CN104778846B - A kind of method for controlling traffic signal lights based on computer vision - Google Patents
A kind of method for controlling traffic signal lights based on computer vision Download PDFInfo
- Publication number
- CN104778846B CN104778846B CN201510136919.8A CN201510136919A CN104778846B CN 104778846 B CN104778846 B CN 104778846B CN 201510136919 A CN201510136919 A CN 201510136919A CN 104778846 B CN104778846 B CN 104778846B
- Authority
- CN
- China
- Prior art keywords
- time
- crossing
- vehicle
- traffic
- green light
- 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.)
- Expired - Fee Related
Links
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/07—Controlling traffic signals
- G08G1/08—Controlling traffic signals according to detected number or speed of vehicles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention provides a kind of method for controlling traffic signal lights based on computer vision, the method first pass through to be arranged on the camera collection at Traffic signal control crossing to video process, it is taken out series of frames, then these frames are processed by image processing techniques, extract the position of vehicle, quantity information also calculates the vehicle real-time distance to crossing, real-time speed and the average speed of all vehicles also predict that vehicle arrives the time at crossing, finally by the analysis of vehicle location being determined the guarantee vehicle time by crossing, show that greensignal light also needs the time kept by selecting K central point clustering algorithm or Clara clustering algorithm that the vehicle program arrival crossing time carries out cluster analysis flexibly, thereby determine that the time of traffic light alternate transition.The present invention is capable of Traffic signal control dynamic, adaptive, reduces the probability that traffic congestion occurs, and reaches to optimize the purpose of Traffic signal control.
Description
Technical field
The present invention designs a kind of method for controlling traffic signal lights based on computer vision, and the method passes through image procossing skill
Art, data mining technology improve the efficiency of traffic light.Belong to the intersection skill of information physical emerging system and data mining
Art application.
Background technology
Image processing techniques is to be analyzed image with computer, to reach the technology of results needed, also known as image processing.
Image procossing refers generally to Digital Image Processing.Digital picture refers to pass through with equipment such as industrial camera, video camera, scanneies
The big two-dimensional array that shooting obtains, the element of this array is referred to as pixel, and its value is referred to as gray value.Image procossing skill
Art generally comprise compression of images, strengthen and restore, mating, describe and identify 3 parts.Common system has health
Resistance to viewing system, figure intelligence system etc., be currently the technology the most gradually risen.
Information physical emerging system is the multi-dimensional complicated system of a COMPREHENSIVE CALCULATING, network and physical environment, passes through 3C
Organically blending of (Computation, Communication, Control) technology cooperates with the degree of depth, it is achieved heavy construction
The real-time perception of system, dynamically control and information service.Information physical emerging system realizes calculating, communicating and physical system
Integrated design, can make that system is relatively reliable, efficient, real-time collaborative, have important and be widely applied prospect.Closely
Nian Lai, information physical emerging system has become domestic and international academia and the important directions of scientific and technological circle's research and development the most, it is contemplated that
Also the industrial field that business circles are first developed will be become.Carry out the research of information physical emerging system with application for accelerating China
Cultivate and advance industrialization significant with information-based fusion.
Data mining is an iterative process, and it searches valuable, unusual fresh information from substantial amounts of data,
It is people and computer result of the joint efforts;It human expert describe the knowledge of problem and target and computer search capability it
Between seeking balance, in the hope of obtaining best result.Data mining is one of field with fastest developing speed in computer industry, with
Before it is a theme in computer science and statistics, nowadays, it developed rapidly become one independent
Field.The most powerful advantage of data mining is that it can be many methods and technology application and substantial amounts of problem set.
Data mining is a non-human act carried out on large data sets, so the target market of its maximum is whole data bins
Storehouse, Data Mart and decision support industry, including such as retail, manufacture, telecommunications, medical treatment, insure, the industry such as transport.
The concept of light stream is that first Gibson put forward in nineteen fifty.It is that space motion object is observing imaging plane
On the instantaneous velocity of pixel motion, be to utilize in image sequence between pixel change and consecutive frame in time domain
Dependency finds previous frame with the corresponding relation existed between present frame, thus calculates the motion of object between consecutive frame
A kind of method of information.It is said that in general, light stream is due to the motion of the movement of foreground target itself, camera in scene, or
Produced by the associated movement of both persons.Its computational methods can be divided three classes: based on region or feature-based matching
Method;Method based on frequency domain;Method based on gradient;In simple terms, light stream is that space motion object is in observation imaging
" instantaneous velocity " of the pixel motion in plane.The research of light stream be utilize pixel intensity data in image sequence time
Territory change and dependency determine " moving " of respective location of pixels.The purpose of research optical flow field is contemplated to from picture sequence
In row, approximation obtains the sports ground being not directly available.
Summary of the invention
Technical problem: the most most basic Traffic signal control mode is the timing control system of single crossing, also cries
Static line Ore-controlling Role.I.e. one day only by a timing scheme or the volume of traffic multiple timing schemes of employing of pressing different periods,
Lack motility, it is impossible to reach optimum or suboptimal control, it is an object of the invention to provide a kind of friendship based on computer vision
Ventilating signal lamp control method, the method increases traffic flow during Traffic signal control, the consideration of waiting time,
Improve Traffic signal control efficiency, reduce occur traffic congestion probability and
Reduce and block up the time.
Technical scheme: method for controlling traffic signal lights based on computer vision of the present invention is by the place to video
Reason extracts traffic flow data, and distributes by the analysis of traffic flow data realizes the traffic light dynamic time.
Method for controlling traffic signal lights based on computer vision of the present invention comprises the following steps:
Step 1): setting traffic light and be shown as green light, user sets the process interval time of traffic video, green light
The expansion time that the basic time kept and green light keep, described traffic video is provided by user, and this video is located in handing over
Ventilating signal lamp controls the video that the photographic head continuous acquisition at crossing arrives, and gathers in the range of the car between two traffic light
Information, including quantity, the positional information of vehicle;The basic time that described green light keeps is that green light ensures that vehicle is by green
The time of lamp, described green light keep the expansion time be according to obtain vehicle data analysis process after green light extend time
Between.
Step 2): extract the interval time that user sets from the traffic video that the crossing that traffic light are green light gathers
Interior untreated video, this video is from the beginning of the frame after last processed video trailer, according to the interposition of this video
The two field picture put calculates each car distance to crossing, and concrete handling process is: enter this two field picture weighted mean method
Row gray processing processes and obtains the gray value function f of image (i, j), wherein (i j) is the coordinate of pixel;With drawing of second order
General Laplacian operater carries out rim detection to the picture after gray processing, distinguishes automobile image element and foreground image elements, connects
Image function f to input (i, j) does following process, it is thus achieved that the image function g of output (i, j):
Wherein, T is threshold value, if pictorial element is automobile, then set g (i, j)=1, if pictorial element is background, then
Setting g (i, j)=0;Remove image edge center noise spot with the median filtering algorithm of spatial domain and reduce error, according to
The distribution situation in each pixel field determines whether this point is deleted;Extract the pixel coordinate value of all vehicles and calculate every
Car is to the real-time distance of stop line;
Described weighted mean method is according to importance and other index, is added with different weights by the three of picture components
Weight average;Described second order Laplace operator be the divergence of the gradient of a function in Euclidean space be given micro-
Divide operator;The space that described spatial domain is made up of pictorial element.Described median filtering method is digital picture or numeral sequence
In row, the value of some Mesophyticum of each point value in one neighborhood of this point replaces, and allows the actual value that the pixel value of surrounding is close,
Thus the method eliminating isolated noise spot.
Step 3): calculate the speed of each car successively according to the pixel coordinate value of all vehicles extracted, each car tests the speed
Specifically comprise the following steps that
Step 31): choose fixing road background image and on image, choose two fixing velocity projections lines;
Step 32): obtain the pixel coordinate of vehicle, according to formula:Calculate the speed v that vehicle is real-time1;Root
According to formulaCalculating is by the average speed of all vehicles of two pre-set velocity projection lines and with this speed
Spend the average speed v as all vehicles2.Wherein e pixel distance and the ratio of actual range;Δ le is that pixel is poor, Δ t
For time difference, le1For vehicle at the pixel of the first rule velocity projections line, le2For vehicle in the second rule velocity projections
The pixel of line, t1For target at the time of Article 1 velocity projections line, t2For target in time of Article 2 velocity projections line;
N is by two velocity projections line vehicle numbers;
Step 4): whether there is fleet to be believed by traffic according to the real-time Distance Judgment at each crossing, Che Dao road calculated
Signal lamp state is the crossing of green light, if having, navigates to a car away from crossing distance maximum and every a default time
Sheet just makees an one-time detection to the last car and by green light or reaches the green light basic time of user preset and by user
The green light basic time preset is with being newly this actually used time.The described condition becoming a fleet is front and back two
Car by the time interval of stop line less than 3 seconds;
Step 5): pre-to distance and each car of average speed the two parameter prediction of stop line according to the vehicle recorded
Meter arrives the time at crossing and writes data set, according to the data vehicle that i.e. vehicle has been waited in line at crossing that speed is zero
Number and time calculate the weighted cumulative latency value of all waits in this crossing vehicle;
Step 6): select different algorithms according to the size of the data set of acquired relevant time, if data are higher than one
Individual preset critical then performs step 8) otherwise continue down to perform.Described selection algorithms of different purpose is: in data volume
Hour select general algorithm to reduce system loading and to improve accuracy, select when data volume is bigger another kind of algorithm to guarantee
SRT;
Step 7): utilizing K central point clustering algorithm that time data carries out cluster analysis, described K central point cluster is calculated
Method is a kind of to select apoplexy due to endogenous wind from the nearest object of meansigma methods as the algorithm of cluster centre, and this algorithm is suitable for data on a small scale,
Specifically comprise the following steps that
Step 71): receive data set;Red and green two states K central point clustering algorithm is had to incite somebody to action according to traffic lights
Data set is polymerized to two classes;
Step 72): the time conduct of that apoplexy due to endogenous wind maximum selecting cluster centre point less is concentrated in two data being divided into
The time that greensignal light extends;
Step 8): utilizing Clara clustering algorithm that time data carries out cluster analysis, described Clara clustering algorithm is one
Planting clustering method based on sampling, it can process substantial amounts of data and have fireballing advantage, specifically comprising the following steps that
Step 81): the data set of reception;Red and green two states Clara clustering algorithm is had to incite somebody to action according to traffic lights
Data set is polymerized to two classes;
Step 82): using the time maximum for an apoplexy due to endogenous wind less for cluster centre point as green extension;
Step 9): using the prolongation time of the greensignal light of acquisition as countdown initial value, when countdown is zero or reaches
When the time expanded by green light to user preset, this crossroad traffic signal lamp is turned red light by green light and will use by control system automatically
The green light that family is preset expands the time according to being newly this actually used time, by weighting in crossing that traffic light are red light
The traffic light at cumulative latency value maximum crossing are turned green light by red light.Go back to step 2).
Beneficial effect: the present invention proposes the intelligent traffic lamp control method of a kind of Information physics emerging system.
Traffic light network is optimized by the method based on computer vision, reduces the vehicle average latency, decreases
There is the probability of traffic congestion, and Intelligent treatment traffic congestion.Specifically, improved method of the present invention have as
Under beneficial effect:
(1) Traffic signal control of the present invention uses dynamic time division methods, compares being manually set of static state
The control method of time can set the traffic lights time intelligently according to real-time traffic flow, improves the effect of transportation
Rate, reduces the time that vehicle waits;
(2) the weighted cumulative waiting time of the present invention can effectively prevent traffic congestion and traffic congestion is occurring
The jam situation in each section of crossing intelligent decision, prior-release congested link vehicle, the mesh of the traffic pressure that reaches to releive
's.
Accompanying drawing explanation
Fig. 1 is the method flow diagram of the present invention.
Detailed description of the invention
One, the video of the camera collection being arranged on road is processed
In being embodied as, mainly analyze due to this patent and process the traffic behavior institute that traffic light are green light crossing
Initially set traffic light with system and be shown as green light, be set by the user and process the interval time of traffic video, green light guarantor
The expansion time that the basic time held and green light keep, carry from the traffic video that the crossing that traffic light are green light gathers
Taking the untreated video in the interval time that family sets, this video frame after last processed video trailer is opened
Begin, calculate each car distance to crossing according to a two field picture in the centre position of this video.Set interval time and carry
The purpose of the frame taking untreated video centre position is reduction system operating load, it is to avoid because video to be processed in every
One frame and cause systematic function to decline, and analysis and result to data after having no effect on.Concrete each frame figure
As handling process is: extract a two field picture from the video of continuous acquisition, this two field picture weighted mean method is carried out gray scale
Change processes and obtains the gray value function f of image (i, j), wherein (i j) is the coordinate of pixel;Calculate with the Laplce of second order
Son carries out rim detection to the picture after gray processing, distinguishes automobile image element and foreground image elements, then to input
Image f (i, j) does following process, it is thus achieved that the image g of output (i, j):
Wherein, T is threshold value, if pictorial element is vehicle, then set g (i, j)=1, if pictorial element is background, then set
G (i, j)=0;Removing image edge center noise spot with the median filtering algorithm of spatial domain, removing noise spot can have
Effect reduces error, improves the precision of the vehicle pixel coordinate obtained, and the distribution situation according to judging each pixel field is come
Determine band point whether to delete finally vehicle image is reduced into a point to represent.Finally extract the pixel coordinate of all vehicles
Value.
Two, each car distance, the real-time speed of each car and average speed of all cars to stop line is calculated respectively
In being embodied as, extract the pixel coordinate value of vehicle, according to formula d=e × Δ le1Calculate each car respectively to stopping
The distance of line, wherein e is the ratio of pixel distance and actual range, is the parameter needing to measure in advance;Δle1On image
Each car is poor to the pixel of stop line.
Extract the pixel coordinate of vehicle, according to formula:Calculate the real-time speed of each car, wherein Δ le respectively2
Poor for the pixel on same adjacent two two field pictures of car, Δ t is the time difference of adjacent two frames;
Road image pretreatment is obtained the road background image fixed and on image, chooses two fixing velocity projections
Line, obtains the headstock position of each car by the pixel coordinate during velocity projections line preset and corresponding time, because
The actual traveling on road when medium velocity is of vehicle is engraved in change, and this time just arriving crossing for prediction vehicle brings
Difficulty, so needing the average speed travelled in certain specific road section with all vehicles to arrive the speed at crossing as prediction vehicle
Parameter, here according to formulaCalculate all vehicles average speed by two pre-set velocity projection lines
Spend the speed parameter arriving crossing as prediction vehicle.Wherein le1It is the vehicle pixel at the first rule velocity projections line,
le2It is the vehicle pixel at the second rule velocity projections line, t1It is the vehicle time at Article 1 velocity projections line, t2It is
Vehicle is in the time of Article 2 velocity projections line;N is by two velocity projections line vehicle numbers.
Three, analytical data determines the time that this crossing greensignal light maintains, and this time is to include basic green time and open up
The green time of exhibition
In being embodied as, fleet whether is had to pass through to hand over according to the real-time Distance Judgment at each crossing, Che Dao road calculated
Ventilating signal lamp state is the crossing of green light, using front and back's two cars by the time interval of same position within 3 seconds as
One fleet, if having, navigates to last car of fleet, it is judged that be last according to being this car distance in fleet
Crossing distance is maximum, just makees last car of one-time detection to the last fleet every a default timeslice and passes through green light
Or reach the basic time that default green light keeps, terminate the basic time that now green light keeps, and update green light holding
Basic time be actually used time in this stage.
Distance and each car of average speed the two parameter prediction anticipated arrival road according to the vehicle recorded to stop line
Mouthful time and write data set, according to the algorithm how many selections of the data set of acquired relevant time are different, if number
Then K central point focusing solutions analysis is otherwise utilized with Clara clustering algorithm higher than a preset critical according to collection.Select not
It is to reduce system loading and to improve accuracy at data volume hour selection general algorithm with algorithm purpose, bigger in data volume
Time select another kind of algorithm to guarantee SRT, be thus effectively saved overhead, beneficially raising system
Service life.Red and green two states is had to utilize the algorithm selected that time data is polymerized to two further according to traffic lights
Individual data set, when two data concentrate the time of that apoplexy due to endogenous wind maximum selecting cluster centre point less as countdown
Between, the expansion time that countdown terminates or the green light of user preset that reaches keeps, the expansion time that now green light keeps
Terminating and update the expansion time that green light keeps is actually used time in this stage.
Four, calculate the weighted cumulative latency value of an all car in crossing, and perform the transformation of traffic light
In being embodied as, count according to vehicle number and the time that the data that speed is zero i.e. vehicle has been waited in line at crossing
Calculate the w=∑ k of all waits in this crossing vehicleiT, wherein kiFor weights, elapsing these weights over time can be increasingly
Greatly, t is the waiting time of vehicle.It is red for introducing weighted cumulative latency value and can effectively assessing each traffic light
The vehicle flowrate situation at lamp crossing, the crossing that prior-release weighted cumulative latency value is high.Then green time is terminated
When crossroad traffic signal lamp is turned red light by green light and is selected total weighted cumulative to wait in the crossing that traffic light are red light
Between be worth the traffic light at crossing of maximum and turned green light by red light and empty this crossing weighted cumulative latency value, finally return
Return the first step to repeat.
Claims (2)
1. a method for controlling traffic signal lights based on computer vision, it is characterised in that the step that the method is comprised
For:
Step 1): setting traffic light and be shown as green light, user sets the process interval time of traffic video, green light
The expansion time that the basic time kept and green light keep, described traffic video is provided by user, and this video is located in handing over
Ventilating signal lamp controls the video that the photographic head continuous acquisition at crossing arrives, and gathers in the range of the car between two traffic light
Quantity and positional information;The basic time that described green light keeps is that green light ensures the vehicle time by green light, described
The expansion time that green light keeps is the time extended according to green light after the vehicle data analysis process obtained;
Step 2): extract the interval time that user sets from the traffic video that the crossing that traffic light are green light gathers
Interior untreated video, this video is from the beginning of the frame after last processed video trailer, according to the interposition of this video
The two field picture put calculates each car distance to crossing, and concrete handling process is: enter this two field picture weighted mean method
Row gray processing processes and obtains the gray value function f of image (i, j), wherein (i j) is the coordinate of pixel;With drawing of second order
General Laplacian operater carries out rim detection to the picture after gray processing, distinguishes automobile image element and foreground image elements, connects
Image function f to input (i, j) does following process, it is thus achieved that the image function g of output (i, j):
Wherein, T is threshold value, if pictorial element is automobile, then set g (i, j)=1, if pictorial element is background, then
Setting g (i, j)=0;Remove image edge center noise spot with the median filtering algorithm of spatial domain and reduce error, according to
The distribution situation in each pixel field determines whether this point is deleted;Extract the pixel coordinate value of all vehicles and calculate every
Car is to the real-time distance of stop line;
Step 3): calculate the speed of each car successively according to the pixel coordinate value of all vehicles extracted, each car tests the speed
Specifically comprise the following steps that
Step 31): choose fixing road background image and on image, choose two fixing velocity projections lines;
Step 32): obtain the pixel coordinate of vehicle, according to formula:Calculate the speed v that vehicle is real-time1;Root
According to formulaCalculating is by the average speed of all vehicles of two pre-set velocity projection lines and with this speed
Spend the average speed v as all vehicles2;Wherein e pixel distance and the ratio of actual range;Δ le is that pixel is poor, Δ t
For time difference, le1For vehicle at the pixel of the first rule velocity projections line, le2For vehicle at Article 2 velocity projections line
Pixel, t1For target at the time of Article 1 velocity projections line, t2For target in time of Article 2 velocity projections line;
N is by two velocity projections line vehicle numbers;
Step 4): whether there is fleet to be believed by traffic according to the real-time Distance Judgment at each crossing, Che Dao road calculated
Signal lamp state is the crossing of green light, if having, navigates to a car away from crossing distance maximum and every a default time
Sheet just makees an one-time detection to the last car and by green light or reaches the green light basic time of user preset and by user
The green light basic time preset is with being newly this actually used time;The condition becoming a fleet is led to for two cars front and back
Cross the time interval of stop line less than 3 seconds;
Step 5): pre-to distance and each car of average speed the two parameter prediction of stop line according to the vehicle recorded
Meter arrives the time at crossing and writes data set, according to the data vehicle that i.e. vehicle has been waited in line at crossing that speed is zero
Number and time calculate the weighted cumulative latency value of all waits in this crossing vehicle;
Step 6): select different algorithms according to the size of the data set of acquired relevant time, if data are higher than one
Individual preset critical then performs step 8) otherwise continue down to perform;Described selection algorithms of different is: in data volume hour
Select general algorithm, select another kind of algorithm when data volume is bigger;
Step 7): utilizing K central point clustering algorithm that time data carries out cluster analysis, described K central point cluster is calculated
Method is a kind of to select apoplexy due to endogenous wind from the nearest object of meansigma methods as the algorithm of cluster centre, and this algorithm is suitable for data on a small scale,
Specifically comprise the following steps that
Step 7.1): receive data set;Red and green two states K central point clustering algorithm is had according to traffic lights
Data set is polymerized to two classes;
Step 7.2): the time catch cropping of that apoplexy due to endogenous wind maximum selecting cluster centre point less is concentrated in two data being divided into
The time extended for greensignal light;
Step 8): utilizing Clara clustering algorithm that time data carries out cluster analysis, described Clara clustering algorithm is one
Planting clustering method based on sampling, it can process substantial amounts of data and have fireballing advantage, specifically comprising the following steps that
Step 8.1): the data set of reception;Red and green two states Clara clustering algorithm is had to incite somebody to action according to traffic lights
Data set is polymerized to two classes;
Step 8.2): using the time maximum for an apoplexy due to endogenous wind less for cluster centre point as green extension;
Step 9): using the prolongation time of the greensignal light of acquisition as countdown initial value, when countdown is zero or reaches
When the time expanded by green light to user preset, this crossroad traffic signal lamp is turned red light by green light and will use by control system automatically
The green light that family is preset expands the time according to being newly this actually used time, by weighting in crossing that traffic light are red light
The traffic light at cumulative latency value maximum crossing are turned green light by red light;Go back to step 2).
Method for controlling traffic signal lights based on computer vision the most according to claim 1, it is characterised in that
Described weighted mean method is according to importance and other index, is weighted putting down with different weights by the three of picture components
All;Described second order Laplace operator is that the differential that the divergence of the gradient of a function in Euclidean space is given is calculated
Son;The space that described spatial domain is made up of pictorial element;Described median filtering method is in digital picture or Serial No.
The value of some Mesophyticum of each point value in one neighborhood of this point replaces, and allows the pixel value of surrounding close to actual value, thus disappears
Except isolated noise point.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510136919.8A CN104778846B (en) | 2015-03-26 | 2015-03-26 | A kind of method for controlling traffic signal lights based on computer vision |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510136919.8A CN104778846B (en) | 2015-03-26 | 2015-03-26 | A kind of method for controlling traffic signal lights based on computer vision |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104778846A CN104778846A (en) | 2015-07-15 |
CN104778846B true CN104778846B (en) | 2016-09-28 |
Family
ID=53620286
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510136919.8A Expired - Fee Related CN104778846B (en) | 2015-03-26 | 2015-03-26 | A kind of method for controlling traffic signal lights based on computer vision |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104778846B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110610608A (en) * | 2019-08-20 | 2019-12-24 | 江苏金晓电子信息股份有限公司 | Traffic state identification method based on binocular camera |
Families Citing this family (17)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105118308B (en) * | 2015-10-12 | 2017-03-15 | 青岛大学 | Urban road intersection traffic signal optimization method based on cluster intensified learning |
CN105243857A (en) * | 2015-10-30 | 2016-01-13 | 桂林市腾瑞电子科技有限公司 | Intelligent traffic light control system |
CN105894831B (en) * | 2016-05-12 | 2020-11-03 | 郑永春 | Intelligent traffic control device |
CN106297328A (en) * | 2016-08-25 | 2017-01-04 | 深圳市元征科技股份有限公司 | A kind of method for controlling traffic signal lights and device |
KR102652023B1 (en) * | 2016-10-28 | 2024-03-26 | 삼성에스디에스 주식회사 | Method and apparatus for real time traffic information provision |
CN106683448B (en) * | 2017-01-25 | 2020-04-28 | 东南大学 | Road signal lamp control system based on CPS and method thereof |
CN108932857B (en) * | 2017-05-27 | 2021-07-27 | 西门子(中国)有限公司 | Method and device for controlling traffic signal lamp |
CN107170255A (en) * | 2017-06-19 | 2017-09-15 | 上海斐讯数据通信技术有限公司 | The intelligent control method and system of a kind of crosswalk traffic-light |
CN108320538A (en) * | 2018-04-17 | 2018-07-24 | 济南浪潮高新科技投资发展有限公司 | A kind of intelligent traffic signal lamp control system and method based on mist calculate node |
CN109035808A (en) * | 2018-07-20 | 2018-12-18 | 上海斐讯数据通信技术有限公司 | A kind of traffic lights switching method and system based on deep learning |
CN110873571B (en) * | 2018-08-30 | 2021-09-07 | 北京嘀嘀无限科技发展有限公司 | Intersection stop line data acquisition method, system and device |
CN109147331B (en) * | 2018-10-11 | 2021-07-27 | 青岛大学 | Road congestion state detection method based on computer vision |
CN109919883B (en) * | 2018-12-03 | 2021-06-22 | 南京三宝科技股份有限公司 | Traffic video data acquisition method based on gray level conversion |
CN110335477A (en) * | 2019-08-27 | 2019-10-15 | 广东工业大学 | A kind of traffic lamp control method and device |
CN110796864B (en) * | 2019-11-06 | 2022-05-17 | 阿波罗智联(北京)科技有限公司 | Intelligent traffic control method, device, electronic equipment and storage medium |
CN111681430B (en) * | 2020-04-30 | 2022-03-29 | 安徽科力信息产业有限责任公司 | Method for predicting number of stop lines of signal lamp intersection in future in real time |
CN111951576A (en) * | 2020-08-17 | 2020-11-17 | 国为(南京)软件科技有限公司 | Traffic light control system based on vehicle identification and method thereof |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6760061B1 (en) * | 1997-04-14 | 2004-07-06 | Nestor Traffic Systems, Inc. | Traffic sensor |
US20050206728A1 (en) * | 2002-04-15 | 2005-09-22 | Janssen Theodorus M | Method and device for controlling a red light camera |
CN1920897A (en) * | 2006-09-15 | 2007-02-28 | 中控科技集团有限公司 | Signal controlling machine and signal controlling system |
CN201600784U (en) * | 2009-10-13 | 2010-10-06 | 山东农业大学 | Intelligent traffic control device |
CN102360531A (en) * | 2011-09-30 | 2012-02-22 | 哈尔滨工业大学 | Intelligent traffic light control method and system based on wireless sensor network |
CN202352081U (en) * | 2011-09-30 | 2012-07-25 | 哈尔滨工业大学 | Intelligent traffic light control system based on wireless sensor network |
CN203931194U (en) * | 2014-06-19 | 2014-11-05 | 杨霞 | A kind of many arrays are the intelligent traffic lamp device of conversion and control number of seconds in real time |
CN104464319A (en) * | 2014-12-12 | 2015-03-25 | 武汉理工大学 | Temporary traffic control method used for environment that part lanes are enclosed |
-
2015
- 2015-03-26 CN CN201510136919.8A patent/CN104778846B/en not_active Expired - Fee Related
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110610608A (en) * | 2019-08-20 | 2019-12-24 | 江苏金晓电子信息股份有限公司 | Traffic state identification method based on binocular camera |
Also Published As
Publication number | Publication date |
---|---|
CN104778846A (en) | 2015-07-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104778846B (en) | A kind of method for controlling traffic signal lights based on computer vision | |
WO2020173226A1 (en) | Spatial-temporal behavior detection method | |
US10037607B2 (en) | Topology determination for non-overlapping camera network | |
CN111104903B (en) | Depth perception traffic scene multi-target detection method and system | |
KR102652023B1 (en) | Method and apparatus for real time traffic information provision | |
CN110111335A (en) | A kind of the urban transportation Scene Semantics dividing method and system of adaptive confrontation study | |
CN107862376A (en) | A kind of human body image action identification method based on double-current neutral net | |
WO2021180030A1 (en) | Behavior recognition method and system, electronic device and computer-readable storage medium | |
CN112836713A (en) | Image anchor-frame-free detection-based mesoscale convection system identification and tracking method | |
CN110717387A (en) | Real-time vehicle detection method based on unmanned aerial vehicle platform | |
CN112232351B (en) | License plate recognition system based on deep neural network | |
CN108769550B (en) | Image significance analysis system and method based on DSP | |
CN101739827A (en) | Vehicle detecting and tracking method and device | |
CN102722889B (en) | A kind of image background acquisition methods and device | |
Zhao et al. | Deep fully convolutional regression networks for single image haze removal | |
CN104978738A (en) | Method of detection of points of interest in digital image | |
US9361690B2 (en) | Video based method and system for automated side-by-side traffic load balancing | |
CN112560701B (en) | Face image extraction method and device and computer storage medium | |
CN111046773A (en) | Method for judging water retention in pavement based on image technology | |
CN111191610A (en) | People flow detection and processing method in video monitoring | |
CN112085767B (en) | Passenger flow statistical method and system based on deep optical flow tracking | |
Wen-juan et al. | Application of vision sensing technology in urban intelligent traffic control system | |
CN115171079B (en) | Vehicle detection method based on night scene | |
Pillai et al. | Fine-Tuned EfficientNetB4 Transfer Learning Model for Weather Classification | |
WO2021259033A1 (en) | Facial recognition method, electronic device, and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
EXSB | Decision made by sipo to initiate substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20160928 |