CN110501018A - A kind of traffic mark board information collecting method for serving high-precision map producing - Google Patents
A kind of traffic mark board information collecting method for serving high-precision map producing Download PDFInfo
- Publication number
- CN110501018A CN110501018A CN201910745743.4A CN201910745743A CN110501018A CN 110501018 A CN110501018 A CN 110501018A CN 201910745743 A CN201910745743 A CN 201910745743A CN 110501018 A CN110501018 A CN 110501018A
- Authority
- CN
- China
- Prior art keywords
- information
- traffic
- mark board
- traffic mark
- location information
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
- G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
- G01C21/26—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network
- G01C21/28—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00 specially adapted for navigation in a road network with correlation of data from several navigational instruments
- G01C21/30—Map- or contour-matching
- G01C21/32—Structuring or formatting of map data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/582—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of traffic signs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/62—Text, e.g. of license plates, overlay texts or captions on TV images
- G06V20/63—Scene text, e.g. street names
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/14—Image acquisition
- G06V30/146—Aligning or centring of the image pick-up or image-field
- G06V30/1475—Inclination or skew detection or correction of characters or of image to be recognised
- G06V30/1478—Inclination or skew detection or correction of characters or of image to be recognised of characters or characters lines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
Abstract
The present invention relates to high-precision map field of information acquisition, disclosing a kind of traffic mark board information collecting method for serving high-precision map producing and system includes obtaining traffic scene image information to be identified;Obtain the location information of traffic mark board in the picture;Obtain the location information of traffic sign in the picture;Ownership judgement is carried out using the location information of location information and traffic mark board in the picture of traffic sign in the picture, road signs information is belonged in corresponding traffic mark board information;Obtain location information of the traffic mark board in high-precision map, road signs information and corresponding traffic mark board information are added in high-precision cartographic information, convenient for the storage of data, and the positioning of traffic mark board is more accurate, convenient for using high-precision map information navigation etc. in application, the correct guidance provided.
Description
Technical field
The present invention relates to high-precision map field of information acquisition, in particular to a kind of friendship for serving high-precision map producing
Logical sign board information collecting method.
Background technique
High-precision map includes a large amount of auxiliary driving information, and wherein road signs information is particularly important, traffic sign letter
An important module of the acquisition of breath as high-precision map producing, existing technological means are mostly that road traffic sign detection combines
GPS position measurement.
At present in traffic mark board context of detection, the main method using vision-based detection obtains the position letter of traffic mark board
Breath, such as a kind of road traffic sign detection and recognition methods based on residual error SSD model of Patent No. CN108960198A, by right
Image carries out multiple dimensioned piecemeal;It uses residual error network ResNet101 as the basic network of SSD, constructs residual error SSD model, carry out
Network training completes the detection and identification with generalization ability, realizes to multiclass difference size mark in the true traffic scene of China
The effective detection and identification of will board.Such as a kind of view-based access control model attention mechanism of Patent No. CN108256467A and geometrical characteristic
Method for traffic sign detection, the characteristics of according to traffic sign, by the organic knot of the geometrical characteristic of vision noticing mechanism and traffic sign
It closes, Conventional visual attention mechanism is improved, introduce the geometrical characteristic constraint of traffic sign, exclusive PCR is realized and handed over
The ranging of logical sign board.
In currently existing scheme, the detection and identification of traffic mark board in the picture are pertained only to, or to traffic mark board
Ranging localization research, and the demand of more detailed road signs information is needed in high-precision cartographic information, such as turn
The information for being used to that user be guided to travel to information, speed-limiting messages etc..
Summary of the invention
For background technique problem encountered, simply there is detailed traffic mark the purpose of the present invention is to provide a kind of
The traffic mark board information collecting method for serving high-precision map producing and system of will information.
In order to achieve the above object, the present invention adopts the following technical scheme: a kind of friendship for serving high-precision map producing
Logical sign board information collecting method includes: to obtain traffic scene image information to be identified;Obtain traffic mark board in the picture
Location information;Obtain the location information of traffic sign in the picture;Using the location information of traffic sign in the picture and
The location information of traffic mark board in the picture carries out ownership judgement, and road signs information is associated with corresponding traffic mark board
In information;Location information of the traffic mark board in high-precision map is obtained, by road signs information and corresponding traffic sign
Board information is added in high-precision cartographic information.
Preferably, using the traffic sign in the deep learning object detection method of anchor-free simultaneously detection image
Board and traffic sign obtain the location information and traffic mark of the parallel minimum circumscribed rectangle frame of traffic mark board in the picture respectively
The location information of the parallel minimum circumscribed rectangle of will in the picture.
Preferably, the location information of the parallel minimum circumscribed rectangle frame using traffic mark board in the picture obtains positioning
Terminal is at a distance from traffic mark board;Positioning is obtained in conjunction with real-time dynamic carrier phase difference location technology and Inertial Measurement Unit
The location information of terminal;In conjunction with the location information of positioning terminal and the range information of the positioning terminal and traffic mark board, obtain
Take location information of the traffic mark board in high-precision map.
Preferably, four corner location information for obtaining traffic mark board calculate positioning eventually by the method for binocular ranging
The actual range of pixel in the rectangle that end is formed with four angle point lines, and using the average distance of actual range as the positioning
Terminal is at a distance from traffic mark board.
Preferably, parallel minimum circumscribed rectangle frame is expanded as into rectangle around its central point outward and increases frame, is increased in rectangle
By critical point detection in big frame, four corner location information of traffic mark board are obtained.
Preferably, pass through four corner location information correcting images of traffic mark board;Image after correction is subjected to OCR
Identification, detects supplementary textual information;Supplementary textual information is associated in corresponding traffic mark board information, and will auxiliary text
Word information is added in high-precision cartographic information.
