CN109635782A - A method of obtaining unmanned required static traffic information - Google Patents
A method of obtaining unmanned required static traffic information Download PDFInfo
- Publication number
- CN109635782A CN109635782A CN201811651097.7A CN201811651097A CN109635782A CN 109635782 A CN109635782 A CN 109635782A CN 201811651097 A CN201811651097 A CN 201811651097A CN 109635782 A CN109635782 A CN 109635782A
- Authority
- CN
- China
- Prior art keywords
- traffic
- signboard
- information
- obtaining
- traffic lights
- 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 23
- 230000003068 static effect Effects 0.000 title claims abstract description 18
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 6
- 238000003708 edge detection Methods 0.000 claims abstract description 4
- 230000009466 transformation Effects 0.000 claims abstract description 4
- 238000012549 training Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 4
- 238000010586 diagram Methods 0.000 description 6
- 238000001514 detection method Methods 0.000 description 3
- 238000013480 data collection Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 201000009487 Amblyopia Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
Classifications
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- 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
- G06V20/588—Recognition of the road, e.g. of lane markings; Recognition of the vehicle driving pattern in relation to the road
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/09—Arrangements for giving variable traffic instructions
- G08G1/0962—Arrangements for giving variable traffic instructions having an indicator mounted inside the vehicle, e.g. giving voice messages
- G08G1/0967—Systems involving transmission of highway information, e.g. weather, speed limits
- G08G1/096708—Systems involving transmission of highway information, e.g. weather, speed limits where the received information might be used to generate an automatic action on the vehicle control
- G08G1/096725—Systems involving transmission of highway information, e.g. weather, speed limits where the received information might be used to generate an automatic action on the vehicle control where the received information generates an automatic action on the vehicle control
Abstract
The invention discloses a kind of methods for obtaining unmanned required static traffic information, comprising: the samples pictures of production traffic lights and traffic signboard；Pass through the contour feature of traffic lights and traffic signboard, filter out the image-region of doubtful traffic lights and traffic signboard, the sample to be predicted for obtaining traffic lights and traffic signboard, is separately input to corresponding convolutional neural networks for the image-region of traffic lights and traffic signboard and identifies；Lane line sample to be predicted filters other non-lane informations by color character, and marginal information is then extracted by the dynamic threshold of edge detection, then carries out Hough transformation and obtain straight line pose, and the straight line pose includes position, length and angle information.The present invention can acquire identification road information, provide road static information for the kinetic control system of unmanned vehicle, determine the traveling strategy of vehicle by capturing belisha beacon information, road sign board and lane line.
Description
Technical field
The invention belongs to unmanned technical fields, and in particular to it is a kind of obtain it is unmanned needed for static traffic information
Method.
Background technique
Pilotless automobile is one kind of intelligent automobile, also referred to as wheeled mobile robot, is relied primarily on interior in terms of
Intelligent driving instrument based on calculation machine system realizes unpiloted purpose.Vehicle periphery ring is perceived using onboard sensor
Border, and according to road, vehicle location and obstacle information obtained is perceived, the steering and speed of vehicle are controlled, to make vehicle
It can reliably and securely be travelled on road.
In the prior art for the identification side of static traffic information (traffic lights, traffic signboard and lane line etc.)
There are many methods, for example, Publication No. CN201710060178.9 patent document disclose it is a kind of carried out by automobile data recorder it is red
The method and system of green light identification, the method and system of traffic lights identification is carried out by automobile data recorder, wherein method includes: to obtain
Take the image information of traffic lights；According to the image information of traffic lights, the status information of analysis identification traffic lights；According to traffic lights
Status information issues the user with prompt.
Notification number is that the patent document of CN201220075023.5 discloses a kind of vehicle-mounted traffic lights identification display device, is wrapped
Signal receiver, controller and display are included, the signal receiver, display are electrically connected with controller respectively.It will be red green
Lamp is converted into character image and sound and light signal, and driver can be reminded to pay attention to the traffic lights of road ahead, be especially amblyopia and color
Blind motorist provides convenience.
There is identification targeted species are single, and identification range is smaller, and speed is slower etc. in existing traffic signboard recognition methods
Problem.
Summary of the invention
To solve the above-mentioned problems, the present invention provides a kind of method for obtaining unmanned required static traffic information, knows
Static traffic information needed for other Unmanned Systems provides necessary decision information for motion control unit.
The technical solution of the present invention is as follows: a kind of method for obtaining unmanned required static traffic information, including following step
It is rapid:
(1) lane line, traffic lights and traffic signboard are shot using starlight grade camera；
(2) samples pictures of traffic lights and traffic signboard are made；
(3) contour feature for passing through traffic lights and traffic signboard, filters out doubtful traffic lights and traffic mark
Know the image-region of board, the sample to be predicted of traffic lights and traffic signboard is obtained, by traffic lights and traffic mark
The image-region of board is separately input to corresponding convolutional neural networks and is identified；
Lane line sample to be predicted filters other non-lane informations by color character, then passes through the dynamic of edge detection
Threshold value extracts marginal information, then carries out Hough transformation and obtain straight line pose, and the straight line pose includes position, length and angle
It spends information and obtains the lane line pose of the right and left via the calculating ratio pair of straight line pose.
Preferably, the focal length of the starlight grade camera of acquisition traffic lights and traffic signboard is 12mm.
Preferably, the focal length of the starlight grade camera of acquisition lane line is 4mm.
Preferably, the samples pictures of traffic lights derive from the video flowing using section acquisition, traffic signboard
Samples pictures carry out screening on the basis of German GTSRB picture library and obtain.
Preferably, the contour feature of the traffic lights and traffic signboard includes number of edges, the height for forming profile
And at least one of width.
Preferably, samples pictures to be transformed to the image of 64x64, RGB channel is both configured to 255.
Preferably, 80% samples pictures are for instructing in the samples pictures of the traffic lights and traffic signboard
Practice, 20% samples pictures are for testing.
Preferably, the sample to be predicted screened in the step (3) is transformed to the image of 64x64, RGB channel is all
It is set as 255.
Compared with prior art, the beneficial effects of the present invention are embodied in:
The present invention can allow the detection of traffic lights and traffic signboard to reach in test segment recognition accuracy
To 94%.In the case where GPU is not used, operation can every frame < 100ms on earth.The present invention can acquire identification road information,
Road static information is provided for the kinetic control system of unmanned vehicle, by capturing belisha beacon information, road sign board is (as limited
Speed, lane type prohibit and the information such as stop) and lane line determine the traveling strategy of vehicle, so as to combine laser SLAM avoidance,
Differential GPS and GPS global map path planning make it possible automatic Pilot.
Detailed description of the invention
Fig. 1 is flow diagram of the invention.
Fig. 2 is object samples signal of the training set pretreatment-mxn object samples → 64x64 containing white background in the present invention
Figure.
Fig. 3 is the CNN network architecture-traffic lights and traffic signboard schematic diagram in the present invention.
Fig. 4 is the schematic diagram for identifying-turning left red light and round red light using section traffic lights.
Fig. 5 is the schematic diagram for identifying-turning left green light and circular green light using section traffic lights.
Fig. 6 is using road section traffic volume Sign Board identification-speed limit and the schematic diagram that no parking.
Fig. 7 is lane line-straight-line detection schematic diagram.
Specific embodiment
Embodiment 1
Traffic lights are identified that traffic signboard identification, Lane detection, obtains traffic in real time by the present embodiment
Information
The present embodiment is 12mm and 4mm starlight grade camera using focal length.Focal length is that the starlight grade camera of 12mm is used for
The identification of traffic lights and traffic signboard.Lane row of the starlight grade camera that focal length is 4mm for front side detects.It is burnt
Mainly capture the video information on ground or more away from the starlight grade camera for 12mm, and focal length be the starlight grade camera of 4mm then
Predominantly terrestrial information.Focusing 4mm camera calibrated, eliminates pattern distortion to a certain extent.
Method particularly includes: respectively traffic lights and traffic signboard make sample picture, and signal lamp atlas mainly comes
Derived from the screen stream in the acquisition of application section, and the atlas of Sign Board is that screening is carried out on the basis of German GTSRB picture library, is protected
It stays in domestic consistent sample, and adds the Sign Board atlas in related application section.Same sample will also by different size,
Different shading values are expanded.
By contour feature, such as form the number of edges of profile, high or wide threshold value, the ratio of width to height filters out doubtful traffic signals
The image-region of lamp and traffic signboard.And the contour area that intercepts in video flowing can be saved by the method to make training
Collection, including background image, thus identifying the contour area of suspect objects belongs to which class object or background using section,
Because the image-region filtered out by contour feature can have not comprising object itself, it may be possible to meet the background of contour feature.
Target area is being contained when making sample for the interference of identification in order to reduce data collection background component in sample
Because intercepting rectangle along target edges in the picture in domain, then all samples are converted to the image of 64x64, that is, make object
Foreground picture.Image should not stretch, and no side object is easy distortion, influence discrimination, but data collection background is left white, i.e. RGB is logical
Road is all set as 255.
When making sample, in addition random 80% sample 20% is used for test sample for training.Respectively traffic is believed
Signal lamp and the corresponding convolutional neural networks of Sign Board training.
The image-region of traffic lights or traffic signboard is separately input to corresponding convolutional neural networks to know
Not.Equally during prediction, the sample to be predicted obtained by the screening of profile characteristic should also be converted thereof into as what is mentioned in 4
The sample that 64x64 background is left white, for predicting.
Wherein, lane line is then to first pass through color character to filter other non-lane informations, passes through the dynamic of edge detection afterwards
Threshold value come extract edge letter (ADAPTIV_THRESH_MEAN), then carry out Hough transformation obtain straight line pose (position, length,
Angle), it (can be counted later by a variety of existing methods, such as the present invention can pass through the line segment of cluster by statistics
Positional relationship and slope difference degree, if be classified as Ganlei, then calculate fall within such so the position at the center of line segment
Set with G-bar to get to its pose), obtain the lane line pose of the right and left.
Claims (8)
1. a kind of method for obtaining unmanned required static traffic information, which comprises the following steps:
(1) lane line, traffic lights and traffic signboard are shot using starlight grade camera；
(2) samples pictures of traffic lights and traffic signboard are made；
(3) contour feature for passing through traffic lights and traffic signboard, filters out doubtful traffic lights and traffic signboard
Image-region, the sample to be predicted of traffic lights and traffic signboard is obtained, by traffic lights and traffic signboard
Image-region is separately input to corresponding convolutional neural networks and is identified；
Lane line sample to be predicted filters other non-lane informations by color character, then passes through the dynamic threshold of edge detection
Marginal information is extracted, then carries out Hough transformation and obtains straight line pose, the straight line pose includes position, length and angle letter
Breath, via the calculating ratio pair of straight line pose, obtains the lane line pose of the right and left.
2. obtaining the method for unmanned required static traffic information as described in claim 1, which is characterized in that acquisition traffic
The focal length of the starlight grade camera of signal lamp and traffic signboard is 12mm.
3. obtaining the method for unmanned required static traffic information as described in claim 1, which is characterized in that acquisition lane
The focal length of the starlight grade camera of line is 4mm.
4. obtaining the method for unmanned required static traffic information as described in claim 1, which is characterized in that traffic signals
The samples pictures of lamp derive from the video flowing using section acquisition, and the samples pictures of traffic signboard are in German GTSRB picture library
On the basis of carry out screening obtain.
5. obtaining the method for unmanned required static traffic information as described in claim 1, which is characterized in that the traffic
The contour feature of signal lamp and traffic signboard includes at least one of number of edges, height and the width for forming profile.
6. obtaining the method for unmanned required static traffic information as described in claim 1, which is characterized in that by sample graph
Piece is transformed to the image of 64x 64, and RGB channel is both configured to 255.
7. obtaining the method for unmanned required static traffic information as described in claim 1, which is characterized in that the traffic
80% samples pictures are for training in the samples pictures of signal lamp and traffic signboard, and 20% samples pictures are for testing.
8. obtaining the method for unmanned required static traffic information as described in claim 1, which is characterized in that the step
(3) sample to be predicted screened in is transformed to the image of 64x 64, and RGB channel is both configured to 255.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811651097.7A CN109635782A (en) | 2018-12-31 | 2018-12-31 | A method of obtaining unmanned required static traffic information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811651097.7A CN109635782A (en) | 2018-12-31 | 2018-12-31 | A method of obtaining unmanned required static traffic information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109635782A true CN109635782A (en) | 2019-04-16 |
Family
ID=66056287
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811651097.7A Pending CN109635782A (en) | 2018-12-31 | 2018-12-31 | A method of obtaining unmanned required static traffic information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109635782A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110188631A (en) * | 2019-05-14 | 2019-08-30 | 重庆大学 | A kind of freeway tunnel car light dividing method |
| CN111079680A (en) * | 2019-12-23 | 2020-04-28 | 北京三快在线科技有限公司 | Temporary traffic signal lamp detection method and device and automatic driving equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106373426A (en) * | 2016-09-29 | 2017-02-01 | 成都通甲优博科技有限责任公司 | Computer vision-based parking space and illegal lane occupying parking monitoring method |
| CN107609472A (en) * | 2017-08-04 | 2018-01-19 | 湖南星云智能科技有限公司 | A kind of pilotless automobile NI Vision Builder for Automated Inspection based on vehicle-mounted dual camera |
| WO2018201835A1 (en) * | 2017-05-03 | 2018-11-08 | 腾讯科技（深圳）有限公司 | Signal light state recognition method, device and vehicle-mounted control terminal and motor vehicle |
-
2018
- 2018-12-31 CN CN201811651097.7A patent/CN109635782A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106373426A (en) * | 2016-09-29 | 2017-02-01 | 成都通甲优博科技有限责任公司 | Computer vision-based parking space and illegal lane occupying parking monitoring method |
| WO2018201835A1 (en) * | 2017-05-03 | 2018-11-08 | 腾讯科技（深圳）有限公司 | Signal light state recognition method, device and vehicle-mounted control terminal and motor vehicle |
| CN107609472A (en) * | 2017-08-04 | 2018-01-19 | 湖南星云智能科技有限公司 | A kind of pilotless automobile NI Vision Builder for Automated Inspection based on vehicle-mounted dual camera |
Non-Patent Citations (1)
| Title |
|---|
| 李海霞等: "汽车辅助驾驶系统交通信号灯识别", 《电子技术与软件工程》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110188631A (en) * | 2019-05-14 | 2019-08-30 | 重庆大学 | A kind of freeway tunnel car light dividing method |
| CN111079680A (en) * | 2019-12-23 | 2020-04-28 | 北京三快在线科技有限公司 | Temporary traffic signal lamp detection method and device and automatic driving equipment |
| CN111079680B (en) * | 2019-12-23 | 2023-09-19 | 北京三快在线科技有限公司 | Temporary traffic signal lamp detection method and device and automatic driving equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1930863B1 (en) | Detecting and recognizing traffic signs | |
| US11836985B2 (en) | Identifying suspicious entities using autonomous vehicles | |
| EP2501585B1 (en) | Method of control of a vehicle headlamp and headlamp arrangement | |
| EP2501584B1 (en) | Method for controlling a headlamp arrangement for vehicle and headlamp arrangement | |
| US10671071B2 (en) | Control method and control device of automatic driving vehicle | |
| EP2057583B1 (en) | Automatic high beam control method | |
| CN106909937A (en) | Traffic lights recognition methods, control method for vehicle, device and vehicle | |
| CN107978165A (en) | Intersection identifier marking and signal lamp Intellisense method based on computer vision | |
| CN111508260A (en) | Vehicle parking space detection method, device and system | |
| CN105260701B (en) | A kind of front vehicles detection method suitable under complex scene | |
| DE102009050505A1 (en) | Clear path detecting method for vehicle i.e. motor vehicle such as car, involves modifying clear path based upon analysis of road geometry data, and utilizing clear path in navigation of vehicle | |
| CN110298216A (en) | Vehicle deviation warning method based on lane line gradient image adaptive threshold fuzziness | |
| CN104361350A (en) | Traffic sign identification system | |
| CN104021378A (en) | Real-time traffic light recognition method based on space-time correlation and priori knowledge | |
| CA2967138A1 (en) | Method of controlling a traffic surveillance system | |
| CN105654073A (en) | Automatic speed control method based on visual detection | |
| KR102204037B1 (en) | Alignment error correction system for collected location information by triangle mesh composed rand-mark and specified object | |
| CN113687310A (en) | Object detection system for an automated vehicle | |
| CN107609472A (en) | A kind of pilotless automobile NI Vision Builder for Automated Inspection based on vehicle-mounted dual camera | |
| CN109635782A (en) | A method of obtaining unmanned required static traffic information | |
| CN108108680A (en) | A kind of front vehicle identification and distance measuring method based on binocular vision | |
| KR101134857B1 (en) | Apparatus and method for detecting a navigation vehicle in day and night according to luminous state | |
| Li et al. | A low-cost and fast vehicle detection algorithm with a monocular camera for adaptive driving beam systems | |
| KR102204051B1 (en) | Mms image editting system having blur menu for image | |
| CN113989772A (en) | Traffic light detection method and device, vehicle and readable storage medium |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190416 |