple continuous video frames acquired aiming at the target area as video frames.

S12: performing splicing treatment on all video frames in the image sequence to obtain an orthographic image and a digital surface model corresponding to the target area; the orthophoto map and the digital surface model are the same size.

In one embodiment, the Pix4D software is used to stitch all video frames in the image sequence and output an orthophoto map and a digital surface model.

Specifically, all video frames in the image sequence are input to Pix4D software, and a processing module in the Pix4D software selects a "3D map" template and outputs the template as an orthographic image, a DSM image, a 3D texture and point cloud data. In a specific embodiment, a project is newly built in the Pix4D software, all video frames contained in the image sequence are imported into the Pix4D software, the Pix4D software reads characteristic information in each video frame, and information such as longitude and latitude, altitude, pitch angle and the like when the unmanned aerial vehicle collects the video frames is obtained. And splicing all video frames in the image sequence through Pix4D software, and outputting an orthographic image and a digital surface model corresponding to the target area. The digital surface model refers to a ground elevation model comprising the heights of surface buildings, bridges, trees and the like. Wherein the digital surface model is also called DSM image.

In one embodiment, the orthographic image and the digital surface model are each corrected based on the position of the same calibration location in each video frame. Specifically, the same calibration positions in all video frames are aligned and overlapped, so that the contents in different video frames are in the same coordinate system, and the position accuracy of each target in the obtained orthophoto map is improved.

Specifically, the step of detecting the class corresponding to each pixel in step S2 includes the following steps.

In this embodiment, since the size of the orthographic image corresponding to the target area is more than one hundred million pixels, the feature element recognition model cannot directly detect the type of each pixel in the orthographic image.

In order to facilitate detection of each pixel in the orthophoto map, the orthophoto map needs to be cropped to obtain a plurality of orthophoto sub-maps with preset sizes. Each of the orthophoto map has position information in the orthophoto map.

In an embodiment, the classification detection is performed on each pixel in the orthophoto image sub-graph through the ground feature element recognition model, so as to obtain a detection classification corresponding to each pixel. The detection category is one of a road line category, a building category, a green belt category, a sidewalk category, a safety island category, a channelized island category, a vegetation category and a water body category.

And splicing all the orthophoto image subgraphs to obtain an orthophoto image, and taking the detection category of each pixel in each orthophoto image subgraph as the detection category of the pixel at the corresponding position in the orthophoto image.

In one embodiment, the training steps of the feature element recognition model are as follows.

Acquiring a plurality of orthophoto training pictures with preset sizes, wherein each pixel in the orthophoto training pictures has a corresponding labeling category; carrying out category prediction on each pixel in the orthophoto training diagram through an initial network to obtain probability values of the pixels belonging to each preset category; determining a preset category corresponding to the maximum probability value as the preset category of the pixel; and training the initial network based on error values between preset categories and labeling categories corresponding to pixels in the orthographic image training diagram to obtain a ground feature element identification model. Preset categories include, but are not limited to, roadway line categories, building categories, green belt categories, sidewalk categories, security island categories, channelized island categories, vegetation categories, and water categories.

Referring to fig. 5 and 6, fig. 5 is a flowchart illustrating an embodiment of step S2 in the map model construction method provided in fig. 1; fig. 6 is a schematic structural diagram of an embodiment of a feature element recognition model provided by the present invention.

In this embodiment, the feature element recognition model includes a coding module, a feature enhancement module, and a decoding module that are sequentially cascaded. Wherein, the feature element identification model adopts a Cascade-Encoder-Decoder framework.

S21: and carrying out feature extraction on each pixel in the orthophoto image subgraph through the coding module to obtain a pixel feature map.

In an embodiment, the encoding module includes a plurality of first feature extraction units that are sequentially cascaded, the first feature extraction unit includes a linear encoding layer and a first feature extraction layer that are sequentially cascaded, and the other first feature extraction units include a block diagram stitching layer and a second feature extraction layer that are sequentially cascaded.

Specifically, each pixel in an orthophoto image sub-image is converted into one-dimensional data through a linear coding layer, and one-dimensional characteristics corresponding to each pixel are obtained; carrying out feature extraction on one-dimensional features of pixels through a first feature extraction layer to obtain a first feature map; the first feature images of the pixels are spliced and fused through a block image splicing layer, so that a combined feature image is obtained; and carrying out feature extraction on the combined feature images of the corresponding pixels of the second feature extraction layer to obtain a pixel feature image, wherein the size of the pixel feature image is smaller than that of the first feature image.

In one embodiment, the coding module may be a Swin transducer structure. The coding module comprises four first feature extraction units which are sequentially cascaded, wherein the first feature extraction unit comprises a linear coding layer and two first feature extraction layers which are sequentially cascaded, and the second first feature extraction unit comprises a block diagram splicing layer and two second feature extraction layers; the third first feature extraction unit comprises a block diagram splicing layer and eighteen second feature extraction layers; the fourth first feature extraction unit includes a block diagram stitching layer and two second feature extraction layers. The first feature extraction layer and the second feature extraction layer may be one of a multi-head self-attention module of a window, a multi-head self-attention module of a moving window, or a multi-layer perceptron. The first feature extraction layer and the second feature extraction layer may be the same or different. The block diagram splicing layer is used for splicing and fusing two adjacent original feature diagrams and deepening the number of channels.

The size of the orthophoto image sub-image is H.times.W, firstly, a layer convolution is adopted to code the orthophoto image sub-image into a characteristic image of H/4*W/4.times.16.times.C according to the size of 1/4, wherein C represents the channel number of the input image. The size of the feature map output by the first feature extraction unit is 1/4 of the size of the feature map after the orthographic image sub-image encoding; the size of the feature map output by the second first feature extraction unit is 1/8 of the size of the feature map after the orthographic image sub-image encoding; the size of the feature map output by the third first feature extraction unit is 1/16 of the size of the feature map after the orthographic image sub-image encoding; the size of the feature map output by the fourth first feature extraction unit is 1/32 of the size of the feature map after the orthophoto sub-image encoding.

S22: and carrying out feature extraction on the pixel feature map through a feature enhancement module to obtain a target feature map.

In an embodiment, the feature enhancement module includes a plurality of second feature extraction units cascaded in sequence. The feature enhancement module may be a neg module.

Specifically, a first feature extraction unit and a second feature extraction unit are used for carrying out feature extraction and up-sampling on the pixel feature map to obtain a third feature map; performing feature fusion on the third feature map and the first feature map through a second feature extraction unit to obtain a fourth feature map; the third feature map is the same size as the first feature map; and carrying out feature extraction on the fourth feature map through a second feature extraction unit to obtain a target feature map.

In a specific embodiment, the feature enhancement module is a FPN (Feature Pyramid Networks) structure. The feature enhancement module comprises four second feature extraction units which are sequentially cascaded. The first and second feature extraction units perform feature extraction on the pixel feature images and perform up-sampling processing to obtain third feature images; the second feature extraction unit is used for splicing the third feature image and the feature image output by the third first feature extraction unit, and extracting the features to obtain a first sub-feature image; after the first sub-feature image is subjected to up-sampling processing, the first sub-feature image and the feature image output by the second first feature extraction unit are spliced through a third second feature extraction unit, and feature extraction is carried out to obtain a second sub-feature image; and after the second sub-feature image is subjected to up-sampling processing, the second sub-feature image and the feature image output by the first feature extraction unit are spliced and feature-extracted through a fourth second feature extraction unit to obtain a target feature image. Wherein the convolution kernel of the second feature extraction unit includes, but is not limited to 3*3, the number of channels of the second feature extraction unit is 256.

S23: and carrying out feature fusion on the target feature images corresponding to the pixels through a decoding module to obtain fusion feature images.

In an embodiment, the structure of the decoding module and the structure of the feature enhancement module may be the same. The decoding module may also have an FPN structure.

Specifically, the decoding module includes four third feature extraction units that are sequentially cascaded. The first third feature extraction unit performs feature extraction on the target feature map to obtain a third sub-feature map; the second third feature extraction unit is used for splicing the third sub-feature image and the feature image output by the third second feature extraction unit, and extracting features to obtain a fourth sub-feature image; splicing the fourth sub-feature image and the feature image output by the second feature extraction unit through a third feature extraction unit, and extracting features to obtain a fifth sub-feature image; and splicing the fifth sub-feature image and the feature image output by the first and second feature extraction units through a fourth and third feature extraction unit, and extracting features to obtain a fusion feature image. Wherein the convolution kernel of the third feature extraction unit includes, but is not limited to 1*1, the number of channels of the third feature extraction unit is 128. In an embodiment, the dimensions of the target feature map, the third sub-feature map, the feature map output by the third second feature extraction unit, the feature map output by the second feature extraction unit of the fourth sub-feature map, the fifth sub-feature map, and the feature map output by the first second feature extraction unit are interpolated to the same dimension, and then corresponding splicing and fusion are performed. For example, the size may be 1/8H 1/8W.

S24: and determining the detection category of the pixel based on the fusion feature map corresponding to the pixel.

Specifically, the detection category of the pixel is obtained by convolving the fusion feature map corresponding to the pixel.

And determining a plurality of detection categories corresponding to the orthophoto map according to the detection categories of the pixels in the orthophoto map. The detection types of the pixels in the orthophoto image subgraph are counted, and the counted detection types are taken as a plurality of detection types corresponding to the orthophoto image subgraph. The detection category may be a road line category, a building category, a green belt category, a sidewalk category, a security island category, a channelized island category, a vegetation category, and a water body category.

Specifically, the step of determining the connected domain corresponding to each detection category in the target area in step S3 specifically includes the following steps.

Traversing all detection categories, and selecting one detection category as a target category; assigning all pixels corresponding to the target category in the orthophotomap as a first identifier; and determining connected domain information corresponding to the target category based on all pixels corresponding to the first identifier, wherein the connected domain information comprises connected domains.

In a specific embodiment, all pixels corresponding to the target class in the orthographic image are assigned as the first identifier, and all pixels except the pixels corresponding to the target class in the orthographic image are assigned as the second identifier, that is, the orthographic image is binarized to obtain a mask image corresponding to the target class. Each mask map has at least one corresponding connected domain. In one embodiment, the first identifier is 1 and the second identifier is 0.

And taking the connected region composed of the first identifiers as the connected region according to the mask map corresponding to the target category. Numbering the corresponding connected domains in the mask map of the target class to obtain the connected domain numbers of the connected domains. And assigning the connected domain numbers to all pixels contained in the connected domain, and counting the total number of the connected domains in the mask map corresponding to the target class. And determining an external rectangular frame of the connected domain according to the connected domain. Specifically, the smallest circumscribed rectangular frame of the connected domain is taken as the circumscribed rectangular frame of the connected domain.

Specifically, the step of determining the height information of each connected domain in step S4 specifically includes the following steps.

In an embodiment, the connected domain information includes a connected domain, a connected domain number, an external rectangular frame, and a total number of connected domains corresponding to the detection category.

Referring to fig. 7 and 8, fig. 7 is a flowchart illustrating an embodiment of step S4 in the map model construction method provided in fig. 1; fig. 8 is a flowchart of a specific embodiment of step S4 in the map model construction method provided in fig. 1.

Because the road line is arranged on the ground, the height information of the road line corresponding to the road line type can be determined as the height information of the ground, and then the height information of the object to be detected is determined based on the difference value between the height value of the object to be detected corresponding to other types and the height information of the ground.

S41: mapping the position of the connected domain corresponding to the road line category in the orthographic image to a digital surface model, and determining a first height value corresponding to the road line.

In particular, since the orthophotos and the digital surface models are identical in size. And multiplying the identifiers of the pixels in the connected domain in the mask diagram corresponding to the road line category by the height value of the corresponding position in the digital surface model to obtain the height value corresponding to the pixels in the connected domain. The height values corresponding to the pixels of all the connected domains corresponding to the road line category can be averaged to obtain a first height value corresponding to the road line.

S42: mapping the position of the connected domain corresponding to the category to be detected in the orthographic image to a digital surface model, and determining a second height value corresponding to the target to be detected.

Specifically, in response to the connected domain number of the connected domain not being greater than the total number of connected domains of the detection class corresponding to the connected domain, all pixels in the connected domain are assigned to 1, and all other pixels except the connected domain in the circumscribed rectangular frame corresponding to the connected domain are assigned to 0. And multiplying the assignment of each pixel corresponding to the connected domain in the external rectangular frame by the numerical value of the corresponding position in the digital surface model to obtain the height value corresponding to each pixel.

And determining a second height value of the target to be detected corresponding to the connected domain based on the height values of all pixels corresponding to the connected domain in the external rectangular frame. In an embodiment, an average value of height values of all pixels corresponding to the connected domain in the external rectangular frame is used as a second height value of the target to be detected corresponding to the connected domain.

And (3) continuing to calculate the height information of the other connected domain corresponding to the category to be detected through the step until the second height values of all the connected domains corresponding to the category to be detected are calculated.

S43: and determining the height information of the object to be detected based on the difference value between the first height value and the second height value.

Specifically, a difference value between the second height value corresponding to the connected domain and the first height value corresponding to the road line is determined as the height information of the object to be detected corresponding to the connected domain.

In an embodiment, the height information of the object to be detected is associated with all pixels included in the corresponding connected domain.

And calculating the height information of each connected domain corresponding to each detection type in the orthographic image by the method.

In an embodiment, a matrix with a height value of 0, which is the same as the size of the orthophoto map, is newly created to save the height information of all connected domains corresponding to each detection category. Specifically, the number of the matrices may be the same as the number of the detection categories in the orthographic image, or the number of the matrices may be one, that is, the height values of the connected domains of all the detection categories corresponding to the orthographic image are stored in the same matrix.

Referring to fig. 9, fig. 9 is a flowchart of an embodiment of step S5 in the map model construction method provided in fig. 1.

Specifically, the step of constructing the map model of the target region in step S5 based on the height information of the connected domains and the positions of the connected domains in the orthophoto map specifically includes the following steps.

S51: and vectorizing detection types corresponding to the pixels in the orthographic image and height information corresponding to the connected domains to generate contour lines of the connected domains.

Specifically, according to the detection category corresponding to each pixel in the orthographic image, the target corresponding to the connected domain to which the pixel belongs can be determined. I.e. all objects contained in the orthophoto map can be determined by the detection class for all pixels in the orthophoto map. For example, if the detection category is a road line category, it is determined that the target corresponding to the road line category is a road line, and if the detection category is a building category, it is determined that the target corresponding to the building category is a building.

The vectorization process is to convert all the targets detected in the orthophotomap and the height results corresponding to the targets from a raster file to a vector file in a three-dimensional data format, for example, shp format or geojson format. And vectorizing all targets detected in the orthophotomap and height results corresponding to the targets to obtain point data, line data and surface data. For example, sidewalks, buildings and safety islands are all composed of vector surfaces, i.e. the data type of the vectorized connected domain is surface data. The markers are composed of vector points, namely the data type of the vectorized connected domain is point data. The track lines are composed of vector lines, namely the data types of the vectorized connected domains are line data.

Specifically, the contour lines corresponding to the connected domains have data types, wherein the data types comprise point data, line data and surface data; the coordinate positions of the contour points included in the contour lines belong to a spherical coordinate system.

The data type of the contour line, the object, the corresponding height information and other information are all stored in the vectorized contour line and used as characteristic values of the contour line.

The vectorization operation is implemented using a third party library, such as by GDAL (Geospatial Data Abstraction Library ).

In one embodiment, the format of the vectorized vector file is determined according to the dimensions of the map model to be built. For example, in 2D and 2.5D scenes, the format of the vector file needs to be converted into shp format, and then the vector file is directly displayed through the shp format; whereas in a 3D scene model data presentation must be used. The shp format or the geojson format of the vector file is converted into a grid body, and then the grid body is stored, and then the GDAL is used for reading the vector file and modeling the VTK library.

In this embodiment, the map model is mainly constructed for the surface data, so the format of the vectorization file of the contour line of the connected domain obtained in step S51 is converted into a planar mesh or a columnar mesh.

That is, the vector file corresponding to the orthographic image is obtained by vectorizing the detection type corresponding to each pixel in the orthographic image and the height information corresponding to each connected domain, the layers of the detection types are stored in the vector file, the contour lines of all the corresponding connected domains are stored in the layers of the detection types, and each contour line has a corresponding characteristic value.

S52: and converting the coordinate position of each contour point in the spherical coordinate system in the contour line corresponding to the connected domain into the plane coordinate position in the plane coordinate system in response to the data type corresponding to the connected domain being the face data.

Specifically, whether a vector file corresponding to an orthophoto map is opened or not is detected, if a popup window interface is displayed on an interface, the map layers of each detection class in the vector file and all corresponding connected domains are stored in the map layers of each detection class are directly read.

Since the stereoscopic map model is constructed based on the face data in the present embodiment. Therefore, whether the data type corresponding to the connected domain is the surface data is judged first, and if the data type corresponding to the connected domain is the surface data, the coordinate position of each contour point in the spherical coordinate system in the contour line corresponding to the connected domain is converted into the plane coordinate position in the plane coordinate system. For example, the plane coordinate position is UTM (Universal Transverse Mercator Grid System, universal transverse ink card grid system) projection coordinates.

S53: and constructing a map model of the target area based on the plane coordinate positions of the contour points in the connected domain and the height information of the connected domain.

Referring to fig. 10, fig. 10 is a flowchart illustrating an embodiment of step S53 in the map model construction method provided in fig. 9.

S531: and carrying out bottom surface modeling based on the plane coordinate positions of all contour points in the connected domain, and generating a bottom surface model.

Specifically, the bottom surface model is generated by performing bottom surface modeling through the VTK Delaunay 2D based on the plane coordinate positions of all the contour points corresponding to the connected domain. If only a planar grid body needs to be constructed, the plane coordinate positions of all contour points corresponding to the connected domain are used as endpoint data of the vtkOBJWriter and constraint data to carry out bottom surface modeling, and a bottom surface model constructed by all the contour points corresponding to the connected domain is stored as an obj file.

S532: and determining the plane coordinate position of the vertex contour point based on the plane coordinate position of the contour point and the height information corresponding to the connected domain.

Specifically, the plane coordinate position of the contour point and the height information of the connected domain are added to obtain the plane coordinate position of the vertex contour point.

S533: and carrying out top surface modeling based on the plane coordinate positions of the vertex contour points to generate a top surface model.

Specifically, if a columnar grid body needs to be constructed, top surface modeling is performed based on the plane coordinate positions of the vertex contour points, and a top surface model is generated. Specifically, plane coordinate positions corresponding to the vertex contour points are read, the plane coordinate positions of the vertex contour points are used as endpoint data and constraint data of vtkDallaunay 2D to conduct top surface modeling, and a top surface model constructed by all contour points corresponding to the connected domain is stored as obj files.

S534: and carrying out side modeling based on the contour points and the vertex contour points to generate a side model.

Specifically, all contour points on the same surface in the top surface model and the bottom surface model are traversed, two adjacent top surface contour points in the top surface model and two adjacent contour points in the bottom surface model are selected to form triangular surfaces, and a corresponding side surface model is generated based on all the generated triangular surfaces.

S535: and combining the bottom surface model, the top surface model and the side surface model to obtain a three-dimensional map model of the target area.

Specifically, the vtkphenddpolydata is used to combine the top model, the bottom model and the side model into one model to obtain a map model of the target area. The model is stored as obj file using vtkOBJWriter.

According to the map model construction method provided by the embodiment, an orthographic image and a digital surface model corresponding to a target area are obtained; performing category detection on each pixel in the orthophoto map of the target area to obtain a detection category corresponding to each pixel; determining a connected domain corresponding to each detection category in the target area based on the pixel category corresponding to each pixel in the orthographic image; mapping the position of each connected domain in the orthographic image to a digital surface model, and determining the height information of each connected domain; and constructing a map model of the target area based on the height information of the connected domains and the positions of the connected domains in the orthophoto map. The method comprises the steps of obtaining a connected domain in a target area by detecting each pixel in an orthographic image of the target area, determining the position of a ground object contained in the target area, determining the height information of the ground object corresponding to the connected domain based on the area data of the connected domain mapped in a digital surface model, and constructing a map model of the target area according to the position information and the height information of the ground object contained in the target area, so that the accuracy of the constructed map model is improved.

Referring to fig. 11, fig. 11 is a schematic diagram of a map modeling apparatus according to an embodiment of the present invention. The present embodiment provides a map model construction apparatus 60, and the map model construction apparatus 60 includes an acquisition module 61, a category detection module 62, an analysis module 63, a calculation module 64, and a construction module 65.

The acquisition module 61 is configured to acquire an orthophoto map and a digital surface model corresponding to the target region.

The class detection module 62 is configured to perform class detection on each pixel in the orthophoto map of the target area, so as to obtain a detection class corresponding to each pixel.

The analysis module 63 is configured to determine a connected domain corresponding to each detection class in the target area based on the pixel class corresponding to each pixel in the orthophoto map.

The calculation module 64 is configured to map the position of each connected domain in the orthophotomap to the digital surface model, and determine the height information of each connected domain.

The construction module 65 is configured to construct a map model of the target area based on the height information of the connected domains and the positions of the connected domains in the orthophoto map.

According to the map model building device provided by the embodiment, an orthographic image and a digital surface model corresponding to a target area are obtained through an obtaining module; the class detection module detects the class of each pixel in the orthophoto map of the target area to obtain a detection class corresponding to each pixel; the analysis module determines a connected domain corresponding to each detection category in the target area based on the pixel category corresponding to each pixel in the orthographic image; the calculation module maps the position of each connected domain in the orthophoto map to the digital surface model, and determines the height information of each connected domain; the construction module constructs a map model of the target area based on the height information of the connected domains and the positions of the connected domains in the orthophoto map. The method comprises the steps of obtaining a connected domain in a target area by detecting each pixel in an orthographic image of the target area, determining the position of a ground object contained in the target area, determining the height information of the ground object corresponding to the connected domain based on the area data of the connected domain mapped in a digital surface model, and constructing a map model of the target area according to the position information and the height information of the ground object contained in the target area, so that the accuracy of the constructed map model is improved.

Referring to fig. 12, fig. 12 is a schematic diagram of a frame of an embodiment of a terminal of the present application. The terminal 80 comprises a memory 81 and a processor 82 coupled to each other, the processor 82 being adapted to execute program instructions stored in the memory 81 for implementing the steps of any of the map model building method embodiments described above. In one particular implementation scenario, terminal 80 may include, but is not limited to: the microcomputer, server, and the terminal 80 may also include mobile devices such as a notebook computer and a tablet computer, which are not limited herein.

In particular, the processor 82 is configured to control itself and the memory 81 to implement the steps of any of the map model building method embodiments described above. The processor 82 may also be referred to as a CPU (Central Processing Unit ). The processor 82 may be an integrated circuit chip having signal processing capabilities. The processor 82 may also be a general purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), a Field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware components. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. In addition, the processor 82 may be commonly implemented by an integrated circuit chip.

Referring to fig. 13, fig. 13 is a schematic diagram illustrating an embodiment of a computer readable storage medium 90 according to the present application. The computer readable storage medium 90 stores program instructions 901 executable by a processor, the program instructions 901 for implementing the steps of any one of the map model building method embodiments described above.

In some embodiments, functions or modules included in an apparatus provided by the embodiments of the present disclosure may be used to perform a method described in the foregoing method embodiments, and specific implementations thereof may refer to descriptions of the foregoing method embodiments, which are not repeated herein for brevity.

The foregoing description of various embodiments is intended to highlight differences between the various embodiments, which may be the same or similar to each other by reference, and is not repeated herein for the sake of brevity.

In the several embodiments provided in the present application, it should be understood that the disclosed methods and apparatus may be implemented in other manners. For example, the apparatus embodiments described above are merely illustrative, e.g., the division of modules or units is merely a logical functional division, and there may be additional divisions of actual implementation, e.g., units or components may be combined or integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be an indirect coupling or communication connection via some interfaces, devices or units, which may be in electrical, mechanical, or other forms.

In addition, each functional unit in each embodiment of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.

The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied essentially or in part or all or part of the technical solution contributing to the prior art or in the form of a software product stored in a storage medium, including several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) or a processor (processor) to perform all or part of the steps of the methods of the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

The foregoing is only the embodiments of the present invention, and therefore, the patent protection scope of the present invention is not limited thereto, and all equivalent structures or equivalent flow changes made by the content of the present specification and the accompanying drawings, or direct or indirect application in other related technical fields, are included in the patent protection scope of the present invention.

Claims

1. The map model construction method is characterized by comprising the following steps of:

mapping the position of each connected domain in the orthographic image to the digital surface model, and determining the height information of each connected domain;

and constructing a map model of the target area based on the height information of the connected domains and the positions of the connected domains in the orthographic image.

2. The method for constructing a map model according to claim 1, wherein,

The acquiring the orthographic image and the digital surface model corresponding to the target area comprises the following steps:

acquiring an image sequence, wherein the image sequence comprises a plurality of video frames which are acquired aiming at the target area, and the video frames contain targets;

performing splicing processing on all the video frames in the image sequence to obtain the orthographic image and the digital surface model corresponding to the target area; the orthographic image and the digital surface model are the same size.

3. The method for constructing a map model according to claim 2, wherein,

the acquiring an image sequence, the image sequence including a plurality of video frames acquired for the target region, includes:

the calibration position is distributed and controlled in the target area;

after the step of performing stitching processing on all the video frames in the image sequence to obtain the orthographic image and the digital surface model of the target area, the method further comprises:

and correcting the orthographic image and the digital surface model respectively based on the position of the same calibration position in each video frame.

4. The method for constructing a map model according to claim 1, wherein,

the step of performing class detection on each pixel in the orthographic image of the target area to obtain a detection class corresponding to each pixel includes:

cutting the orthographic image to obtain a plurality of orthographic image subgraphs with preset sizes;

performing category detection on each pixel in the orthophoto image subgraph through a ground feature element identification model to obtain the detection category corresponding to each pixel; the ground feature element identification model comprises a coding module, a characteristic enhancement module and a decoding module which are sequentially cascaded;

and taking the detection category of each pixel in each orthographic image sub-image as the detection category of the pixel at the corresponding position in the orthographic image.

5. The method for constructing a map model according to claim 4, wherein,

the step of performing class detection on each pixel in the orthophoto image sub-graph through the ground feature element recognition model to obtain the detection class corresponding to each pixel, includes:

performing feature extraction on each pixel in the orthophoto image subgraph through the coding module to obtain a pixel feature map;

Extracting the characteristics of the pixel characteristic map through the characteristic enhancement module to obtain a target characteristic map;

performing feature fusion on the target feature map corresponding to the pixel through the decoding module to obtain a fusion feature map;

6. The map model construction method according to claim 5, wherein the encoding module includes a plurality of first feature extraction units that are sequentially cascaded, a first one of the first feature extraction units includes a linear encoding layer and a first feature extraction layer that are sequentially cascaded, and the other first feature extraction units include a block map stitching layer and a second feature extraction layer that are sequentially cascaded;

the step of extracting features of each pixel in the orthophoto image sub-graph by the encoding module to obtain a pixel feature graph includes:

converting each pixel in the orthophoto image subgraph into one-dimensional data through the linear coding layer to obtain one-dimensional characteristics corresponding to each pixel;

performing feature extraction on one-dimensional features of the pixels through the first feature extraction layer to obtain a first feature map;

The first feature images of the pixels are spliced and fused through the block image splicing layer, so that a combined feature image is obtained;

and carrying out feature extraction on the combined feature map corresponding to the pixels through the second feature extraction layer to obtain the pixel feature map, wherein the size of the pixel feature map is smaller than that of the first feature map.

7. The map model construction method according to claim 6, wherein the feature enhancement module includes a plurality of second feature extraction units that are sequentially cascaded,

the feature extraction is performed on the pixel feature map by the feature enhancement module to obtain a target feature map, including:

performing feature extraction and up-sampling on the pixel feature map through a first feature extraction unit and a second feature extraction unit to obtain a third feature map;

performing feature extraction on the fourth feature map through a second feature extraction unit to obtain the target feature map;

the step of performing feature fusion on the target feature map corresponding to the pixel through the decoding module to obtain a fusion feature map, including:

And interpolating the third feature map, the fourth feature map and the target feature map to the same size through the decoding module and carrying out feature fusion to obtain a fusion feature map.

8. The method for constructing a map model according to claim 1, wherein,

the determining, based on the pixel class corresponding to each pixel in the orthographic image, the connected domain corresponding to each detection class in the target area includes:

traversing all the detection categories, and selecting one detection category as a target category;

assigning all the pixels corresponding to the target class in the orthographic image as a first identifier;

and determining connected domain information corresponding to the target category based on all the pixels corresponding to the first identifier, wherein the connected domain information comprises the connected domain.

9. The map model construction method according to claim 8, wherein the detection category includes at least a road route category and a category to be detected; the road line category is a detection category of the road line, and the category to be detected is a detection category of the target to be detected;

mapping the position of each connected domain in the orthographic image to the digital surface model, and determining the height information of each connected domain comprises the following steps:

Mapping the position of the connected domain corresponding to the road line category in the orthographic image to the digital surface model, and determining a first height value corresponding to the road line;

mapping the position of the connected domain corresponding to the category to be detected in the orthographic image to the digital surface model, and determining a second height value corresponding to the target to be detected;

10. The method of claim 9, wherein,

mapping the position of each connected domain in the orthographic image to the digital surface model, and determining the height information of each connected domain, and further comprising:

and correlating the height information of the object to be detected with all the pixels contained in the corresponding connected domain.

11. The map model construction method according to claim 9, wherein the connected domain information further includes a connected domain number, an circumscribed rectangular frame, and a total number of connected domains corresponding to the detection category;

mapping the position of the connected domain corresponding to the category to be detected in the orthographic image to the digital surface model, and determining a second height value corresponding to the target to be detected, including:

In response to the number of the connected domains not being larger than the total number of the connected domains of the detection categories corresponding to the connected domains, the pixels corresponding to the connected domains in an external rectangular frame of the connected domains are all assigned to 1, and the pixels except the connected domains in the external rectangular frame are all assigned to 0;

multiplying the assigned value of each pixel in the circumscribed rectangular frame by the numerical value of the corresponding position in the digital surface model to obtain a height value corresponding to each pixel;

and determining a second height value of the object to be detected corresponding to the connected domain based on the height values of all the pixels corresponding to the connected domain in the circumscribed rectangular frame.

12. The method for constructing a map model according to claim 11, wherein,

the determining, based on the height values of all the pixels corresponding to the connected domain in the circumscribed rectangular frame, a second height value of the target to be detected corresponding to the connected domain includes:

and taking the average value of the height values of all the pixels corresponding to the connected domain in the circumscribed rectangular frame as a second height value of the target to be detected corresponding to the connected domain.

13. The method for constructing a map model according to claim 1, wherein,

The constructing a map model of the target area based on the height information of the connected domain and the position of each connected domain in the orthographic image includes:

vectorizing detection categories corresponding to the pixels in the orthophoto map and height information corresponding to the connected domains to generate contour lines of the connected domains respectively; the contour lines corresponding to the connected domains are provided with data types, wherein the data types comprise point data, line data and surface data; the coordinate positions of the contour points contained in the contour lines belong to a spherical coordinate system;

converting the coordinate position of each contour point in the spherical coordinate system in the contour line corresponding to the connected domain into a plane coordinate position in a plane coordinate system in response to the data type corresponding to the connected domain being the surface data;

and constructing a map model of the target area based on the plane coordinate position of each contour point in the connected domain and the height information of the connected domain.

14. The method of claim 13, wherein,

the constructing a map model of the target area based on the plane coordinate position of each contour point in the connected domain and the height information of the connected domain includes:

Performing bottom surface modeling based on the plane coordinate positions of the contour points in the connected domain, and generating a bottom surface model;

determining the plane coordinate position of the vertex contour point based on the plane coordinate position of the contour point and the height information corresponding to the connected domain;

and combining the bottom surface model, the top surface model and the side surface model to obtain the map model of the target area.

15. A terminal comprising a memory, a processor and a computer program stored in the memory and running on the processor, the processor being configured to execute program data to implement the steps of the map model construction method according to any one of claims 1 to 14.