Preferably, the location information of the parallel minimum circumscribed rectangle frame of traffic mark board in the picture includes traffic mark board
Centre coordinate data, width data and altitude information;Believe the position of the parallel minimum circumscribed rectangle of traffic sign in the picture
Breath includes the centre coordinate data of traffic sign;Compare the centre coordinate data of traffic sign and the centre coordinate of traffic mark board
Data, width data and altitude information belong to road signs information in corresponding traffic mark board information.
Preferably, the classification information of the major class of traffic sign is obtained, the major class includes prompt mark, caution sign and taboo
Only indicate, then using the classification information of major class as the target detection of same class, and the result that will test is as traffic sign point
The input of class device, obtains the classification information of the group of traffic sign, and the group includes speed limit, limit for height, and taboo is stopped.
Preferably, a kind of computer readable storage medium, is stored thereon with computer program, and the computer program is located
Manage the step of realizing any of the above-described the method when device executes.
Preferably, a kind of traffic mark board information acquisition system for serving high-precision map producing includes: that image obtains
Module, for obtaining traffic scene image information to be identified;Traffic mark board obtains module, is scheming for obtaining traffic mark board
Location information as in;Traffic sign obtains module, for obtaining the location information of traffic sign in the picture;Belong to mould
Block, for being belonged to using the location information of location information and traffic mark board in the picture of traffic sign in the picture
Judgement, road signs information is associated in corresponding traffic mark board information;Locating module exists for obtaining traffic mark board
Road signs information and corresponding traffic mark board information are added to high-precision cartographic information by the location information in high-precision map
In.
Compared with prior art, the present invention provides a kind of traffic mark board information for serving high-precision map producing to adopt
Set method characterized by comprising obtain traffic scene image information to be identified;Obtain the position of traffic mark board in the picture
Confidence breath;Obtain the location information of traffic sign in the picture;Utilize traffic sign location information in the picture and friendship
The location information of logical sign board in the picture carries out ownership judgement, and road signs information is associated with corresponding traffic mark board and is believed
In breath;Location information of the traffic mark board in high-precision map is obtained, by road signs information and corresponding traffic mark board
Information is added in high-precision cartographic information.After road signs information is associated in corresponding traffic mark board information, then it will hand over
Logical flag information and corresponding traffic mark board information are added in high-precision cartographic information, convenient for the storage of data, Er Qiejiao
The positioning of logical sign board is more accurate, convenient for using high-precision map information navigation etc. in application, the correct guidance provided so that
The traffic sign for belonging to same traffic mark board appears in same position on high-precision map, appears in reality with actual traffic mark
Position consistency on the road of border convenient for providing correct prompt, and also improves the aesthetics of high-precision map.
Detailed description of the invention
Fig. 1 is a kind of traffic mark board information collection side for serving high-precision map producing provided in an embodiment of the present invention
The flow diagram of method;
Fig. 2 is a kind of composition figure for the traffic mark board information acquisition system for serving high-precision map producing of the present invention;
Through the above attached drawings, it has been shown that the specific embodiment of the disclosure will be hereinafter described in more detail.These attached drawings
It is not intended to limit the scope of this disclosure concept by any means with verbal description, but is by referring to specific embodiments
Those skilled in the art illustrate the concept of the disclosure.
Specific embodiment
Hereafter exemplary embodiment is illustrated by detailed, embodiment generation described in provided embodiment
Table part better embodiment of the invention, and simultaneously not all embodiments.Based on the embodiments of the present invention and picture and text, this
Field technical staff can be obtained every other embodiment without creative labor, all will be in the present invention
Within the scope of protection.
The application scenarios of the embodiment of the present invention are described as follows, and each intelligent navigation terminal realizes that high-precision is navigated and driven automatically
It is increasing to sail the demand to the high-precision map of electronics such as automobile, high-precision map includes a large amount of auxiliary driving information, such as
Current lane information, traffic mark board information and traffic lights information etc., wherein traffic mark board information is particularly important, therefore, is
The high-precision cartographic information with all kinds of detailed road signs informations is obtained, the embodiment of the invention provides one kind to serve
The traffic mark board information collecting method of high-precision map producing, this method can be applied in each acquisition terminal system, should
Acquisition terminal system includes CCD camera system and data processing system, and data processing system is primarily to storage and processing are clapped
Related roads data, road signs information, the traffic mark board information taken the photograph back.It includes the number that this method, which also can be applied to other,
According in the processing terminal system of processing module, the embodiment of the present invention is not particularly limited this.The acquisition terminal system can be with
Relevant operation is carried out by carrier of automobile.
CCD camera is binocular in the present embodiment, and CCD camera can be three mesh or more in other embodiments, again not
Limitation.The CCD camera can shoot the region comprising traffic mark board, be believed with obtaining the image comprising traffic mark board
Breath, the data processing system can store and process the image information.
The acquisition terminal system not only may include the CCD camera system and data processing system of binocular, can also include
Other hardware and system.For example processor, the embodiment of the present invention are not specifically limited in this embodiment.The processor can be with the CCD phase
Machine connection, the processor can carry out relevant processing to the image information comprising traffic mark board.
The acquisition terminal system not only may include CCD camera system and data processing system, can also be including other
System, such as data transmission system, what which can will acquire includes road signs information and traffic mark board
The high-precision cartographic information of information is transferred to cloud or server.To realize high-precision navigation or the automobile etc. of intelligent terminal
Automatic Pilot, the embodiment of the present invention are also not specifically limited this.
It should also be noted that, acquisition terminal system can include by the acquisition of above-mentioned CCD camera in the embodiment of the present invention
The image of traffic sign and traffic mark board, in other embodiments, acquisition terminal can also be obtained by other means comprising handing over
The image of logical mark and traffic mark board.
In the embodiment of the present invention, acquisition terminal system can use CCD camera binocular distance measurement method ranging, can also be with
With three mesh telemetrys etc., to improve the accuracy of ranging.
If Fig. 1 is a kind of traffic mark board information collection for serving high-precision map producing provided in an embodiment of the present invention
The flow diagram of method.This method comprises:
S1 obtains traffic scene image information to be identified.
Traffic mark board be with graphical symbol and text transmitting specific information, to manage traffic, instruction direction of traffic with
Guarantee the facility of the coast is clear and traffic safety.Suitable for highways and streets and all accommodation roads, the property with decree
Matter, vehicle, pedestrian must comply with.Since the surface color of traffic mark board is all more gorgeous, effectively increases it and distinguish energy
Power takes the clear image comprising traffic mark board convenient for CCD camera.
Specifically, all directions are equipped with CCD camera on acquisition terminal, CCD camera claps the scene of all directions
It takes the photograph, obtains traffic scene image information, some in the traffic scene image include traffic mark board image and traffic indication map
Picture, traffic scene image some do not include traffic mark board image and Traffic Sign Images, it should be noted that acquisition terminal removes
Outside the various attribute informations for recording road, the interest such as food and drink, hotel, market, gas station, the parking lot on road periphery can be also obtained
Point information etc..The monitoring range for obtaining traffic scene image is larger, can comprehensively collect road information, and equipment cost compared with
It is low.
S2 obtains the location information of traffic mark board in the picture.
Specifically, process mainly includes to described image using the deep learning object detection method of anchor-free
Image coding and decoding two parts, wherein image coding using VGG16 be used as diaphyseal portion, repeatedly stacking 3x3 small rolls
The maximum pond layer of product core and 2x2 is conducive to obtain semantic information in perception open country.And image decoding will using the method for up-sampling
Coded portion amplifies, and is embodied as being carried out continuously 2 times of deconvolution operation, and on the basis of decoding feature using 1x1's
Convolution kernel carries out regression forecasting, directly obtains the location information of traffic mark board in the picture, and obtained result is each traffic
Sign board minimum circumscribed rectangle in the picture.
Specifically, step 1: network inputs size is fixed, obtain detection image after by image scaling be fixed dimension
640x352x3;
Step 2: inputting an image into VGG16 network, and forward inference calculates, and carries out the operation of a large amount of convolution sum pond;
Step 3: after the completion of forward inference calculates, the characteristics of image encoded is the 1/32 of original image, having a size of
20x11x512;
Step 4: 8 times of up-sampling decoding operates, specially continuous 2x2 de-convolution operation are carried out to coding characteristic;
Step 5: after up-sampling, decoding characteristics of image is obtained, having a size of 160x88x128;
Step 6: regression forecasting is carried out to decoding feature using the convolution kernel of 1x1, obtains class prediction and box prediction letter
Breath;
Step 7: in conjunction with class prediction and box predictive information, the box information for being wherein predicted as traffic mark board is extracted.
S3 obtains the location information of traffic sign in the picture.
Specifically, traffic sign, is the road equipment for transmitting guidance, limitation, warning or instruction information with text or symbol.
Also known as road sign, road signs.It is usually with safety, the traffic mark that setting is eye-catching, clear, bright in traffic sign
Will implements traffic administration, guarantees traffic safety, smoothly important measures.Traffic sign mainly includes prompt mark, warning
The major class classification such as mark and prohibitory sign.Wherein, prompt mark plays indicative function, shares 29 groups, and color is blue bottom, Bai Tu
Case, such as straight trip, turning, roundabout ahead etc. to the left.Caution sign is mainly telltale, shares 49 groups, and color is yellow bottom,
Black surround, black pattern, such as right-angled intersection, continuous turning, upper abrupt slope etc..Prohibitory sign plays the role of forbidding certain behavior, altogether
There are 43 groups, most of color is white background, red circle, Hong Gang, black pattern, and setting is needing to forbid or now near vehicle etc..Example
Such as, no through traffic, No entry.
Specifically, process mainly includes to described image using the deep learning object detection method of anchor-free
Image coding and decoding two parts it is consistent with the acquisition traffic mark board method of location information in the picture, obtain
The location information of traffic sign in the picture is taken, obtained result is each traffic sign minimum circumscribed rectangle in the picture.
S4 is carried out using the location information of the location information and traffic mark board of traffic sign in the picture in the picture
Ownership judgement, road signs information is associated in corresponding traffic mark board information.
Compare the centre coordinate data of traffic sign and the centre coordinate data of traffic mark board, width data and high degree
According to road signs information is belonged in corresponding traffic mark board information.
It include multiple traffic signs on one traffic mark board, specifically, being sat by the center of judgement detection traffic sign
Whether mark is within the scope of the boundary rectangle of traffic mark board, if so, judge that the traffic sign is attributed on the traffic mark board,
And corresponding road signs information is associated in corresponding traffic mark board information, process mainly comprises the steps that
Step 1: obtaining the location information of traffic mark board in the picture using anchor-free object detection method, false
If the centre coordinate for obtaining traffic mark board is A [x1, y1], wide high respectively W, H;
Step 2: the location information of traffic sign in the picture is obtained using anchor-free object detection method, it is assumed that
It obtains to prompt traffic sign, and prompting the centre coordinate of traffic sign is B [x2, y2], wide high respectively W2, H2;
Step 3: will be prompted to the image within the scope of traffic sign and be input to sorter network, such as alexnet, carry out further
Classification, such as obtain is classified as turn right mark;
Step 4: by judging x2 whether in [x1-W/2, x1+W/2] range, whether y2 is at [y1-H/2, xy1+H/2]
In range, if so, the traffic mark board that right-hand rotation traffic sign ownership centre coordinate is A is judged, if it is not, then carrying out one
The judgement of traffic mark board is finished until being matched to corresponding traffic mark board or traversal;
Step 5: if judging, traffic sign belongs to corresponding traffic mark board, which is associated with phase
In the traffic mark board information answered.
S5 obtains location information of the traffic mark board in high-precision map, by road signs information and corresponding traffic
Sign board information is added in high-precision cartographic information.
Specifically, by the method for binocular ranging obtain pixel of the acquisition terminal near traffic mark board central point it
Between distance, by RTK technology obtain acquisition terminal location information, the two be added after obtain traffic mark board accurately
Location information in figure.Finally, high-precision cartographic information will be added to for road signs information and corresponding traffic mark board information
In, the high-precision cartographic information with detailed traffic flag information is obtained, it, will convenient for being applied or automatic Pilot application when navigation
After road signs information is associated in corresponding traffic mark board information, then by road signs information and corresponding traffic mark board
Information is added in high-precision cartographic information, and convenient for the storage of data, and the positioning of traffic mark board is more accurate, convenient for utilizing
High-precision map information navigation etc. is in application, the correct guidance provided, so that belonging to the traffic sign of same traffic mark board
Same position on high-precision map is appeared in, the position consistency on real road is appeared in actual traffic mark, convenient for providing
Correctly prompt, and also improve the aesthetics of high-precision map.
The location information of the parallel minimum circumscribed rectangle frame of traffic mark board in the picture includes the center of traffic mark board
Coordinate data, width data and altitude information;The location information of the parallel minimum circumscribed rectangle of traffic sign in the picture includes
The centre coordinate data of traffic sign;
Specifically, step 1: obtaining the location information of traffic mark board using anchor-free object detection method, such as
The centre coordinate for obtaining traffic mark board is A [x1, y1], wide high respectively W, H;
Step 2: due to the minimum circumscribed rectangle that obtained detection box is traffic mark board, in the inside meeting of boundary rectangle
Certain ranging will be caused if the part to be considered as to a part of traffic mark board comprising the other background informations in part
Error.So the embodiment of the present invention, by the range shorter of required ranging, the center of ranging range is the center of traffic mark board
Coordinate is A [x1, y1], but wide height is respectively W/10, H/10, carries out certain diminution;
Step 3: measuring the pixel in range using the method for binocular ranging, and takes average as Current traffic
The relative position of sign board;
Step 4: location information of the traffic mark board in high-precision map is obtained in conjunction with RTK information.
If Fig. 2 is a kind of traffic mark board information collection for serving high-precision map producing provided in an embodiment of the present invention
The composition figure of system.The system includes:
S10, image collection module, for obtaining traffic scene image information to be identified;
S20, traffic mark board obtains module, for obtaining the location information of traffic mark board in the picture;
S30, traffic sign obtains module, for obtaining the location information of traffic sign in the picture;
S40 belongs to module, for using traffic sign location information in the picture and traffic mark board in the picture
Location information carry out ownership judgement, road signs information is associated in corresponding traffic mark board information;
S50, locating module believe traffic sign for obtaining location information of the traffic mark board in high-precision map
Breath and corresponding traffic mark board information are added in high-precision cartographic information.
It should be noted that using the friendship in the deep learning object detection method while detection image of anchor-free
Logical sign board and traffic sign, obtain respectively the parallel minimum circumscribed rectangle frame of traffic mark board in the picture location information and
The location information of the parallel minimum circumscribed rectangle of traffic sign in the picture.Specifically, using the deep learning of anchor-free
Object detection method detects image, can directly acquire the class of traffic mark board position in the picture and major class simultaneously
The position of other and traffic sign in the picture, that is, obtain the parallel minimum circumscribed rectangle frame of traffic mark board in the picture
Parallel minimum circumscribed rectangle of the location information with traffic sign in the picture location information.Currently used traffic sign inspection
Measured data collection has T100 and CCTSDB etc., these data sets can classify to traffic sign while detection, at this moment
It is divided into the classification for major class, that is, generally can be divided into prompt mark, caution sign and prohibitory sign.And it specifically turns to, directly
The classification of the group of row etc. also needs to carry out secondary classification.
It should be noted that generally directly can not more specifically be divided due to the unbalanced problem of data sample quantity
Class, such as specially speed limit limit for height mark.The embodiment of the present invention is divided into three categories, second for the first time by classifying twice
It is secondary to carry out detailed group classification again.Specifically, obtaining the classification information of the major class of traffic sign, the major class includes prompt mark
Will, caution sign and prohibitory sign, then using the classification information of major class as the target detection of same class, and the result that will test
As the input of traffic sign classifier, the classification information of the group of traffic sign is obtained, the group includes speed limit, limit for height,
Taboo is stopped.Because the classification that carries out in target detection is easy to be influenced by target background, and carries out the figure after single goal detection
As only including target object and a small amount of background information substantially, the influence of background can be effectively reduced.And for classifier
The problem of training sample, acquisition and the corresponding data enhancing of progress that can be more convenient, reduction data nonbalance.
It should be noted that including multiple traffic signs on traffic mark board, i.e., multiple traffic signs are located at same
On traffic mark board, in high-precision map, need for road signs information to be recorded corresponding traffic mark board information
In.In this way convenient for the accurately aesthetics of the storage of diagram data and high-precision map.Moreover, if respectively by a traffic
When traffic sign on sign board positions respectively, it is likely that due to deviation, so that being located at the traffic sign on the same traffic sign
Different several places, the false judgment for causing user to guide section are set positioned at high-precision map is upper.The number dispersed in this way
According to being also not easy to store.
It should be noted that the location information of the parallel minimum circumscribed rectangle frame using traffic mark board in the picture, is obtained
Take positioning terminal at a distance from traffic mark board;Positioning terminal is exactly acquisition terminal.Acquisition terminal is set in acquisition equipment, positioning
Terminal includes the located in connection device for acquiring data and needing, such as RTK module and IMU module, in conjunction with real-time dynamic carrier phase
Differential position and Inertial Measurement Unit obtain the location information of positioning terminal;It still can be accurate in the case where weak signal
Obtain the location information of positioning terminal.
In conjunction with the location information of positioning terminal and the range information of the positioning terminal and traffic mark board, traffic mark is obtained
Location information of the will board in high-precision map.Four corner location information for obtaining traffic mark board, pass through binocular ranging
Method calculates the actual range of pixel in the rectangle that positioning terminal and four angle point lines are formed, and by the average departure of actual range
From as the positioning terminal at a distance from traffic mark board.It should be noted that due to leading in parallel minimum circumscribed rectangle frame
Often include traffic mark board and other background informations, is easy to cause large effect to last distance measurement result.The present invention is real
Example is applied on the basis of target detection, by critical point detection, such as mtcnn, obtains traffic in parallel minimum circumscribed rectangle frame
Four angle points of sign board, and the actual range of pixel in angle point line is calculated, and using average distance as current road signs
Board is at a distance from positioning terminal.Finally in conjunction with positioning terminal location information and current road signs board Distance positioning terminal away from
From obtaining the world coordinates data of current road signs board, high-precision cartographic information be added.
Specifically, parallel minimum circumscribed rectangle frame is expanded as rectangle around its central point outward increases frame, increase in rectangle
By critical point detection in big frame, four corner location information of traffic mark board are obtained.Because of parallel minimum circumscribed rectangle frame
The periphery that traffic mark board in the picture may tightly be pasted, due to the influence of photo angle etc., traffic mark board is not face,
Four angle points of traffic mark board in image may not be enclosed in frame by minimum circumscribed rectangle frame, so first will be outside minimum
Connect rectangle frame around its central point expand as outward rectangle increase frame after, then detect traffic mark board four corner locations letter
Breath, but traffic mark board and other background informations are generally included in frame since rectangle increases, it is easy to last distance measurement result
Cause large effect.So pixel and positioning terminal in the rectangular extent for again obtaining four angle point lines it is practical away from
From, and using the average distance of the actual range as the distance of current road signs board.
Specifically, passing through four corner location information correcting images of traffic mark board;Image after correction is subjected to OCR
Identification, detects supplementary textual information;Supplementary textual information is associated in corresponding traffic mark board information, and will auxiliary text
Word information is added in high-precision cartographic information.Not only there is road signs information on one traffic mark board, there are also supplementary text letters
Breath, these also occupy the significant portion of entire sign board, carry out critical point detection on the basis of traffic mark board detection, and
Image flame detection is carried out by the key point of detection, and the image corrected carries out OCR identification as the input of RCNN.It will corresponding text
Word information association is added in high-precision cartographic information into corresponding traffic mark board information.First image flame detection is detected again,
Convenient for the accuracy rate of detection, because of shooting angle problem etc., text may be it is inclined, after correction, text face place, be convenient for
The accuracy rate of detection is improved, the accuracy rate of the high-precision map also just improved is correctly guided convenient for giving user, gone out safely
Row promotes user experience.
The invention also discloses a kind of computer readable storage mediums, are stored thereon with computer program, the computer
The step of any of the above-described the method is realized when program is executed by processor.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.
Within the scope of the knowledge and ability level for meeting those skilled in the art, the various embodiments or skill that are mentioned above
Art feature in the absence of conflict, can be combined with each other as other alternative embodiment, these are not by sieve one by one
Alternative embodiment list, that the limited quantity formed is combined by the technical characteristic of limited quantity, still fall within the invention discloses
Technical scope in, be also that those skilled in the art can be appreciated in conjunction with attached drawing and above or infer and obtain.
Finally it is stressed again that embodiment cited hereinabove is only used for for more typical, preferred embodiment of the invention
Be described in detail, explain technical solution of the present invention, in order to reader's understanding, the protection scope that is not intended to limit the invention or
Using.
Therefore, any modification, equivalent replacement, improvement and so on and the skill that obtains within the spirit and principles in the present invention
Art scheme should be all included within protection scope of the present invention.
Claims (10)
1. a kind of traffic mark board information collecting method for serving high-precision map producing characterized by comprising
Obtain traffic scene image information to be identified;
Obtain the location information of traffic mark board in the picture;
Obtain the location information of traffic sign in the picture;
Ownership is carried out using the location information of location information and traffic mark board in the picture of traffic sign in the picture to sentence
It is disconnected, road signs information is associated in corresponding traffic mark board information;
Location information of the traffic mark board in high-precision map is obtained, road signs information and corresponding traffic mark board are believed
Breath is added in high-precision cartographic information.
2. according to the method described in claim 1, it is characterized by: using anchor-free deep learning target detection side
It is outer to obtain the parallel minimum of traffic mark board in the picture respectively for traffic mark board and traffic sign in method while detection image
Connect the location information of parallel minimum circumscribed rectangle of the location information of rectangle frame with traffic sign in the picture.
3. according to the method described in claim 2, it is characterized by:
Using the location information of the parallel minimum circumscribed rectangle frame of traffic mark board in the picture, positioning terminal and traffic mark are obtained
The distance of will board;
The location information of positioning terminal is obtained in conjunction with real-time dynamic carrier phase difference location technology and Inertial Measurement Unit;
In conjunction with the location information of positioning terminal and the range information of the positioning terminal and traffic mark board, traffic mark board is obtained
Location information in high-precision map.
4. according to the method described in claim 3, it is characterized by: four corner location information of acquisition traffic mark board, lead to
The actual range of pixel in the rectangle that the method for crossing binocular ranging calculates positioning terminal and four angle point lines are formed, and will be practical
The average distance of distance is as the positioning terminal at a distance from traffic mark board.
5. according to the method described in claim 4, it is characterized by:
Parallel minimum circumscribed rectangle frame is expanded as into rectangle around its central point outward and increases frame, increases in frame in rectangle and passes through pass
The detection of key point, obtains four corner location information of traffic mark board.
6. according to the method described in claim 4, it is characterized by:
Pass through four corner location information correcting images of traffic mark board;
Image after correction is subjected to OCR identification, detects supplementary textual information;
Supplementary textual information is associated in corresponding traffic mark board information, and high-precision map is added in supplementary textual information
In information.
7. according to the method described in claim 2, it is characterized by:
The location information of the parallel minimum circumscribed rectangle frame of traffic mark board in the picture includes the centre coordinate of traffic mark board
Data, width data and altitude information;
The location information of the parallel minimum circumscribed rectangle of traffic sign in the picture includes the centre coordinate data of traffic sign;
Compare the centre coordinate data of traffic sign and centre coordinate data, width data and the altitude information of traffic mark board,
Road signs information is belonged in corresponding traffic mark board information.
8. according to the method described in claim 1, it is characterized by: the classification information of the major class of acquisition traffic sign, described big
Class includes prompt mark, caution sign and prohibitory sign, then using the classification information of major class as the target detection of same class, and
Input of the result that will test as traffic sign classifier obtains the classification information of the group of traffic sign, the group packet
Speed limit, limit for height are included, taboo is stopped.
9. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program quilt
The step of any one of claim 1-8 the method is realized when processor executes.
10. a kind of traffic mark board information acquisition system for serving high-precision map producing characterized by comprising
Image collection module, for obtaining traffic scene image information to be identified;
Traffic mark board obtains module, for obtaining the location information of traffic mark board in the picture;
Traffic sign obtains module, for obtaining the location information of traffic sign in the picture;
Belong to module, for believing using the position of location information and traffic mark board in the picture of traffic sign in the picture
Breath carries out ownership judgement, and road signs information is associated in corresponding traffic mark board information;
Locating module, for obtaining location information of the traffic mark board in high-precision map, by road signs information and accordingly
Traffic mark board information be added in high-precision cartographic information.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910745743.4A CN110501018B (en) | 2019-08-13 | 2019-08-13 | Traffic sign information acquisition method for high-precision map production |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910745743.4A CN110501018B (en) | 2019-08-13 | 2019-08-13 | Traffic sign information acquisition method for high-precision map production |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110501018A true CN110501018A (en) | 2019-11-26 |
CN110501018B CN110501018B (en) | 2021-11-02 |
Family
ID=68587509
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910745743.4A Active CN110501018B (en) | 2019-08-13 | 2019-08-13 | Traffic sign information acquisition method for high-precision map production |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110501018B (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111275000A (en) * | 2020-02-18 | 2020-06-12 | 广州敏视数码科技有限公司 | Traffic sign board detection method based on historical positioning data |
CN111291660A (en) * | 2020-01-21 | 2020-06-16 | 天津大学 | Anchor-free traffic sign identification method based on void convolution |
CN112071078A (en) * | 2020-09-01 | 2020-12-11 | 交科院检测技术(北京)有限公司 | Traffic engineering environment intelligent detection system |
CN112101299A (en) * | 2020-11-02 | 2020-12-18 | 立得空间信息技术股份有限公司 | Automatic traffic sign extraction method and system based on binocular CCD camera |
CN112132853A (en) * | 2020-11-30 | 2020-12-25 | 湖北亿咖通科技有限公司 | Method and device for constructing ground guide arrow, electronic equipment and storage medium |
CN112149624A (en) * | 2020-10-16 | 2020-12-29 | 腾讯科技(深圳)有限公司 | Traffic identification image processing method and device |
CN112509329A (en) * | 2020-12-05 | 2021-03-16 | 武汉中海庭数据技术有限公司 | Traffic sign data processing method, electronic device and storage medium |
WO2021104180A1 (en) * | 2019-11-29 | 2021-06-03 | 上海商汤临港智能科技有限公司 | Map generation method, positioning method, apparatus, device, storage medium, and computer program |
CN113011212A (en) * | 2019-12-19 | 2021-06-22 | 北京四维图新科技股份有限公司 | Image recognition method and device and vehicle |
CN113074749A (en) * | 2021-06-07 | 2021-07-06 | 湖北亿咖通科技有限公司 | Road condition detection and update method, electronic equipment and computer-readable storage medium |
CN113628168A (en) * | 2021-07-14 | 2021-11-09 | 深圳海翼智新科技有限公司 | Target detection method and device |
CN113673281A (en) * | 2020-05-14 | 2021-11-19 | 百度在线网络技术(北京)有限公司 | Speed limit information determining method, device, equipment and storage medium |
CN113947764A (en) * | 2021-12-06 | 2022-01-18 | 腾讯科技(深圳)有限公司 | Image processing method, device, equipment and storage medium |
Citations (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN201269758Y (en) * | 2008-09-22 | 2009-07-08 | 交通部公路科学研究所 | Vehicle mounted full automatic detection recording system for traffic signs |
US20110255743A1 (en) * | 2010-04-13 | 2011-10-20 | International Business Machines Corporation | Object recognition using haar features and histograms of oriented gradients |
JP2012168061A (en) * | 2011-02-15 | 2012-09-06 | Jvc Kenwood Corp | Navigation device, navigation method and program |
CN104637300A (en) * | 2015-02-06 | 2015-05-20 | 南京理工大学 | Information-based general survey analysis and display system for road traffic signs |
CN105718860A (en) * | 2016-01-15 | 2016-06-29 | 武汉光庭科技有限公司 | Positioning method and system based on safe driving map and binocular recognition of traffic signs |
CN106326858A (en) * | 2016-08-23 | 2017-01-11 | 北京航空航天大学 | Road traffic sign automatic identification and management system based on deep learning |
CN106372577A (en) * | 2016-08-23 | 2017-02-01 | 北京航空航天大学 | Deep learning-based traffic sign automatic identifying and marking method |
CN106525057A (en) * | 2016-10-26 | 2017-03-22 | 陈曦 | Generation system for high-precision road map |
CN106980855A (en) * | 2017-04-01 | 2017-07-25 | 公安部交通管理科学研究所 | Traffic sign quickly recognizes alignment system and method |
US20170236015A1 (en) * | 2016-02-16 | 2017-08-17 | Hitachi, Ltd. | Image processing device, alarming apparatus, image processing system, and image processing method |
CN107679508A (en) * | 2017-10-17 | 2018-02-09 | 广州汽车集团股份有限公司 | Road traffic sign detection recognition methods, apparatus and system |
CN107957266A (en) * | 2017-11-16 | 2018-04-24 | 北京小米移动软件有限公司 | Localization method, device and storage medium |
CN108090413A (en) * | 2017-11-21 | 2018-05-29 | 武汉中海庭数据技术有限公司 | A kind of traffic mark board correlating method and device |
CN108230357A (en) * | 2017-10-25 | 2018-06-29 | 北京市商汤科技开发有限公司 | Critical point detection method, apparatus, storage medium, computer program and electronic equipment |
US20180188060A1 (en) * | 2016-12-30 | 2018-07-05 | DeepMap Inc. | Traffic Sign Feature Creation for High Definition Maps Used for Navigating Autonomous Vehicles |
CN108846333A (en) * | 2018-05-30 | 2018-11-20 | 厦门大学 | Sign board landmark data collection generates and vehicle positioning method |
CN109255279A (en) * | 2017-07-13 | 2019-01-22 | 深圳市凯立德科技股份有限公司 | A kind of method and system of road traffic sign detection identification |
CN109409366A (en) * | 2018-10-30 | 2019-03-01 | 四川长虹电器股份有限公司 | Distorted image correction method and device based on Corner Detection |
-
2019
- 2019-08-13 CN CN201910745743.4A patent/CN110501018B/en active Active
Patent Citations (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN201269758Y (en) * | 2008-09-22 | 2009-07-08 | 交通部公路科学研究所 | Vehicle mounted full automatic detection recording system for traffic signs |
US20110255743A1 (en) * | 2010-04-13 | 2011-10-20 | International Business Machines Corporation | Object recognition using haar features and histograms of oriented gradients |
JP2012168061A (en) * | 2011-02-15 | 2012-09-06 | Jvc Kenwood Corp | Navigation device, navigation method and program |
CN104637300A (en) * | 2015-02-06 | 2015-05-20 | 南京理工大学 | Information-based general survey analysis and display system for road traffic signs |
CN105718860A (en) * | 2016-01-15 | 2016-06-29 | 武汉光庭科技有限公司 | Positioning method and system based on safe driving map and binocular recognition of traffic signs |
US20170236015A1 (en) * | 2016-02-16 | 2017-08-17 | Hitachi, Ltd. | Image processing device, alarming apparatus, image processing system, and image processing method |
CN106326858A (en) * | 2016-08-23 | 2017-01-11 | 北京航空航天大学 | Road traffic sign automatic identification and management system based on deep learning |
CN106372577A (en) * | 2016-08-23 | 2017-02-01 | 北京航空航天大学 | Deep learning-based traffic sign automatic identifying and marking method |
CN106525057A (en) * | 2016-10-26 | 2017-03-22 | 陈曦 | Generation system for high-precision road map |
US20180188060A1 (en) * | 2016-12-30 | 2018-07-05 | DeepMap Inc. | Traffic Sign Feature Creation for High Definition Maps Used for Navigating Autonomous Vehicles |
CN106980855A (en) * | 2017-04-01 | 2017-07-25 | 公安部交通管理科学研究所 | Traffic sign quickly recognizes alignment system and method |
CN109255279A (en) * | 2017-07-13 | 2019-01-22 | 深圳市凯立德科技股份有限公司 | A kind of method and system of road traffic sign detection identification |
CN107679508A (en) * | 2017-10-17 | 2018-02-09 | 广州汽车集团股份有限公司 | Road traffic sign detection recognition methods, apparatus and system |
CN108230357A (en) * | 2017-10-25 | 2018-06-29 | 北京市商汤科技开发有限公司 | Critical point detection method, apparatus, storage medium, computer program and electronic equipment |
CN107957266A (en) * | 2017-11-16 | 2018-04-24 | 北京小米移动软件有限公司 | Localization method, device and storage medium |
CN108090413A (en) * | 2017-11-21 | 2018-05-29 | 武汉中海庭数据技术有限公司 | A kind of traffic mark board correlating method and device |
CN108846333A (en) * | 2018-05-30 | 2018-11-20 | 厦门大学 | Sign board landmark data collection generates and vehicle positioning method |
CN109409366A (en) * | 2018-10-30 | 2019-03-01 | 四川长虹电器股份有限公司 | Distorted image correction method and device based on Corner Detection |
Non-Patent Citations (2)
Title |
---|
HEE SEOK LEE等: "Simultaneous Traffic Sign Detection and Boundary Estimation Using Convolutional Neural Network", 《IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS》 * |
徐仙伟等: "基于深度学习的交通标志识别算法", 《计算机时代》 * |
Cited By (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021104180A1 (en) * | 2019-11-29 | 2021-06-03 | 上海商汤临港智能科技有限公司 | Map generation method, positioning method, apparatus, device, storage medium, and computer program |
CN113011212A (en) * | 2019-12-19 | 2021-06-22 | 北京四维图新科技股份有限公司 | Image recognition method and device and vehicle |
CN111291660A (en) * | 2020-01-21 | 2020-06-16 | 天津大学 | Anchor-free traffic sign identification method based on void convolution |
CN111291660B (en) * | 2020-01-21 | 2022-08-12 | 天津大学 | Anchor-free traffic sign identification method based on void convolution |
CN111275000B (en) * | 2020-02-18 | 2023-05-02 | 广州敏视数码科技有限公司 | Traffic sign board detection method based on historical positioning data |
CN111275000A (en) * | 2020-02-18 | 2020-06-12 | 广州敏视数码科技有限公司 | Traffic sign board detection method based on historical positioning data |
CN113673281A (en) * | 2020-05-14 | 2021-11-19 | 百度在线网络技术(北京)有限公司 | Speed limit information determining method, device, equipment and storage medium |
CN113673281B (en) * | 2020-05-14 | 2023-10-03 | 百度在线网络技术(北京)有限公司 | Speed limit information determining method, device, equipment and storage medium |
CN112071078A (en) * | 2020-09-01 | 2020-12-11 | 交科院检测技术(北京)有限公司 | Traffic engineering environment intelligent detection system |
CN112149624A (en) * | 2020-10-16 | 2020-12-29 | 腾讯科技(深圳)有限公司 | Traffic identification image processing method and device |
CN112101299A (en) * | 2020-11-02 | 2020-12-18 | 立得空间信息技术股份有限公司 | Automatic traffic sign extraction method and system based on binocular CCD camera |
CN112132853A (en) * | 2020-11-30 | 2020-12-25 | 湖北亿咖通科技有限公司 | Method and device for constructing ground guide arrow, electronic equipment and storage medium |
CN112509329A (en) * | 2020-12-05 | 2021-03-16 | 武汉中海庭数据技术有限公司 | Traffic sign data processing method, electronic device and storage medium |
CN113074749A (en) * | 2021-06-07 | 2021-07-06 | 湖北亿咖通科技有限公司 | Road condition detection and update method, electronic equipment and computer-readable storage medium |
CN113074749B (en) * | 2021-06-07 | 2021-08-20 | 湖北亿咖通科技有限公司 | Road condition detection and update method, electronic equipment and computer-readable storage medium |
CN113628168A (en) * | 2021-07-14 | 2021-11-09 | 深圳海翼智新科技有限公司 | Target detection method and device |
CN113947764A (en) * | 2021-12-06 | 2022-01-18 | 腾讯科技(深圳)有限公司 | Image processing method, device, equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN110501018B (en) | 2021-11-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110501018A (en) | A kind of traffic mark board information collecting method for serving high-precision map producing | |
US11280630B2 (en) | Updating map data | |
CN106462727A (en) | Systems and methods for lane end recognition | |
CN103770704A (en) | System and method for recognizing parking space line markings for vehicle | |
CN108388641A (en) | A kind of means of transportation based on deep learning ground drawing generating method and system | |
CN110427860A (en) | A kind of Lane detection method, apparatus and storage medium | |
JP2010205160A (en) | Method for notifying speed-limit sign recognition result | |
CN109635737A (en) | Automobile navigation localization method is assisted based on pavement marker line visual identity | |
CN109426800A (en) | A kind of method for detecting lane lines and device | |
CN110969592A (en) | Image fusion method, automatic driving control method, device and equipment | |
CN108090459A (en) | A kind of road traffic sign detection recognition methods suitable for vehicle-mounted vision system | |
Billones et al. | Vehicle detection and tracking using corner feature points and artificial neural networks for a vision-based contactless apprehension system | |
US20220219700A1 (en) | Apparatus, method, and computer program for generating map | |
Bu et al. | A UAV photography–based detection method for defective road marking | |
JP2012215442A (en) | Own position determination system, own position determination program, own position determination method | |
CN111316324A (en) | Automatic driving simulation system, method, equipment and storage medium | |
CN111881752B (en) | Guardrail detection classification method and device, electronic equipment and storage medium | |
Guo et al. | Detection of occluded road signs on autonomous driving vehicles | |
CN116052124A (en) | Multi-camera generation local map template understanding enhanced target detection method and system | |
JP6031915B2 (en) | Image processing apparatus and program | |
TWI743637B (en) | Traffic light recognition system and method thereof | |
JP2022003548A (en) | Recognition system and program for the same | |
CN113255578A (en) | Traffic identification recognition method and device, electronic equipment and storage medium | |
Zhang et al. | Integration of scanning and image processing algorithms for lane detection based on fuzzy method | |
US20220316906A1 (en) | Apparatus and Method for Generating Navigational Plans |
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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |