CN113887321A - Method, equipment and storage medium for generating lane-level simulation road network - Google Patents
Method, equipment and storage medium for generating lane-level simulation road network Download PDFInfo
- Publication number
- CN113887321A CN113887321A CN202111056068.8A CN202111056068A CN113887321A CN 113887321 A CN113887321 A CN 113887321A CN 202111056068 A CN202111056068 A CN 202111056068A CN 113887321 A CN113887321 A CN 113887321A
- Authority
- CN
- China
- Prior art keywords
- lane
- sample
- street view
- image
- recognition model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000004088 simulation Methods 0.000 title claims abstract description 38
- 238000003860 storage Methods 0.000 title claims abstract description 11
- 238000012549 training Methods 0.000 claims description 36
- 230000004888 barrier function Effects 0.000 claims description 4
- 230000004927 fusion Effects 0.000 claims description 3
- 230000008569 process Effects 0.000 description 5
- 230000008901 benefit Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000009123 feedback regulation Effects 0.000 description 2
- 238000004445 quantitative analysis Methods 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 239000004575 stone Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention relates to the field of simulated road networks, in particular to a method, equipment and a storage medium for generating a lane-level simulated road network. The method comprises the steps of firstly obtaining an image corresponding to the street view and a map of the position of the street view, applying an image recognition algorithm to the street view image, recognizing lane information in the street view image, equivalently extracting lanes from the street view image, and then matching the extracted lanes to the map to obtain the simulated road network. This makes it possible to obtain information about the individual lanes from the map as a whole. Compared with the method of manually acquiring lane information on the spot, the method of the invention reduces the workload on one hand and can improve the accuracy of the acquired simulated road network on the other hand. In addition, compared with the prior art in which lane information is manually collected, the image recognition of the method is high in speed, so that the method can update the simulation road network in time.
Description
Technical Field
The invention relates to the field of simulated road networks, in particular to a method, equipment and a storage medium for generating a lane-level simulated road network.
Background
With the development of cities, the urban road network is continuously enlarged and the traffic flow is increased rapidly, and the requirements of urban-level traffic simulation on the lane-level road network are rapidly increased. The method has the advantages that the running state of the road network is accurately simulated and evaluated, the quantitative analysis and the feedback regulation are carried out on the road network, and the method is a key and difficult point for realizing the technologies such as network signal control evolution, dynamic traffic distribution and the like. The traffic simulation road network is an excellent tool for finely designing and quantitatively evaluating the traffic condition of vehicles on the lanes. Most of the existing simulation road networks acquire lane information manually on site and then draw the lane information on a map manually to obtain the simulation road network containing the lane information, but the manual drawing of the simulation road network increases the labor cost on one hand, and the condition that data is lost and the collection is missed exists when a driver collects the lane information, so that the accuracy of the drawn simulation road network is reduced, namely the simulation road network deviates from the actual road network, and the use of the simulation road network is further influenced.
In summary, the simulated road network obtained by the prior art is low in accuracy.
Thus, there is a need for improvements and enhancements in the art.
Disclosure of Invention
In order to solve the technical problems, the invention provides a method, equipment and a storage medium for generating a lane-level simulation road network, and solves the problem of low accuracy of the simulation road network obtained in the prior art.
In order to achieve the purpose, the invention adopts the following technical scheme:
in a first aspect, the present invention provides a method for generating a lane-level simulation road network, wherein the method comprises:
obtaining street view images and coordinate information corresponding to the street view images;
applying an image recognition model to the street view image to perform lane recognition to obtain lane information corresponding to the street view image;
and fusing the lane information and the map to obtain a simulated road network.
In one implementation, the applying an image recognition model to the street view image to perform lane recognition to obtain lane information corresponding to the street view image includes:
acquiring a sample lane and a sample street view picture;
constructing an image recognition model;
training the image recognition model according to the sample lane and the sample street view picture to obtain the trained image recognition model;
and performing lane recognition on the image recognition model after the street view image is trained to obtain lane information in the street view image.
In one implementation, the training the image recognition model according to the sample lane and the sample street view picture to obtain the trained image recognition model includes:
obtaining a sample lane line corresponding to the sample lane according to the sample lane;
and training the image recognition model according to the sample lane line and the sample street view picture to obtain the trained image recognition model.
In one implementation, the training the image recognition model according to the sample lane line and the sample street view picture to obtain the trained image recognition model includes:
obtaining a sample grid lane line in the sample lane lines according to the sample lane lines;
obtaining the sample street view picture containing the grid lane lines in the sample street view picture according to the sample street view picture;
and training the image recognition model according to the sample grid lane line and the sample street view picture containing the grid lane line to obtain the trained image recognition model.
In one implementation, the training the image recognition model according to the sample grid lane line and the sample street view picture containing the grid lane line to obtain the trained image recognition model includes:
obtaining a sample bus lane in the sample lanes according to the sample lanes;
obtaining the sample bus grid lane lines corresponding to the sample bus lanes in the sample grid lane lines according to the sample grid lane lines;
obtaining a sample bus street view picture containing the sample bus grid lane line in the sample street picture according to the sample street picture;
and training the image recognition model according to the sample bus grid lane line and the sample bus street view picture to obtain the trained image recognition model.
In one implementation, the training the image recognition model according to the sample lane line and the sample street view picture to obtain the trained image recognition model includes:
obtaining sample lane lines positioned at two sides of the barrier in the sample lane lines according to the sample lane lines;
removing a sample lane line on one side of the obstacle;
obtaining the sample street view picture containing the barrier in the sample street view picture according to the sample street view picture;
and training the image recognition model according to the sample street view picture containing the obstacle and the sample lane line on the removed side to obtain the trained image recognition model.
In one implementation, the training the image recognition model according to the sample lane line and the sample street view picture to obtain the trained image recognition model includes:
setting a sample label for the sample lane line to obtain a label corresponding to the sample lane;
inputting the sample street view picture into the image recognition model to obtain an output result of the image recognition model;
and training the image recognition model according to the output result and the label corresponding to the sample lane to obtain the trained image recognition model.
In one implementation, the fusing the lane information with the map to obtain a simulated road network includes:
obtaining the matching position of the street view image on the map according to the map and the street view image;
and fusing the lane information corresponding to the street view image to a matching position on the map.
In one implementation, the obtaining a matching position of the street view image on the map according to the map and the street view image includes:
obtaining each image block in the street view image according to the street view image;
arranging the image blocks according to a set direction to obtain the arranged image blocks;
obtaining each matching image matched with each image block on the map according to the map and each image block;
and obtaining the matching position of each matching image on the map according to the map and each matching image.
In one implementation, the obtaining each image block in the street view image according to the street view image includes:
obtaining the position of the lane in the street view image according to the street view image;
and obtaining each image block corresponding to the position in the street view image according to the position.
In a second aspect, an embodiment of the present invention further provides an apparatus for generating a lane-level simulated road network, where the apparatus includes the following components:
the information acquisition module is used for acquiring street view images and maps corresponding to the street view images;
the lane recognition module is used for carrying out lane recognition on the street view image by applying an image recognition model to obtain lane information corresponding to the street view image;
and the fusion module is used for fusing the lane information and the map to obtain a simulated road network.
In a third aspect, an embodiment of the present invention further provides a terminal device, where the terminal device includes a memory, a processor, and a program stored in the memory and executable on the processor for generating a lane-level simulated road network, and when the processor executes the program for generating a lane-level simulated road network, the steps of the method for generating a lane-level simulated road network are implemented.
In a fourth aspect, the embodiment of the present invention further provides a computer readable storage medium, where a program for generating a lane-level simulated road network is stored, and when the program for generating a lane-level simulated road network is executed by a processor, the steps of the method for generating a lane-level simulated road network are implemented.
Has the advantages that: the method comprises the steps of firstly obtaining an image corresponding to the street view and a map of the position of the street view, applying an image recognition algorithm to the street view image, recognizing lane information in the street view image, equivalently extracting a lane from the street view image, and then placing the extracted lane on the map to obtain a simulated road network (simulating an actual lane on the map to obtain the road network formed by the actual lane). This makes it possible to obtain information about the individual lanes from the map as a whole. Compared with the manual on-site collection of lane information, the method for acquiring the lane information by the image recognition method reduces the workload on one hand and can improve the accuracy of the acquired simulated road network on the other hand. In addition, compared with the prior art in which lane information is manually collected, the image recognition of the method is high in speed, so that the method can update the simulation road network in time.
Drawings
FIG. 1 is an overall flow chart of the present invention;
FIG. 2 is a flow chart of a portion of an embodiment of the present invention;
FIG. 3 is a multi-lane line of the present invention;
FIG. 4 is a lane-free line of the present invention;
FIG. 5 is a grid lane line of the present invention;
FIG. 6 is a lane line with an obstacle according to the present invention;
fig. 7 is a lane-level simulation road network according to the present invention.
Detailed Description
The technical scheme of the invention is clearly and completely described below by combining the embodiment and the attached drawings of the specification. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Research shows that with the development of cities, the continuous expansion of urban road networks and the rapid increase of traffic flow, the requirements of urban-level traffic simulation on lane-level road networks are rapidly increased. The method has the advantages that the running state of the road network is accurately simulated and evaluated, the quantitative analysis and the feedback regulation are carried out on the road network, and the method is a key and difficult point for realizing the technologies such as network signal control evolution, dynamic traffic distribution and the like. The traffic simulation road network is an excellent tool for finely designing and quantitatively evaluating the traffic condition of vehicles on the lanes. Most of the existing simulation road networks acquire lane information manually on site and then draw the lane information on a map manually to obtain the simulation road network containing the lane information, but the manual drawing of the simulation road network increases the labor cost on one hand, and the condition that data is lost and the collection is missed exists when a driver collects the lane information, so that the accuracy of the drawn simulation road network is reduced, namely the simulation road network deviates from the actual road network, and the use of the simulation road network is further influenced.
In order to solve the technical problems, the invention provides a method, equipment and a storage medium for generating a lane-level simulation road network, and solves the problem of low accuracy of the simulation road network obtained in the prior art. In specific implementation, the method comprises the steps of firstly obtaining an image corresponding to the street view and a map of the position of the street view, applying an image recognition algorithm to the street view image to recognize lane information in the street view image, and then placing the extracted street on the map to obtain the simulated road network. This makes it possible to obtain information about the individual lanes from the map as a whole. Compared with the manual on-site collection of lane information, the method for acquiring the lane information by the image recognition method reduces the workload on one hand and can improve the accuracy of the acquired simulated road network on the other hand.
For example, there are A, B, C lanes in a block, and by capturing a block image of the block at high altitude, and applying an image recognition model to the captured block image, the image recognition model can recognize the layout of A, B, C three lanes in the block image, and then find the positions on the map corresponding to the actual positions of A, B, C three lanes in the block map. The positions are A, B, C on the map, and the simulated road network can be obtained according to A, B, C positions of the three lanes on the map.
Exemplary method
The method for generating the lane-level simulation road network according to the embodiment can be applied to terminal equipment, and the terminal equipment can be a mobile phone, a computer and the like with an image playing function. In this embodiment, as shown in fig. 1, the method for generating a lane-level simulated road network specifically includes the following steps:
s100, obtaining the street view image and a map corresponding to the street view image.
The street view image in this embodiment is obtained through an open source street view map platform, and the map corresponding to the street view is also an open source map.
S200, applying an image recognition model to the street view image to perform lane recognition, and obtaining lane information corresponding to the street view image.
Step S200 includes two parts: and training the image recognition model and applying the image recognition model after training to recognize the lane information in the actual street view image. These are described below:
the training of the image recognition model comprises the following steps:
s201, acquiring a sample lane and a sample street view picture.
The embodiment collects a large amount of sample databases, the sample databases include sample street view pictures, and each sample street view picture may or may not contain a lane. The sample database also covers sample street view pictures without lanes, so that a negative sample data set is manufactured, and the negative sample data set is mainly used for manufacturing a negative sample set aiming at the condition that no lane line exists in areas such as traffic intersections, turning places and the like, so that the model has the condition of fitting no lane line.
Street views in the real world are collected through an open source street view map platform and used for making a sample database, and attributes corresponding to the street views in the real world are recorded, wherein the attributes comprise shooting time, shooting places (longitude and latitude information) and the like.
S202, constructing an image recognition model.
The image recognition model of the embodiment includes an Attention extension model, a Distributed Local Attention (DLA) model and a neural network model, and the three models are independent of each other.
Attention Expansion model: using a linear transformation matrix WA∈RH'xH(H'>H) Let the attention mapsA [ A ]1；...；AH]Mapping to
And satisfies the following conditions:
the Embedding dimension of the original model is 192, which has only 3 heads, and the dimension of each head is 64. Now the Embedding dimension of the model has been extended to 384, and 6 heads, then the dimension of each head is still 64. This enjoys the benefits of both more heads and high embedding dimensions. The attribute map is used for obtaining fine-grained information of a more local lane line, namely the previous relation of each pixel point in each lane line, and more accurate coordinate point positioning is obtained through the attention map mechanism.
DLA model:
the attribute map and value of DLA model are subjected to feature aggregation operation. In contrast, directly superimposing multiple convolutions may result in a large center of the receptive field, resulting in the model ignoring features at the image boundaries. DLA (distributed Local attachment) can effectively overcome the limitations of DLA (distributed Local attachment) and better jointly model Local features and global features. The Local of the DLA focuses more on the global receptive field, namely the information of the overall contour of each lane line, so that the distances between the lane lines and the background are clearer, the regional position of each lane line is focused more, and the position of a model resolution point is more accurate
S203, training the image recognition model according to the sample lane and the sample street view picture to obtain the trained image recognition model.
The embodiment trains the model by using the lane line corresponding to the lane.
The sample lane lines of the present embodiment include sample grid lane lines and sample multi-lane lines located on both sides of the obstacle. The sample grid lane lines further comprise sample bus grid lane lines and sample no-parking grid lane lines. The sample bus grid lane lines and the sample no-parking grid lane lines are yellow grid lines in bus areas, no-parking areas and the like; the sample lane lines on the two sides of the obstacle comprise sample lane lines on the two sides of the obstacle such as a fence, a stone pier and the like, and the lane lines on the other side of the obstacle can be removed when the application model is used for identifying that the shielded lane lines on the other side of the obstacle are abnormal. In the embodiment, when the database is collected, if the lane lines at the edge part of the road surface are worn and disappear or the condition that the lane lines cannot be seen due to vehicle occlusion is caused, the occluded lane lines are automatically supplemented so as to increase samples in the database. By constructing the sample database, the sample lane line of the embodiment reaches more than 10 lanes, so that the requirement of the training model on the number of samples is met, and the trained model can meet the requirement of recognition of various lane scenes.
After obtaining the sample database, training the image recognition model is started, and step S203 includes steps S2031, S2032, and S2033 as follows:
s2031, setting a sample label for the sample lane line to obtain a label corresponding to the sample lane.
In the embodiment, the labels are actually marked on the multiple lane lines, the model training aims to fit the shape of each lane line in a regression mode, the model has the capability of finely regressing each coordinate point from coarse to fine, and the iteration effect is better and better along with the increase of the training times of the model until the result output by the model is infinitely close to the actually marked lane line labels.
S2032, inputting the sample street view picture into the image recognition model to obtain an output result of the image recognition model.
Because the sample street view picture containing the lane line and the sample street view picture (negative sample) containing no lane line are set in the embodiment, the two types of sample street view pictures are input into the image recognition model, and the model is trained. This enables the model of the present embodiment to output no result for a picture without lane lines.
S2033, training the image recognition model according to the output result and the label corresponding to the sample lane to obtain the trained image recognition model.
For example, the label corresponding to the linear sample lane is a straight line, when the sample street view picture where the linear sample lane is located is input into the value model, if the output of the model after the model identifies the sample street view picture is not matched with the straight line, the parameters of the model are adjusted until the output result is matched with the label, and the training of the model is completed.
And S300, fusing the lane information and the map to obtain a simulated road network.
In the embodiment, the lane information (shape and position) identified by the model is fused to the map where the actual lane is located, so that the lane information of each place can be seen on the map. In the embodiment, a Map Matching algorithm (Map-Matching) is adopted to fuse the lane information with the Map. Step S300 includes steps S301 and S302 as follows:
s301, obtaining the matching position of the street view image on the map according to the map and the street view image.
In the embodiment, a plurality of image blocks are collected in the same street, and the image blocks form a street view image of the street. Then, the image blocks are arranged according to the same azimuth angle, for example, the image blocks are arranged along the direction of an actual lane, an image matched with the image blocks is found in the image on the map, the position of the matched image on the map is the position of the street view image on the map, that is, the position of the lane in the street view image on the map, and the specific process includes: obtaining each image block in the street view image according to the street view image; arranging the image blocks according to a set direction (the trend of a lane) to obtain the arranged image blocks; obtaining each matching image matched with each image block on the map according to the map and each image block; and obtaining the matching position of each matching image on the map according to the map and each matching image.
S302, the lane information corresponding to the street view image is fused to a matching position on the map.
And (4) placing the lane information identified by the model on the matching position to obtain the simulated road network on the map.
By way of example, the overall process of the invention is illustrated by the following example:
as shown in fig. 2, a basic vector road network is obtained based on OSM, road network topology at road level is generated by using the basic vector road network, and then city-level vehicle-mounted view street scenes are obtained by adopting a spatial traversal algorithm. The above process is equivalent to the collection of sample street view pictures in this embodiment, a typical scene (the picture contains multiple lane lines, grid lane lines, and no lane line) is selected from all the obtained sample street view pictures to construct a basic data set, then lane line data is labeled (a sample label is set for the sample lane line), then the sample street view picture with the label is input into a Transfomer lane line recognition algorithm, the algorithm is trained, the street view image to be processed is input into the trained algorithm to recognize the lane line information in the street view image to be processed, the recognized lane line information is fused into a map based on an HMM map matching algorithm, and a high-precision lane-level simulation road network is obtained. As shown in fig. 3, the multi-lane line identified by the present invention is shown. Fig. 4 shows a street view without lane lines according to the present invention. As shown in fig. 5, is an inventive grid lane line. As shown in fig. 6, the lane line of the invention is hidden by a fence (obstacle). Fig. 7 is a lane-level simulation road network obtained by the present invention.
In summary, the invention first obtains an image corresponding to a street view and a map of a location of the street view, applies an image recognition algorithm to the street view image to recognize lane information in the street view image, which is equivalent to extracting a lane from the street view image, and then puts the extracted lane on the map to obtain a simulated road network (simulating an actual lane on the map to obtain a road network formed by the actual lane). This makes it possible to obtain information about the individual lanes from the map as a whole. Compared with the manual on-site collection of lane information, the method for acquiring the lane information by the image recognition method reduces the workload on one hand and can improve the accuracy of the acquired simulated road network on the other hand. In addition, compared with the prior art in which lane information is manually collected, the image recognition of the method is high in speed, so that the method can update the simulation road network in time.
Exemplary devices
The embodiment also provides a device for generating a lane-level simulation road network method, which comprises the following components:
the information acquisition module is used for acquiring street view images and maps corresponding to the street view images;
the lane recognition module is used for carrying out lane recognition on the street view image by applying an image recognition model to obtain lane information corresponding to the street view image;
and the fusion module is used for fusing the lane information and the map to obtain a simulated road network. Based on the above embodiment, the present invention further provides a terminal device, where the terminal device includes a memory, a processor, and a program stored in the memory and operable on the processor for generating a lane-level simulated road network, and the processor implements the steps of the method for generating a lane-level simulated road network when executing the program for generating a lane-level simulated road network.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, databases, or other media used in embodiments provided herein may include non-volatile and/or volatile memory. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
In summary, the present invention discloses a method, a device and a storage medium for generating a lane-level simulation road network, wherein the method comprises: obtaining a street view image and a map corresponding to the street view image; applying an image recognition model to the street view image to perform lane recognition to obtain lane information corresponding to the street view image; and fusing the lane information and the map to obtain a simulated road network. Compared with the manual on-site collection of lane information, the method for acquiring the lane information by the image recognition method reduces the workload on one hand and can improve the accuracy of the acquired simulated road network on the other hand. In addition, compared with the prior art in which lane information is manually collected, the image recognition of the method is high in speed, so that the method can update the simulation road network in time.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (13)
1. A method of generating a lane-level simulated road network, comprising:
obtaining a street view image and a map corresponding to the street view image;
applying an image recognition model to the street view image to perform lane recognition to obtain lane information corresponding to the street view image;
and fusing the lane information and the map to obtain a simulated road network.
2. The method of claim 1, wherein the applying an image recognition model to the street view image for lane recognition to obtain lane information corresponding to the street view image comprises:
acquiring a sample lane and a sample street view picture;
constructing an image recognition model;
training the image recognition model according to the sample lane and the sample street view picture to obtain the trained image recognition model;
and performing lane recognition on the image recognition model after the street view image is trained to obtain lane information in the street view image.
3. The method of claim 2, wherein said training said image recognition model according to said sample lane and said sample street view picture to obtain said trained image recognition model comprises:
obtaining a sample lane line corresponding to the sample lane according to the sample lane;
and training the image recognition model according to the sample lane line and the sample street view picture to obtain the trained image recognition model.
4. The method of generating a lane-level simulated road network according to claim 3, wherein said training said image recognition model according to said sample lane line and said sample street view picture to obtain said image recognition model after training comprises:
obtaining a sample grid lane line in the sample lane lines according to the sample lane lines;
obtaining the sample street view picture containing the grid lane lines in the sample street view picture according to the sample street view picture;
and training the image recognition model according to the sample grid lane line and the sample street view picture containing the grid lane line to obtain the trained image recognition model.
5. The method of generating a lane-level simulated road network according to claim 4, wherein said training said image recognition model according to said sample grid lane line and said sample street view picture containing grid lane line to obtain said image recognition model after training comprises:
obtaining a sample bus lane in the sample lanes according to the sample lanes;
obtaining the sample bus grid lane lines corresponding to the sample bus lanes in the sample grid lane lines according to the sample grid lane lines;
obtaining a sample bus street view picture containing the sample bus grid lane line in the sample street picture according to the sample street picture;
and training the image recognition model according to the sample bus grid lane line and the sample bus street view picture to obtain the trained image recognition model.
6. The method of generating a lane-level simulated road network according to claim 3, wherein said training said image recognition model according to said sample lane line and said sample street view picture to obtain said image recognition model after training comprises:
obtaining sample lane lines positioned at two sides of the barrier in the sample lane lines according to the sample lane lines;
removing a sample lane line on one side of the obstacle;
obtaining the sample street view picture containing the barrier in the sample street view picture according to the sample street view picture;
and training the image recognition model according to the sample street view picture containing the obstacle and the sample lane line on the removed side to obtain the trained image recognition model.
7. The method of generating a lane-level simulated road network according to claim 3, wherein said training said image recognition model according to said sample lane line and said sample street view picture to obtain said image recognition model after training comprises:
setting a sample label for the sample lane line to obtain a label corresponding to the sample lane;
inputting the sample street view picture into the image recognition model to obtain an output result of the image recognition model;
and training the image recognition model according to the output result and the label corresponding to the sample lane to obtain the trained image recognition model.
8. The method of generating a lane-level simulation road network according to claim 1, wherein said fusing the lane information with the map to obtain a simulation road network comprises:
obtaining the matching position of the street view image on the map according to the map and the street view image;
and fusing the lane information corresponding to the street view image to a matching position on the map.
9. The method for generating a lane-level simulation road network according to claim 8, wherein said obtaining the matching position of the street view image on the map according to the map and the street view image comprises:
obtaining each image block in the street view image according to the street view image;
arranging the image blocks according to a set direction to obtain the arranged image blocks;
obtaining each matching image matched with each image block on the map according to the map and each image block;
and obtaining the matching position of each matching image on the map according to the map and each matching image.
10. The method of generating a lane-level simulation road network according to claim 9, wherein said obtaining each image block in said street-view image according to said street-view image comprises:
obtaining the position of the lane in the street view image according to the street view image;
and obtaining each image block corresponding to the position in the street view image according to the position.
11. A device for generating a lane-level simulation road network method is characterized by comprising the following components:
the information acquisition module is used for acquiring street view images and maps corresponding to the street view images;
the lane recognition module is used for carrying out lane recognition on the street view image by applying an image recognition model to obtain lane information corresponding to the street view image;
and the fusion module is used for fusing the lane information and the map to obtain a simulated road network.
12. A terminal device comprising a memory, a processor and a program for generating a road-level simulated road network stored in said memory and operable on said processor, said processor implementing the steps of the method for generating a road-level simulated road network according to any one of claims 1 to 10 when executing said program for generating a road-level simulated road network.
13. A computer-readable storage medium, having stored thereon a program for generating a lane-level simulated road network, which when executed by a processor, carries out the steps of the method of generating a lane-level simulated road network according to any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111056068.8A CN113887321A (en) | 2021-09-09 | 2021-09-09 | Method, equipment and storage medium for generating lane-level simulation road network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111056068.8A CN113887321A (en) | 2021-09-09 | 2021-09-09 | Method, equipment and storage medium for generating lane-level simulation road network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113887321A true CN113887321A (en) | 2022-01-04 |
Family
ID=79008848
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111056068.8A Pending CN113887321A (en) | 2021-09-09 | 2021-09-09 | Method, equipment and storage medium for generating lane-level simulation road network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113887321A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114170301A (en) * | 2022-02-09 | 2022-03-11 | 城云科技（中国）有限公司 | Abnormal municipal facility positioning method and device and application thereof |
| CN115438517A (en) * | 2022-11-07 | 2022-12-06 | 阿里巴巴达摩院(杭州)科技有限公司 | Simulation map generation method, electronic device and computer storage medium |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103954275A (en) * | 2014-04-01 | 2014-07-30 | 西安交通大学 | Lane line detection and GIS map information development-based vision navigation method |
| CN107767696A (en) * | 2016-08-19 | 2018-03-06 | 德韧营运有限责任公司 | Simultaneously generate multilane map and on the map generated positioning vehicle system and method |
| CN108388641A (en) * | 2018-02-27 | 2018-08-10 | 广东方纬科技有限公司 | A kind of means of transportation based on deep learning ground drawing generating method and system |
| JP2019204211A (en) * | 2018-05-22 | 2019-11-28 | トヨタ自動車株式会社 | Map generation system |
| CN110941684A (en) * | 2018-09-21 | 2020-03-31 | 高德软件有限公司 | Production method of map data, related device and system |
| CN111597995A (en) * | 2020-05-15 | 2020-08-28 | 弗徕威智能机器人科技（上海）有限公司 | Lane marking line detection method, system and storage medium |
| CN111815742A (en) * | 2020-09-14 | 2020-10-23 | 蘑菇车联信息科技有限公司 | Lane line generation method and system |
-
2021
- 2021-09-09 CN CN202111056068.8A patent/CN113887321A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103954275A (en) * | 2014-04-01 | 2014-07-30 | 西安交通大学 | Lane line detection and GIS map information development-based vision navigation method |
| CN107767696A (en) * | 2016-08-19 | 2018-03-06 | 德韧营运有限责任公司 | Simultaneously generate multilane map and on the map generated positioning vehicle system and method |
| CN108388641A (en) * | 2018-02-27 | 2018-08-10 | 广东方纬科技有限公司 | A kind of means of transportation based on deep learning ground drawing generating method and system |
| JP2019204211A (en) * | 2018-05-22 | 2019-11-28 | トヨタ自動車株式会社 | Map generation system |
| CN110941684A (en) * | 2018-09-21 | 2020-03-31 | 高德软件有限公司 | Production method of map data, related device and system |
| CN111597995A (en) * | 2020-05-15 | 2020-08-28 | 弗徕威智能机器人科技（上海）有限公司 | Lane marking line detection method, system and storage medium |
| CN111815742A (en) * | 2020-09-14 | 2020-10-23 | 蘑菇车联信息科技有限公司 | Lane line generation method and system |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114170301A (en) * | 2022-02-09 | 2022-03-11 | 城云科技（中国）有限公司 | Abnormal municipal facility positioning method and device and application thereof |
| CN115438517A (en) * | 2022-11-07 | 2022-12-06 | 阿里巴巴达摩院(杭州)科技有限公司 | Simulation map generation method, electronic device and computer storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112212874B (en) | Vehicle track prediction method and device, electronic equipment and computer readable medium | |
| WO2018145602A1 (en) | Lane determination method, device and storage medium | |
| Yang et al. | Generating lane-based intersection maps from crowdsourcing big trace data | |
| CN113887321A (en) | Method, equipment and storage medium for generating lane-level simulation road network | |
| CN112307566A (en) | Vehicle simulation test method, device, equipment and storage medium | |
| CN106980657A (en) | A kind of track level electronic map construction method based on information fusion | |
| CN105160309A (en) | Three-lane detection method based on image morphological segmentation and region growing | |
| CN110689724B (en) | Automatic motor vehicle zebra crossing present pedestrian auditing method based on deep learning | |
| CN106980855A (en) | Traffic sign quickly recognizes alignment system and method | |
| CN108319952A (en) | A kind of vehicle characteristics extracting method and device | |
| CN112633722B (en) | Vehicle-mounted road safety risk assessment system and method | |
| WO2021237768A1 (en) | Data-driven-based system for implementing automatic iteration of prediction model | |
| CN109101743A (en) | A kind of construction method of high-precision road net model | |
| CN112070049A (en) | Semantic segmentation method under automatic driving scene based on BiSeNet | |
| JP5544595B2 (en) | Map image processing apparatus, map image processing method, and computer program | |
| CN112446915B (en) | Picture construction method and device based on image group | |
| CN112837414B (en) | Method for constructing three-dimensional high-precision map based on vehicle-mounted point cloud data | |
| CN113312987B (en) | Recognition method based on unmanned aerial vehicle road surface crack image | |
| CN114758086A (en) | Method and device for constructing urban road information model | |
| CN112833891A (en) | Road data and lane-level map data fusion method based on satellite film recognition | |
| CN117011413A (en) | Road image reconstruction method, device, computer equipment and storage medium | |
| Taal et al. | Procedural Generation of Traffic Signs. | |
| CN114485684B (en) | Map road matching method and system based on ground element topological relation | |
| CN112507977B (en) | Lane line positioning method and device and electronic equipment | |
| CN111325811A (en) | Processing method and processing device for lane line data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |