CN108681717A - City-level traffic video detection equipment quality detection method - Google Patents
City-level traffic video detection equipment quality detection method Download PDFInfo
- Publication number
- CN108681717A CN108681717A CN201810481913.8A CN201810481913A CN108681717A CN 108681717 A CN108681717 A CN 108681717A CN 201810481913 A CN201810481913 A CN 201810481913A CN 108681717 A CN108681717 A CN 108681717A
- Authority
- CN
- China
- Prior art keywords
- data
- section
- car plate
- time
- detecting device
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- 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
- 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
- G06V20/625—License plates
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses City-level traffic video detection equipment quality detection methods.Include the following steps：A. the network topology structure for building each video detecting device in road network, obtains network topology structure table；B. the stochastic model of link travel confidence time is constructed according to network topology structure table；C. correct car data and wrong data excessively in the equipment detection total amount N that video detecting device S is detected are identified, the fallout ratio of video detecting device S is calculated；All video detecting devices are traversed, the fallout ratio of all video detecting devices is obtained；D. stochastic simulation is carried out to missing data according to the link travel confidence time, defines the stability indicator that vehicle amount is crossed in section, calculates the omission factor of video detecting device in network topology structure table.The present invention has the characteristics that promote traffic data quality, reduce workload and promote data to describe accuracy, is the basis of intelligent transportation platform construction.
Description
Technical field
The present invention relates to data quality monitoring technical fields, especially City-level traffic video detection equipment quality detection side
Method.
Background technology
Most domestic city all experienced construction of high-tech traffic system in several years, be mounted with a large amount of traffic data collection equipment,
Have collected huge traffic data.But the difficult weight when vehicle supervision department wants to alleviate urban traffic blocking using big data technology
Weight, the main reason is that during equipment assembles, due to lens quality, packaging technology etc., accuracy rate will slightly under
Drop after putting into use on the spot, is interfered, acquisition by factors such as light, equipment setting angle, vehicle density, car plate hangs
To data unavoidably there is noise.Data set with a large amount of mistakes can not support effective utilization of wisdom traffic system.
And be monitored at present only for device powers down, the indexs such as the error rate of equipment and omission factor were not detected.
Invention content
The object of the present invention is to provide a kind of City-level traffic video detection equipment quality detection methods.The present invention has
Have and promote traffic data quality, reduce workload and promote the characteristics of data describe accuracy, is intelligent transportation platform construction
Basis.
Technical scheme of the present invention：City-level traffic video detection equipment quality detection method, includes the following steps：
A. the network topology structure for building each video detecting device in road network, obtains network topology structure table；
B. according to the upstream-downstream relationship of network topology structure table, the stochastic model of link travel confidence time is constructed；
C. car plate coding rule, car plate are met by the time of video detecting device S and its by upstream section with car plate
Time within the link travel confidence time of stochastic model, car plate occur have regularity be condition construct decision tree, profit
Correct car data and wrong data excessively in the equipment detection total amount N that video detecting device S is detected are identified with decision tree, from
And calculate the fallout ratio of video detecting device S；Other video detecting devices in traverses network topological structure table, obtain network
The fallout ratio of all video detecting devices in topological structure table；
D. stochastic simulation is carried out to missing data according to the link travel confidence time, the stability that vehicle amount is crossed in definition section refers to
Mark, calculates the omission factor of video detecting device in network topology structure table.
In step a described in City-level traffic video detection equipment quality detection method above-mentioned, the structure city
The network topology structure of grade video detecting device, obtains network topology structure table, includes the following steps：
a1:The initial data of all video detecting devices in road network is acquired, same car plate is not by extraction preset time period
With the time that video detecting device detects, vehicle driving trace is built by the chronological order that the car plate is identified to, it will
Vehicle driving trace is changed into device network topological structure table；
a2:In the network topology structure table of a1, according to the data on flows that the video detecting device of upstream and downstream detects, structure
Make statistic
Wherein：NSVehicle amount, N are crossed for preset time period inner section crossingAVehicle amount, N are crossed for upstream sectionODFor NSAnd NATwo
Section car plate is matched to cross vehicle amount；
a3:The statistic constructed to every section in network topology structure table is classified, different types of section case
Type figure carries out outlier detection, cleans topological relation unreasonable in network topology structure table；
a4:Comparison and undertakes the construction of the location data that unit provides at network topology structure table, carried out on map point demarcation and
Live artificial comparison carries out network topology structure table optimization in conjunction with the spatial relation of video detecting device.
In City-level traffic video detection equipment quality detection method above-mentioned, the link travel confidence time determines
Justice is as follows：
Definition：If θ is the parameter of total travel time, parameter space Θ, traveled distance time t1,t2,t3,...,
tnFor from overall sample.To given α (0<α<1), it is assumed that there are two statistic θL=θL(t1,t2,...,tn) and θR=θR
(t1,t2,...,tn), if to arbitrary θ ∈ Θ, have
P{θL≤θ≤θR}≥1-α (2)
Then claim section [θL,θR] be θ confidence level be 1- α the stroke confidence time；Wherein：θLFor confidence lower limit, θRFor
Confidence upper limit；
The upstream-downstream relationship according to network topology structure table described in step b constructs link travel confidence time randomness
Model is to collected traveled distance time θ, sample data t1,t2,t3,...,tn, its distribution function be F (θ)=∑ P
(ti), frequency interval τ, using steepest descent method, adaptive algorithm construct link travel confidence time stochastic model, specifically
Construction includes the following steps:
b1:It is α to take level of significance α initial value0, i.e. (the 1- α of confidence level 1000) %, time to chopping S initial values S0；
b2:To parameter alpha, S takes P0=max P (θ), P0Corresponding journey time initial value is
b3:Define difference operator
Steepest descent method Iteration is：
Here
Stopping criterion for iteration
If (7) formula meets, time to chopping S, the confidence time under confidence level 100 (1- α) % are exportedHereIf S=S0, note
b4：Time to chopping auto-adaptive parameter is v, S=S-v, repeats b2-b3, judges whether to restrain, allowable error limit
For arbitrarily small positive real number ε, if
Then execute b7；Otherwise S=S-v repeats b2-b3, untilIf being still unsatisfactory for the condition of convergence (8), execute
b5；
b5:S=S0+ v repeats b2-b3, judges whether to restrain, if
Then execute b7；Otherwise S=S+v repeats b2-b3, until S is maximum time to chopping 2 hours, if being still unsatisfactory for restraining
Condition (9), output result are " not restraining ", execute b6；
b6:Confidence level auto-adaptive parameter is ω, and α=α-ω execute b2-b5 if α >=0.8 1-, otherwise execute b7；
b7：Delivery stroke confidence time final result
In step c described in City-level traffic video detection equipment quality detection method above-mentioned, identified using decision tree
Go out correct car data and wrong data excessively in the equipment detection total amount N that video detecting device S is detected, it is specific as follows to state step
Suddenly：
C1. the car plate in equipment detection total amount N in detection cycle is traversed, identification meets the car data excessively of car plate coding rule
M；
C2. with car plate by the time of video detecting device S and its by time in upstream section in stochastic model
In the link travel confidence time, car plate occurs having regularity is condition, identifies right-on car plate M in M1。
The specific method of step c2 described in City-level traffic video detection equipment quality detection method above-mentioned is：
Table R and table D are established in video detecting device S；Table R correctly crosses car data for storing, and table D is waited for for storing
What is further judged crosses car data；As unit of day, car plate and upstream and downstream progress that traversal video detecting device S is detected
Match, the car plate that will match to crosses the car plate data that the vehicle time appeared in the link travel confidence time and is judged as that identification is correct, envelope
Block data is filled, is labeled as 1, is stored in table R；Other car plate data markers are 0, in deposit table D；
Car plate data in table D are further judged, specifically according to the trip rule of car plate, selects and provides in table D
Regular car plate data；Described there is the judgment method of regularity to be：The time that a certain car plate occurs is write down, in history
It is traversed in data, if one hour car plate occurred 3 times or more before and after same time period, judges that the car plate has rule
Property；
According to car plate the data update table D and table R with regularity；Specific update is to reject to have from table D
The car plate data of regularity have described in the car plate data deposit table R of regularity, and the car plate data in table R are as complete
The car plate M of total correctness1。
In step c described in City-level traffic video detection equipment quality detection method above-mentioned, the video detection
The fallout ratio of equipment S is according in table D after the car data F excessively and update for not meeting car plate coding rule in equipment detection total amount N
Car plate data calculated, according to the law of large numbers, the fallout ratio of video detecting device S is obtained by following formula：
In step d described in City-level traffic video detection equipment quality detection method above-mentioned, the network topology
The omission factor of video detecting device, calculates in the steps below in structure table：
D1. data cleansing；The video detecting device in topological structure is traversed, by every video detecting device and video detection
One month data of section where equipment daily counted vehicle amount, and the point analysis that peels off is carried out to every video detecting device, rejected
Outlier；
D2. stable relative error is calculated：If section is B, the detection device at section B is Bi, upstream A, because
Section crosses vehicle amount and swims across vehicle amount thereon with regularity, then the relative error that can define section B is
Detection device is BiRelative error be
Above-mentioned N (A) crosses vehicle total amount for upstream A's, and N (B) crosses vehicle total amount, N (B for section B'si) be section B at inspection
Measurement equipment BiCross vehicle total amount；
Car data was crossed according to one month, calculates the relative error and detection device B thereon of each section BiIt is opposite accidentally
Difference；According to the regularity and large sample theory in section, stable relative error is obtained, is denoted asAnd
In formula,N*(A)、N*(B)、N*(Bi) it is respectively N (A), N (B) εB、Mean value；
D3. judge section B missing inspection situations, calculate N*(B) confidence interval：According toVehicle total amount should be had by calculating section
A confidence intervalWhereinδ
For N (Bi) standard deviation；
WhenWhen, it is that upstream missing inspection is not dealt with, output section omission factor is 0；
WhenWhen, judge section missing inspection；
D4. judge detection device BiMissing inspection situation calculates N*(Bi)：Detection at the section B of missing inspection, which is set, to be judged to step d3
Standby Bi, by the relative error of the calculated stabilizations of step d2Detection device B is calculated lateriThere should be vehicle total amountOne
A confidence intervalWhereinδ is N
(Bi) standard deviation；
WhenWhen, it is that upstream missing inspection is not dealt with, output detection device BiOmission factor is 0；
WhenWhen, judge detection device BiMissing inspection, missing inspection quantity Nlou(Bi) be
Omission factor is
D5. update optimization：To there are the detection device B of missing inspectioniCompletion calculates, that is, traverses all detection device Bi, by step
Rapid d1~d4 recalculates omission factor, detection device BiThere are missing inspections for upstream, are counted again after the quantity polishing of missing inspection is updated
Calculate omission factor.
In step d1 described in City-level traffic video detection equipment quality detection method above-mentioned, outlier is being rejected
Afterwards, if the sample total for crossing vehicle amount is inadequate, the similitude clustering in section is carried out according to the link travel confidence time in road network,
Vehicle quantity was simulated according to the similar section of cluster result, obtained vehicle amount, when simulation excludes the outlier of section and upstream.
Advantageous effect：Compared with prior art, the invention has the advantages that：
1. the present invention constructs the network topology of video detecting device：Video detecting device network topology structure is for describing
Structural relation in road network between each video detecting device, if which video detecting device is in same section, which equipment is deposited
Physical distance etc. between upstream-downstream relationship, equipment.In order to without loss of generality, embody the practicability of model, need to be based on regarding
Frequency detection device network topology structure goes to analyze every a pair that there are the stroke confidence times between the section of upstream-downstream relationship, still
The vehicle supervision department in many cities is difficult the network topology structure table for the video detecting device for providing an entirely accurate, mainly
Reason has three：First, build and complete since the traffic video detection system in many cities is all divided into several phases, may be each issue hold
Unit of founding a capital is different, and Equipment Foundations information does not summarize；Second is that since video detecting device itself is fitted without GPS module, lean on
The data that GPS gathers equipment obtains and actual installation position deviation are very big；Third, due to road network structure complexity, equipment packing density
Not enough, between equipment and equipment and it is not all closed area, the upstream-downstream relationship of equipment is difficult to differentiate completely.Therefore, it is necessary to profits
The network topology structure table of video detecting device, the structural relation between apparent equipment are established with real data.Further, since
Often there is equipment newly-built and move situations such as changing and occur, video detecting device network topology structure table needs periodically automatically update.
2. the present invention proposes the concept of link travel confidence time, and establish calculate the link travel confidence time with
Machine model.The link travel confidence time refers to that vehicle is existed under Parking situation by two detection sectional plane journey times when not long
Confidence interval under given confidence level, it is similar to Travel Time Reliability, whether it can be completed on time describing traveler go out
Capable possibility.With Travel Time Reliability the difference is that, the stroke confidence time had both provided on the section of reliable journey time
Limit, and lower limit is provided, the probability that traveler arrives within the given time can be described more accurately.Stroke confidence
Time not only has researching value in terms of link travel time distribution research and traffic congestion identification, in fake-licensed car analysis, vehicle
Also there is important research value in terms of board recognition accuracy analysis and the analysis of road network vehicle dissipation rate.
3. the present invention gets up video detecting device single in road network by building network topology structure organic connections, then
In conjunction with the method for mathematical modeling, the fallout ratio and omission factor of each detection device are successfully described, and by wrong data and correctly
Data are isolated, and greatly improve traffic data quality.
4. the present invention will identify that the data of correct data and suspection are encapsulated as block number respectively when equipment fallout ratio calculates
According to；Also, historical data is packaged by the present invention when finding the regularity of car plate；By the above method, the present invention passes through
" block number evidence " modeling method, by dispersion, segmentation, fragmentation with the relevant information data of vehicle, be packaged into car plate
Number it is " the block number evidence " of major key, and imparts " growth " to " block number evidence ", it can one continually changing master of accurate description
Body, and as the time changes, the accuracy of description will be higher and higher；Meanwhile in statistical work, largely weight can be avoided
It returns to work work, shortens the response time.
5. the present invention constructs the link travel confidence time using the historical data of video detecting device, using steepest descent method
The link travel confidence time is found with adaptive algorithm, using the link travel confidence time as the detection device quality of data and section
The important evidence of similitude is precisely separating out most of video detecting device detection correctly number by the link travel confidence time
According to recycling the regularity of driving to construct decision tree Accurate classification and go out correct and fault monitoring data, regarded to calculate
The fallout ratio of frequency detection device.The omission factor of video detecting device then defines the stability indicator of section vehicle flowrate, according to steady
Qualitative index developing algorithm analyzes the omission factor of equipment, is provided more accurately for the structure of entire Traffic Analysis platform
Data support.
It is periodically automatically updated 6. the network topology structure that the present invention is built can be realized, the link travel confidence time of foundation
Stochastic model can periodically automatically update in use, by vehicle traveling with regularity car plate, encapsulate block data
The differentiation for carrying out fallout ratio realizes the self study of machine by the above method, and then realizes attribute field in " block number evidence "
Automatically updating data.
Description of the drawings
Fig. 1 is that car data is crossed in the part cleaned in the embodiment of the present invention；
Fig. 2 is calibration position of the video detecting device on map in the embodiment of the present invention；
Fig. 3 is the definition schematic diagram of link travel confidence time；
Fig. 4 is the channelized picture that section Golconda road is tested in the embodiment of the present invention；
Fig. 5 is Golconda North Road and rising sun East Road intersection to Golconda North Road and Yan'an East Road intersection frequency in the embodiment of the present invention
Number histogram；
Fig. 6 be Chinese Road and provincial government road intersection to fountain (left figure) in the embodiment of the present invention, BeiJing Xi Road and in
The frequency histogram in five sections Li Chong (right figure) of dam crossing to the South Roads Jia Xiu；
Fig. 7 is five sections Li Chong of the South Roads Zhong Jiaxiu of the embodiment of the present invention to BeiJing Xi Road and the crossings Zhong Ba (left figure), Golconda
The frequency histogram in North Road and Yan'an East Road intersection to Golconda North Road Xin Yin factories (right figure)；
Fig. 8 is Golconda North Road and rising sun East Road intersection to Golconda North Road and Yan'an East Road intersection one in the embodiment of the present invention
Its traveled distance time change situation；
Fig. 9 is that sand rushes the intersections Lu Yuxing Guan Lu to the one day reality in Zun Yi road and high official position road intersection in the embodiment of the present invention
Journey time situation of change；
Figure 10 is the North Roads Zhong Shachong of the embodiment of the present invention and rushes the intersections Lu Yuxing Guan Lu one day to sand up to high bridge intersection
Traveled distance time change situation；
Figure 11 is Golconda South Road and the streets You Zha intersection to the streets You Zha and youth road intersection one day in the embodiment of the present invention
Traveled distance time change situation；
Figure 12 is the decision tree that video detecting device fallout ratio algorithm is built in the embodiment of the present invention.
Specific implementation mode
The present invention is further illustrated with reference to the accompanying drawings and examples, but be not intended as to the present invention limit according to
According to.
Embodiment 1.A kind of City-level traffic video detection equipment quality detection method, includes the following steps：
A. the network topology structure for building each video detecting device in road network, obtains network topology structure table；
B. according to the upstream-downstream relationship of network topology structure table, the stochastic model of link travel confidence time is constructed；
C. car plate coding rule, car plate are met by the time of video detecting device S and its by upstream section with car plate
Time within the link travel confidence time of stochastic model, car plate occur have regularity be condition construct decision tree, profit
Identify that correct car data of crossing (is detection with car plate in the equipment detection total amount N that video detecting device S is detected with decision tree
Object) and wrong data, to calculate the fallout ratio of video detecting device S；Other in traverses network topological structure table regard
Frequency detection device obtains the fallout ratio of all video detecting devices in network topology structure table；When equipment is dispatched from the factory, different manufacturers are known
Other algorithm is had any different, but accuracy rate is all close to 100%.But during equipment assembles, due to originals such as lens quality, packaging technologies
Cause, accuracy rate will slightly decline, about 98%；Put into use on the spot after, by light, equipment setting angle, vehicle density,
The factors such as car plate hang influence, and accuracy rate declines to a great extent；Therefore, there are a part of error detections in equipment detection total amount N
Data and generation part missing data, can calculate inspection as the two indices for judging video detecting device quality
Fallout ratio in measurement equipment and omission factor.Equipment detects total amount N, refers to that a video detecting device S is detected within the unit interval
The total amount of data arrived；Equipment detects total amount N=right amount+false retrieval amount；
D. stochastic simulation is carried out to missing data according to the link travel confidence time, the stability that vehicle amount is crossed in definition section refers to
Mark, calculates the omission factor of video detecting device in network topology structure table.
Based on actual conditions, the video detecting device of electric police, bayonet in the road network in city etc. is all divided into several phases and builds
And if by different manufacturer's construction, there are very big deviations for the equipment GPS data that different vendor provides.The traffic such as government
Administrative department is difficult the device topology for taking out an entirely accurate.In fact, urban road is unlike through street
There is much fork roads without detection device, it is therefore necessary to the topology of equipment is established according to actual flow data in closed section
Structure, the upstream-downstream relationship of apparent equipment room；In addition, equipment exists newly-built and moves the problems such as changing, topological structure need constantly into
Row update.
According to above-mentioned analysis, corresponding algorithm is needed periodically to update the topological relation of equipment room, the more preferable land productivity of ability
With relevance data analysis road network state.
Based on above-mentioned analysis, in abovementioned steps a, the network topology structure of each video detecting device in road network is built, is obtained
Network topology structure table, includes the following steps：
a1：Acquire the initial data of all video detecting devices (all electric polices, bayonet i.e. in road network) in road network, extraction
The time that same car plate is detected by the video detecting device at different electric polices, bayonet (in terms of one month) in preset time period,
Vehicle driving trace is built by the chronological order that the car plate is identified to, vehicle driving trace is changed into device network and is opened up
Flutter structure table；Primary fields record upstream equipment data on flows, upstream device data on flows and road of network topology structure table
The track number of section and direction；The initial data of the present embodiment is to cross car data as data using Guizhou In China province Guiyang road network
Source, as shown in Figure 1, Fig. 1, which is the part cleaned, crosses car data；
a2:In the network topology structure table of a1, according to the data on flows that the video detecting device of upstream and downstream detects, structure
Make statistic
Wherein：NSFor vehicle amount, N are crossed in section crossing (in terms of one month) in preset time periodAVehicle amount, N are crossed for upstream sectionOD
For NSAnd NATwo section car plates are matched to cross vehicle amount；
a3:The statistic constructed to every section in network topology structure table is classified, different types of section case
Type figure carries out outlier detection, cleans topological relation unreasonable in network topology structure table；
a4:Comparison and undertakes the construction of the location data (such as GPS data) that unit provides at network topology structure table, map (such as
Google Maps) on carry out point demarcation (as shown in Figure 2) and scene manually comparison, in conjunction with the spatial position of video detecting device
Relationship carries out network topology structure table optimization；By taking the road network of Guizhou In China province Guiyang as an example, table 1 is Guiyang partial video
The network topology structure table of detection device.
Table 1
The link travel confidence time above-mentioned is defined as follows：
Definition：If θ is the parameter of total travel time, parameter space Θ, traveled distance time t1,t2,t3,...,
tnFor from overall sample.To given α (0<α<1), it is assumed that there are two statistic θL=θL(t1,t2,...,tn) and θR=θR
(t1,t2,...,tn), if to arbitrary θ ∈ Θ, have
P{θL≤θ≤θR}≥1-α (2)
Then claim section [θL,θR] be θ confidence level be 1- α the stroke confidence time；Wherein：θLFor confidence lower limit, θRFor
Confidence upper limit；As shown in Figure 3；
The upstream-downstream relationship according to network topology structure table described in step b constructs link travel confidence time randomness
Model is to collected traveled distance time θ, sample data t1,t2,t3,...,tn, its distribution function be F (θ)=∑ P
(ti), frequency interval τ, using steepest descent method, adaptive algorithm construct link travel confidence time stochastic model, specifically
Construction includes the following steps:
b1:It is α to take level of significance α initial value0, i.e. (the 1- α of confidence level 1000) %, time to chopping S initial values S0；
b2:To parameter alpha, S takes P0=max P (θ), P0Corresponding journey time initial value is
b3:Define difference operator
Steepest descent method Iteration is：
Here
Stopping criterion for iteration
If (7) formula meets, time to chopping S, the confidence time under confidence level 100 (1- α) % are exportedHereIf S=S0, note
b4：Time to chopping auto-adaptive parameter is v, S=S-v, repeats b2-b3, judges whether to restrain, allowable error limit
For arbitrarily small positive real number ε, if
Then execute b7；Otherwise S=S-v repeats b2-b3, untilIf being still unsatisfactory for the condition of convergence (8), execute
b5；
b5:S=S0+ v repeats b2-b3, judges whether to restrain, if
Then execute b7；Otherwise S=S+v repeats b2-b3, until S is maximum time to chopping 2 hours, if being still unsatisfactory for restraining
Condition (9), output result are " not restraining ", execute b6；
b6:Confidence level auto-adaptive parameter is ω, and α=α-ω execute b2-b5 if α >=0.8 1-, otherwise execute b7；
b7：Delivery stroke confidence time final result
Fig. 4 is the channelized picture for wherein testing section Guizhou In China province Guiyang Golconda road；Fig. 5-Fig. 7 is Guizhou In China respectively
Province Guiyang Golconda North Road and rising sun East Road intersection intersect to Golconda North Road and Yan'an East Road intersection, China Road and provincial government road
Mouth arrives five sections Li Chong of fountain, BeiJing Xi Road and the crossings Zhong Ba to the South Roads Jia Xiu, five sections Li Chong of the South Roads Jia Xiu to West Beijing
The dams Lu Yuzhong crossing, Golconda North Road and Yan'an East Road intersection to the frequency histogram in five experiment sections of Golconda North Road Xin Yin factories
Figure.By the analysis of the frequency histogram in above-mentioned experiment section, the stochastic model of link travel confidence time is built for us
Thinking is provided.
Table 2 is Guizhou In China province Guiyang Golconda North Road and rising sun East Road intersection to Golconda North Road and Yan'an East Road intersection
The link travel confidence time；
Table 3 is to be flushed to BeiJing Xi Road and the crossings Zhong Ba and Golconda North Road in the Guizhou In China province Guiyang South Roads Jia Xiu five and prolong
Link travel confidence time of the peace East Road intersection to Golconda North Road Xin Yin factories；
Table 4 is that BeiJing Xi Road and the crossings Zhong Ba and Golconda North Road and Yan'an East Road intersection to treasured are flushed in the South Roads Jia Xiu five
The link travel confidence time of mountain North Road Xin Yin factories
Table 2- tables 4 embody the stroke confidence time to five sections using link travel confidence time stochastic model
Solve convergence process.
Table 2
Table 3
Table 4
According to Dutch national communication policy document《Nota Mobility》The identification standard of standard, journey time threshold value is
1.2 times of average travel time.The link travel confidence time defined accordingly to the present invention has done necessary verification, and table 5 is
The Link Travel Time comparing result that the link travel confidence time defined with the present invention and Nederlands Norm calculate.What Holland provided
Travel Time Reliability is not defined journey time lower limit.And there are some in the data on flows of section in practical road network
What is gone to zero spends the vehicle time, this cannot function as Travel Time Reliability and carries out further data analysis.As a result it shows：Holland
The critical field threshold value of the journey time provided is smaller.The reliable stroke time threshold provided according to Nederlands Norm, in Fig. 8, Fig. 9
Two section confidence levels are 80% and 82%, and two section confidence levels can only achieve 76% and 65%, and this in Figure 10, Figure 11
It wherein also include impossible data set close to zero.It cannot be preferably fitted the traveled distance time, can not accurately be drawn
The netted condition of branch.
Table 5
In aforementioned step c, identified in the equipment detection total amount N that video detecting device S is detected just using decision tree
True car data and wrong data (see Figure 12) excessively, it is specific as follows to state step：
C0. data cleansing removes repetition and uploads data, obtains the equipment detection total amount N of duplicate removal；
C1. the car plate in equipment detection total amount N in detection cycle is traversed, identification meets the car data excessively of car plate coding rule
M；
C2. with car plate by the time of video detecting device S and its by time in upstream section in stochastic model
In the link travel confidence time, car plate occurs having regularity is condition, identifies right-on car plate M in M1。
The specific method of aforementioned step c2 is：Table R and table D are established in video detecting device S；Table R is for storing just
True car data excessively, table D are used to store the car data excessively for waiting for further judging；As unit of day, the S inspections of traversal video detecting device
The car plate and upstream and downstream measured is matched, and the car plate that will match to is spent the vehicle time and appeared in the link travel confidence time
Car plate data are judged as that identification is correct, encapsulate block data, are labeled as 1, are stored in table R；Other car plate data markers are 0, are deposited
Enter in table D；
Car plate data in table D are further judged, specifically according to the trip rule of car plate, selects and provides in table D
Regular car plate data；Described there is the judgment method of regularity to be：The time that a certain car plate occurs is write down, in history
It is traversed and (is such as traversed in preceding 3 months data) in data, if one hour car plate occurs before and after same time period
It crosses 3 times or more, judges that the car plate has regularity；
According to car plate the data update table D and table R with regularity；Specific update is to reject to have from table D
The car plate data of regularity have described in the car plate data deposit table R of regularity, and the car plate data in table R are as complete
The car plate M of total correctness1。
In aforementioned step c, the fallout ratio of the video detecting device S is to be detected in total amount N not meeting according to equipment
Cross car data F and the car plate data in table D after update of car plate coding rule are calculated, according to the law of large numbers, video detection
The fallout ratio of equipment S is obtained by following formula：
In aforementioned step d, the omission factor of video detecting device, is counted in the steps below in the network topology structure table
It calculates：
D1. data cleansing：The video detecting device in topological structure is traversed, by every video detecting device and video detection
One month data of section where equipment daily counted vehicle amount, carried out the point analysis that peels off to every video detecting device, i.e.,：Time
The video detecting device in topological structure is gone through, by one month number of section where every video detecting device and video detecting device
According to vehicle amount was daily counted, the data exception point in one middle of the month of every detection device is rejected, abnormal point generation is possible as
Power-off causes；When equipment is dispatched from the factory, different manufacturers recognizer is had any different, but accuracy rate is all close to 100%.In equipment assembling process
In, due to lens quality, packaging technology etc., accuracy rate will slightly decline, about 98%.After putting into use on the spot, by
It is influenced to factors such as light, equipment setting angle, vehicle density, car plate hangs, accuracy rate declines to a great extent；
D2. stable relative error is calculated：If section is B, the detection device at section B is Bi, upstream A, because
Section crosses vehicle amount and swims across vehicle amount thereon with regularity, then the relative error that can define section B is
Detection device is BiRelative error be
Above-mentioned N (A) crosses vehicle total amount for upstream A's, and N (B) crosses vehicle total amount, N (B for section B'si) be section B at inspection
Measurement equipment BiCross vehicle total amount；
Car data was crossed according to one month, calculates the relative error and detection device B thereon of each section BiIt is opposite accidentally
Difference；According to the regularity and large sample theory in section, stable relative error is obtained, is denoted asAnd
In formula,N*(A)、N*(B)、N*(Bi) it is respectively N (A), N (B) εB、Mean value；
D3. judge section B missing inspection situations, calculate N*(B) confidence interval：According toVehicle total amount should be had by calculating sectionA confidence intervalWherein
δ is N (Bi) standard deviation；
WhenWhen, it is that upstream missing inspection is not dealt with, output section omission factor is 0；
WhenWhen, judge section missing inspection；
D4. judge detection device BiMissing inspection situation calculates N*(Bi)：Detection at the section B of missing inspection, which is set, to be judged to step d3
Standby Bi, by the relative error of the calculated stabilizations of step d2Detection device B is calculated lateriThere should be vehicle total amountOne
A confidence intervalWhereinδ is N
(Bi) standard deviation；
WhenWhen, it is that upstream missing inspection is not dealt with, output detection device BiOmission factor is 0；
WhenWhen, judge detection device BiMissing inspection, missing inspection quantity Nlou(Bi) be
Omission factor is
D5. update optimization：To there are the detection device B of missing inspectioniCompletion calculates, that is, traverses all detection device Bi, by step
Rapid d1~d4 recalculates omission factor, detection device BiThere are missing inspections for upstream, are counted again after the quantity polishing of missing inspection is updated
Calculate omission factor.
In aforementioned step d1, after rejecting outlier, if the sample total for crossing vehicle amount is inadequate, according to section row in road network
The journey confidence time carries out the similitude clustering in section, simulated vehicle quantity according to the similar section of cluster result, obtained vehicle
Amount excludes the outlier of section and upstream when simulation.
Claims (8)
1. City-level traffic video detection equipment quality detection method, which is characterized in that include the following steps：
A. the network topology structure for building each video detecting device in road network, obtains network topology structure table；
B. according to the upstream-downstream relationship of network topology structure table, the stochastic model of link travel confidence time is constructed；
C. with car plate meet car plate coding rule, car plate by time of video detecting device S and its by upstream section when
Between within the link travel confidence time of stochastic model, car plate occur have regularity be condition construct decision tree, using certainly
Plan tree identifies correct car data and wrong data excessively in the equipment detection total amount N that video detecting device S is detected, to count
Calculate the fallout ratio of video detecting device S；Other video detecting devices in traverses network topological structure table, obtain network topology
The fallout ratio of all video detecting devices in structure table；
D. stochastic simulation is carried out to missing data according to the link travel confidence time, defines the stability indicator that vehicle amount is crossed in section,
Calculate the omission factor of video detecting device in network topology structure table.
2. City-level traffic video detection equipment quality detection method according to claim 1, it is characterised in that：
In step a, the network topology structure of the structure City-level video detecting device obtains network topology structure table, wraps
Include following step：
a1:The initial data of all video detecting devices in road network is acquired, same car plate in preset time period is extracted and is regarded by difference
The time that frequency detection device detects builds vehicle driving trace, by vehicle by the chronological order that the car plate is identified to
Driving trace is changed into device network topological structure table；
a2:In the network topology structure table of a1, according to the data on flows that the video detecting device of upstream and downstream detects, construction system
Metering
Wherein：NSVehicle amount, N are crossed for preset time period inner section crossingAVehicle amount, N are crossed for upstream sectionODFor NSAnd NATwo sections
Car plate is matched to cross vehicle amount；
a3:The statistic constructed to every section in network topology structure table is classified, different types of section box figure
Outlier detection is carried out, topological relation unreasonable in network topology structure table is cleaned；
a4:It compares network topology structure table and undertakes the construction of the location data that unit provides, point demarcation and scene are carried out on map
It is artificial to compare, in conjunction with the spatial relation of video detecting device, carry out network topology structure table optimization.
3. City-level traffic video detection equipment quality detection method according to claim 1, it is characterised in that：
The link travel confidence time is defined as follows：
Definition：If θ is the parameter of total travel time, parameter space Θ, traveled distance time t1,t2,t3,...,tnIt is next
From overall sample.To given α (0<α<1), it is assumed that there are two statistic θL=θL(t1,t2,...,tn) and θR=θR(t1,
t2,...,tn), if to arbitrary θ ∈ Θ, have
P{θL≤θ≤θR}≥1-α (2)
Then claim section [θL,θR] be θ confidence level be 1- α the stroke confidence time；Wherein：θLFor confidence lower limit, θRFor confidence
The upper limit；
The upstream-downstream relationship according to network topology structure table described in step b constructs link travel confidence time stochastic model,
It is to collected traveled distance time θ, sample data t1,t2,t3,...,tn, its distribution function be F (θ)=∑ P (ti)、
Frequency interval is τ, and link travel confidence time stochastic model, specific configuration are constructed using steepest descent method, adaptive algorithm
Include the following steps:
b1:It is α to take level of significance α initial value0, i.e. (the 1- α of confidence level 1000) %, time to chopping S initial values S0；
b2:To parameter alpha, S takes P0=max P (θ), P0Corresponding journey time initial value is
b3:Define difference operator
Steepest descent method Iteration is：
Here
Stopping criterion for iteration
If (7) formula meets, time to chopping S, the confidence time under confidence level 100 (1- α) % are exportedHereIf S=S0, note
b4：Time to chopping auto-adaptive parameter is v, S=S-v, repeats b2-b3, judges whether to restrain, and allowable error is limited to appoint
The positive real number ε for anticipating small, if
Then execute b7；Otherwise S=S-v repeats b2-b3, untilIf being still unsatisfactory for the condition of convergence (8), b5 is executed；
b5:S=S0+ v repeats b2-b3, judges whether to restrain, if
Then execute b7；Otherwise S=S+v repeats b2-b3, until S is maximum time to chopping 2 hours, if being still unsatisfactory for the condition of convergence
(9), output result is " not restraining ", executes b6；
b6:Confidence level auto-adaptive parameter is ω, and α=α-ω execute b2-b5 if α >=0.8 1-, otherwise execute b7；
b7：Delivery stroke confidence time final result
4. City-level traffic video detection equipment quality detection method according to claim 1, it is characterised in that：Step c
In, identify correct car data and error number excessively in the equipment detection total amount N that video detecting device S is detected using decision tree
According to specific as follows to state step：
C1. the car plate in equipment detection total amount N in detection cycle is traversed, identification meets the car data M excessively of car plate coding rule；
C2. with car plate by the time of video detecting device S and its by time in upstream section in the section of stochastic model
In the stroke confidence time, car plate occurs having regularity is condition, identifies right-on car plate M in M1。
5. City-level traffic video detection equipment quality detection method according to claim 4, it is characterised in that：
The specific method of step c2 is：
Table R and table D are established in video detecting device S；Table R correctly crosses car data for storing, and table D is waited for for storing into one
What step judged crosses car data；As unit of day, car plate and upstream and downstream that traversal video detecting device S is detected are matched, will
The car plate being matched to crosses the car plate data that the vehicle time appeared in the link travel confidence time and is judged as that identification is correct, and encapsulation is blocking
Data are labeled as 1, are stored in table R；Other car plate data markers are 0, in deposit table D；
Car plate data in table D are further judged, specifically according to the trip rule of car plate, are picked out in table D with rule
The car plate data of rule property；Described there is the judgment method of regularity to be：The time that a certain car plate occurs is write down, in historical data
In traversed, if one hour car plate occurred 3 times or more before and after same time period, judge the car plate have regularity；
According to car plate the data update table D and table R with regularity；Specific update is to reject to have rule from table D
Property car plate data, by the described car plate data deposit table R with regularity, the car plate data in table R are completely just
True car plate M1。
6. City-level traffic video detection equipment quality detection method according to claim 5, it is characterised in that：Step c
In, the fallout ratio of the video detecting device S is that the vehicle excessively that car plate coding rule is not met in total amount N is detected according to equipment
Data F is calculated with the car plate data in table D after update, and according to the law of large numbers, the fallout ratio of video detecting device S is by following
Formula obtains：
7. City-level traffic video detection equipment quality detection method according to claim 1, it is characterised in that：Step d
In, the omission factor of video detecting device, calculates in the steps below in the network topology structure table：
D1. data cleansing；The video detecting device in topological structure is traversed, by every video detecting device and video detecting device
Section one month data in place daily counted vehicle amount, carried out the point analysis that peels off to every video detecting device, rejecting peels off
Point；
D2. stable relative error is calculated：If section is B, the detection device at section B is Bi, upstream A, because of section mistake
Vehicle amount and vehicle amount is swum across thereon with regularity, then the relative error that can define section B is
Detection device is BiRelative error be
Above-mentioned N (A) crosses vehicle total amount for upstream A's, and N (B) crosses vehicle total amount, N (B for section B'si) be section B at detection device
BiCross vehicle total amount；
Car data was crossed according to one month, calculates the relative error and detection device B thereon of each section BiRelative error；Root
According to the regularity and large sample theory in section, stable relative error is obtained, is denoted asAnd
In formula,N*(A)、N*(B)、N*(Bi) it is respectively N (A), N (B) εB、Mean value；D3. judge section B missing inspections
Situation calculates N*(B) confidence interval：According toVehicle total amount should be had by calculating sectionA confidence intervalWhereinδ is N (Bi) standard
Difference；
WhenWhen, it is that upstream missing inspection is not dealt with, output section omission factor is 0；
WhenWhen, judge section missing inspection；
D4. judge detection device BiMissing inspection situation calculates N*(Bi)：Detection device B at the section B of missing inspection is judged step d3i,
By the relative error of the calculated stabilizations of step d2Detection device B is calculated lateriThere should be vehicle total amountOne set
Believe sectionWhereinδ is N (Bi)
Standard deviation；
WhenWhen, it is that upstream missing inspection is not dealt with, output detection device BiOmission factor is 0；
WhenWhen, judge detection device BiMissing inspection, missing inspection quantity Nlou(Bi) be
Omission factor is
D5. update optimization：To there are the detection device B of missing inspectioniCompletion calculates, that is, traverses all detection device Bi, by step d1
~d4 recalculates omission factor, detection device BiThere are missing inspections for upstream, and leakage is recalculated after the quantity polishing of missing inspection is updated
Inspection rate.
8. City-level traffic video detection equipment quality detection method according to claim 7, it is characterised in that：Step d1
In, after rejecting outlier, if the sample total for crossing vehicle amount is inadequate, section is carried out according to the link travel confidence time in road network
Similitude clustering simulated vehicle quantity according to the similar section of cluster result, obtained vehicle amount, and when simulation excludes section and upper
The outlier of trip.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810481913.8A CN108681717B (en) | 2018-05-18 | 2018-05-18 | Quality detection method for urban traffic video detection equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810481913.8A CN108681717B (en) | 2018-05-18 | 2018-05-18 | Quality detection method for urban traffic video detection equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108681717A true CN108681717A (en) | 2018-10-19 |
| CN108681717B CN108681717B (en) | 2021-12-07 |
Family
ID=63805260
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810481913.8A Active CN108681717B (en) | 2018-05-18 | 2018-05-18 | Quality detection method for urban traffic video detection equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108681717B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110807919A (en) * | 2019-10-22 | 2020-02-18 | 浙江大学 | Urban road network traffic operation situation evaluation method based on vehicle passing data |
| CN111223305A (en) * | 2019-12-25 | 2020-06-02 | 华为技术有限公司 | Method and device for detecting illegal card involvement |
| CN112419261A (en) * | 2020-11-19 | 2021-02-26 | 江汉大学 | Visual acquisition method and device with abnormal point removing function |
| CN112419751A (en) * | 2020-10-10 | 2021-02-26 | 同济大学 | Signalized intersection lane queuing length estimation method based on single-section electric alarm data |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100104199A1 (en) * | 2008-04-24 | 2010-04-29 | Gm Global Technology Operations, Inc. | Method for detecting a clear path of travel for a vehicle enhanced by object detection |
| US20140348390A1 (en) * | 2013-05-21 | 2014-11-27 | Peking University Founder Group Co., Ltd. | Method and apparatus for detecting traffic monitoring video |
| US20160042641A1 (en) * | 2013-06-18 | 2016-02-11 | Carnegie Mellon University, A Pennsylvania Non-Profit Corporation | Smart and scalable urban signal networks: methods and systems for adaptive traffic signal control |
| CN105489010A (en) * | 2015-12-29 | 2016-04-13 | 中国城市规划设计研究院 | System and method for monitoring and analyzing fast road travel time reliability |
| CN105513359A (en) * | 2016-01-27 | 2016-04-20 | 北京工业大学 | Method for estimating city expressway traffic states based on mobile detection of smartphones |
| CN106448165A (en) * | 2016-11-02 | 2017-02-22 | 浙江大学 | Road network travel time reliability evaluation method based on online booked car data |
| CN106952474A (en) * | 2017-04-12 | 2017-07-14 | 湖南源信光电科技股份有限公司 | The statistical method of traffic flow detected based on moving vehicle |
| CN107195180A (en) * | 2017-06-08 | 2017-09-22 | 青岛海信网络科技股份有限公司 | A kind of traffic trip track extraction method and device based on the alert data of electricity |
-
2018
- 2018-05-18 CN CN201810481913.8A patent/CN108681717B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100104199A1 (en) * | 2008-04-24 | 2010-04-29 | Gm Global Technology Operations, Inc. | Method for detecting a clear path of travel for a vehicle enhanced by object detection |
| US20140348390A1 (en) * | 2013-05-21 | 2014-11-27 | Peking University Founder Group Co., Ltd. | Method and apparatus for detecting traffic monitoring video |
| US20160042641A1 (en) * | 2013-06-18 | 2016-02-11 | Carnegie Mellon University, A Pennsylvania Non-Profit Corporation | Smart and scalable urban signal networks: methods and systems for adaptive traffic signal control |
| CN105489010A (en) * | 2015-12-29 | 2016-04-13 | 中国城市规划设计研究院 | System and method for monitoring and analyzing fast road travel time reliability |
| CN105513359A (en) * | 2016-01-27 | 2016-04-20 | 北京工业大学 | Method for estimating city expressway traffic states based on mobile detection of smartphones |
| CN106448165A (en) * | 2016-11-02 | 2017-02-22 | 浙江大学 | Road network travel time reliability evaluation method based on online booked car data |
| CN106952474A (en) * | 2017-04-12 | 2017-07-14 | 湖南源信光电科技股份有限公司 | The statistical method of traffic flow detected based on moving vehicle |
| CN107195180A (en) * | 2017-06-08 | 2017-09-22 | 青岛海信网络科技股份有限公司 | A kind of traffic trip track extraction method and device based on the alert data of electricity |
Non-Patent Citations (5)
| Title |
|---|
| CUI QINGHUA ET AL.: "Time-varying confidence interval forecasting of travel time for urban arterials using AＲIMA-GAＲCH model", 《JOURNAL OF SOUTHEAST UNIVERSITY ( ENGLISH EDITION)》 * |
| TRAVIS B. GLICK ET AL.: "Traffic and Transit Travel Time Reliability Indexes and Confidence Intervals", 《TRANSPORTATION RESEARCH RECORD: JOURNAL OF THE TRANSPORTATION RESEARCH BOARD》 * |
| 张勇等: "城市道路网络的行程时间可靠性", 《系统工程理论与实践》 * |
| 李昂等: "城市道路路段行程时间计算模型研究", 《公路工程》 * |
| 杨玲敏等: "交通运输系统工程与信息", 《交通运输系统工程与信息》 * |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110807919A (en) * | 2019-10-22 | 2020-02-18 | 浙江大学 | Urban road network traffic operation situation evaluation method based on vehicle passing data |
| CN111223305A (en) * | 2019-12-25 | 2020-06-02 | 华为技术有限公司 | Method and device for detecting illegal card involvement |
| CN111223305B (en) * | 2019-12-25 | 2022-05-17 | 华为技术有限公司 | Method and device for detecting illegal card involvement |
| CN112419751A (en) * | 2020-10-10 | 2021-02-26 | 同济大学 | Signalized intersection lane queuing length estimation method based on single-section electric alarm data |
| CN112419751B (en) * | 2020-10-10 | 2021-10-08 | 同济大学 | Signalized intersection lane queuing length estimation method based on single-section electric alarm data |
| CN112419261A (en) * | 2020-11-19 | 2021-02-26 | 江汉大学 | Visual acquisition method and device with abnormal point removing function |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108681717B (en) | 2021-12-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108681717A (en) | City-level traffic video detection equipment quality detection method | |
| WO2019047905A1 (en) | Road traffic analysis system, method and apparatus | |
| CN106228808B (en) | City expressway travel time prediction method based on Floating Car space-time grid data | |
| GB2582532A (en) | Method for detecting traffic anomaly of urban road based on peak hour traffic flow rate | |
| CN108010316A (en) | A kind of road traffic multisource data fusion processing method based on road net model | |
| CN105206057B (en) | Detection method and system based on Floating Car resident trip hot spot region | |
| CN112489426B (en) | Urban traffic flow space-time prediction scheme based on graph convolution neural network | |
| CN107134142A (en) | A kind of urban road method for predicting based on multisource data fusion | |
| CN109255288A (en) | A kind of road surface breakage detection method, device and terminal device | |
| CN110362557B (en) | Missing path repairing method based on machine learning and license plate recognition data | |
| US20130090905A1 (en) | Traffic Sensor Management | |
| CN108597227A (en) | Road traffic flow forecasting method under freeway toll station | |
| CN107945510B (en) | Road segment detection method considering traffic demand and road network operation efficiency | |
| CN109241169A (en) | The multi-source heterogeneous data fusion geo-database integration method of power distribution network operation information | |
| CN107180270A (en) | Passenger flow forecasting and system | |
| CN109598930A (en) | One kind detecting overhead closed system automatically | |
| CN106952472A (en) | Congestion level sort method and system between city based on multivariate data | |
| CN105279963B (en) | A kind of method and device for obtaining traffic information | |
| CN108198438A (en) | A kind of heuristic missing road traffic projectional technique based on upstream and downstream intersection vehicle flux | |
| CN107919016A (en) | Traffic flow parameter missing complementing method based on multi-source detector data | |
| CN109063885A (en) | A kind of substation's exception metric data prediction technique | |
| CN110400462A (en) | Track traffic for passenger flow monitoring and pre-alarming method and its system based on fuzzy theory | |
| CN105241464B (en) | A kind of crane rail slash figure fast matching method | |
| CN114005282A (en) | Intelligent city traffic management system and method based on crowd sensing | |
| CN104731970B (en) | The quality testing and control method of the multi-source heterogeneous data of highway |
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 |