CN102768802B - Method for judging road vehicle jam based on finite-state machine (FSM) - Google Patents
Method for judging road vehicle jam based on finite-state machine (FSM) Download PDFInfo
- Publication number
- CN102768802B CN102768802B CN201210240036.8A CN201210240036A CN102768802B CN 102768802 B CN102768802 B CN 102768802B CN 201210240036 A CN201210240036 A CN 201210240036A CN 102768802 B CN102768802 B CN 102768802B
- Authority
- CN
- China
- Prior art keywords
- road
- patch
- wagon flow
- image
- speed
- 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.)
- Active
Links
Landscapes
- Traffic Control Systems (AREA)
- Image Processing (AREA)
Abstract
The invention belongs to the technical field of digital image processing and pattern recognition, and particularly relates to a method for judging a road vehicle jam based on a finite-state machine (FSM). The method comprises the following steps of establishing an FSM model according to GA115-1995 road traffic jam standard and combining with a forming process of an urban road jam, capturing a traffic road image by using a high-definition camera, estimating length and an overall speed of a vehicle queue on a motorway in the image, activating a trigger condition in the FSM according to limitations of the motorway to capacity and speed of vehicles, and judging whether the current road condition is in a jam state or not. Compared with traditional method, the method disclosed by the invention comprehensively considers the forming process of the urban road traffic jam, can dynamically detect the generation and the dissipation of the traffic jam in real time, and supplies an effective basis for traffic management to relevant departments.
Description
Technical field
The invention belongs to Digital Image Processing and mode identification technology, be specifically related to road vehicle in the intelligent traffic monitoring method of discrimination that blocks up.
Background technology
In recent years the rapid fluid resuscitation of China's economy increases, and country obviously increases in the input of transportation industry.A large amount of inputs of road infrastructure, riseing rapidly of motor vehicle number, cause road vehicle to block up, frequent accidents occurs, the problems such as traffic environment deterioration are day by day serious, but the construction of corresponding traffic law spread of education and intelligent transportation manage system relatively lags behind, for traffic administration proposes stern challenge.The effect of traffic administration is that the vehicle on road and pedestrian are reasonably guided, organized and limit, and makes that traffic is safe as far as possible, unimpeded, public hazards are little, less energy consumption.Wherein, effectively to dredge be the important duty of traffic administration for the monitoring of road traffic state and abnormal traffic flow.
GA 115-1995 road traffic block standard definition block on road because vehicle is excessively intensive, the reason such as traffic hazard, engineering construction, act of violating regulations and nature, and the vehicle time delay causing mistake increases and the state of queue length lengthening.Wherein vehicle queue refers on the driveway in road junction or section, and the random vehicle arriving is by process, and the wait occurring after traffic flow is obstructed is by the ranks of vehicle.Although traffic behavior happens suddenly, but can detect identification.If can detect in time and alarm, automatic synchronization is processed scheduling, just can reduce loss, improves to greatest extent the utilization factor of road.Traffic behavior detects the important step as traffic monitoring and traffic events detection, is the important component part of intelligent transportation system (ITS), for traffic administration provides more rapidly and service accurately.Therefore, the automatic detection of traffic behavior is very important in intelligent transportation research.
Current most of traffic data mainly detects acquisition by the magnetic induction loop being embedded under track.This method exists obviously not enough: strong to external system dependence; Construction maintenance is convenient not; System flexibility is poor.Along with developing rapidly of multimedia and computer technology, directly utilizing video image characteristic detection technique to detect traffic route state and warn becomes possibility.And video detection technology is compared and is had many outstanding advantages with other sensor: 1) can extract high-quality vehicle and traffic scene information; 2) can detect larger scene area, reduce the installation number of sensor; 3) video sensor is easy to installment and debugging, and road pavement and building facility can not produce destruction, and the expense of operation and maintenance is relatively low.
Summary of the invention
The object of the invention is to propose a kind of scene face is large, efficiency is high road vehicle method of discrimination that blocks up that adapts to.
The road vehicle that the present invention the proposes method of discrimination that blocks up, adopts a kind of finite state machine model.Concrete is exactly to control according to GA 115-1995 road traffic obstruction standard and city road signal lamp the forming process that intersection blocks up, and sets up finite state machine model (Fig. 2).Finite state machine model comprises three kinds of possible states: unimpeded, walk or drive slowly, block up; Between state, change analysis result with traffic state into condition, analyze according to being wagon flow speed and wagon flow queue length; And the traffic roads image flow analysis that wagon flow speed and wagon flow queue length are taken by high definition camera estimation obtains; Wherein:
The process of described wagon flow queue length estimation is:
(1) image color conversion: by image from the color space conversion of RGB to gray space;
(2) edge extracting: under gray level image, road area is carried out to Edge extraction;
(3) edge strength projection: the tail position that calculates fleet according to the edge strength of image-region;
Described wagon flow speed is utilized adjacent two two field pictures, by the method for template matches, estimates, its computation process is:
(1) a plurality of reference blocks (patch) of road area are set, record the image in previous frame reference block;
(2) template matches, utilizes template matches search patch in the position of present frame;
(3) regional movement calculates, and by the motion-vector of patch in statistical picture, in order to estimation area, moves.
Specifically be described below respectively:
One, set up finite state machine model
Finite state machine model (as Fig. 2), each state represents with circle, comprises three kinds of possible states, be respectively unimpeded, walk or drive slowly and block up, be specially:
(1) unimpeded, represent that road vehicle is sparse, travel speed is not affected by front vehicles;
(2) jogging, represents that roadway occupancy is higher, and Vehicle Speed is subject to impact to a certain degree;
(3) block up, represent that roadway occupancy is high, the travel speed of vehicle is had a strong impact on.
In the state conversion of finite state machine, according to wagon flow length (being vehicle queue length) and wagon flow speed (the vehicle gait of march of queuing up), defined state transition function.If detected, current wagon flow length reaches the upper limit and estimated speed is 0, judges that whole track is in congestion status; If current wagon flow length reaches the upper limit, speed is not 0, and wagon flow is in jogging state; If do not arrive limit and detect wagon flow length, judge that now track is in unimpeded state.
Two, estimate wagon flow length
Obtain the road image that high-definition camera is taken, the image photographing due to current most of CCD camera is rgb space, needs to be first transformed into gray space here and processes.Based on vehicle self, there is the feature of enriching very much textural characteristics, can extract Image edge gradient to gray level image.By image space edge feature statistical means, for single-frame images, analyze motor vehicle queue length on road.Particular content is:
1, by the road image of being taken by high-definition camera, certain region
coloured image be transformed into gray space.Generally CCD camera output image is the coloured image with R, G, tri-passages of B.So the formula of conversion can be expressed as:
Wherein
,
,
represent respectively the channel strength value of corresponding each pixel,
intensity level corresponding to gray space after conversion.
2, gray level image is extracted to edge strength.Define a kind of
boundary operator:
Wherein
,
respectively to calculate
,
direction gradient operator.Definition image is
, wherein
,
be pixel coordinate, the gradient of all directions can be expressed as:
Gradient magnitude can be expressed as:
3, by the edge size of image to being parallel to the cumulative projection of way direction.As way direction in image perpendicular to
axle, can utilize formula (6) vertically to add up:
Wherein width represents the width of road.
4, obtain the length of fleet on road.Set an edge strength threshold value
, solve:
Choose and separate concentrated minimum value:
be exactly the afterbody ordinate of current road fleet, calculate wagon flow measure initial ordinate with
poor, can obtain wagon flow length.
Three, estimate wagon flow speed
Here utilize the variation of consecutive frame image pixel to carry out estimation region motion feature.First define some reference blocks (patch), the area image pixel that in the previous frame that prestores, patch demarcates.Utilize image template matching technology, in next frame image, search for the position of patch.The movement of patch utilizes the motion feature in region to represent, the size of movement velocity represents the travel speed of this region vehicle, and direction of motion represents the moving direction of this region vehicle.Be specially:
1, the position of reference block (patch) is set.At a large amount of patch of track region division, for preserving the area image pixel of previous frame patch demarcation, (generally can adopt
image block as patch).Because granularity is crossed conference and brought background pixel into, cause patch in consecutive frame to be lack of consistency; And the too small meeting of granularity is easily subject to the interference of noise, affect accuracy.
2, the movement that utilizes the mode of template matches to calculate patch.To each patch, can set the maximum magnitude that it moves, be referred to as to search for forms
, adopt SSD formula:
Wherein
,
,
,
the height, the search forms that represent respectively wide, the patch of patch
image pixel and the pixel of patch.The time complexity of Direct calculation formulas (9) is
.In order to reduce the complicacy of calculating, can use convolution theorem can be converted into discrete Fourier transformation, utilize Fast Fourier Transform (FFT) (FFT) by reduced complexity to be
.
3, add up the integrated moving of a series of patch.Because movement and the road direction of vehicle integral body has consistance.So can assess the goodness of fit by computer memory cosine:
Wherein
,
represent respectively the direction of road and the direction that patch moves.Allow Vehicle Driving Cycle and road direction to have certain deviation, an angle threshold is set, as
, filter some noisy patch.
4, calculate wagon flow translational speed.The patch staying after statistics filtered noise, the mean value of calculating mobile vector:
Here
represent region integrated moving vector,
for the quantity of effective patch,
the motion-vector that represents patch.Finally, frame per second
represent wagon flow translational speed.
Principal feature of the present invention has:
(1) the present invention differentiates the process that flow process meets the formation of blocking up, and can detect exactly road condition;
(2) only dependency graph, as video sensor, can carry out equipment layout quickly and easily;
(3) this method has only been utilized texture information and the pixel value difference information of image, and outdoor strong illumination variation is had compared with strong robustness.
Accompanying drawing explanation
Fig. 1 is process flow diagram of the present invention.
Fig. 2 is finite state machine model figure.
Embodiment
As shown in Figure 1, the vedio data of acquisition camera monitoring and controlling traffic road, analyzes by image processing algorithm, and concrete steps are as follows:
1, utilize formula (1) that coloured image is transformed into gray space;
2, by the boundary operator of formula (2) definition, in the lower gradient magnitude of calculating gray level image of formula (5);
3, adopt formula (6) to carry out the cumulative projection of image border, the minimum value of asking for formula (7) under setting threshold just can calculate the afterbody ordinate of current road fleet.Here, threshold value is relevant with the width of track imaging, is generally set as
;
4,, again according to formula (9), calculate
minimum value, obtain the mobile vector of each piece (patch);
5, through trend pass filtering, process (10), retain Vehicle Driving Cycle consistent patch relative to road direction, angle threshold can be set as
;
6, recycling formula (11) calculates patch mobile vector average, estimates wagon flow translational speed, is expressed as frame per second
mobile vector average;
7, current fleet length result of calculation and wagon flow velocity estimation value are imported in finite state machine (Fig. 2), finally obtained current road condition.
Claims (1)
1. the congestion in road method of discrimination based on finite state machine, is characterized in that concrete steps are:
step 1, set up finite state machine model
Finite state machine model comprises three kinds of possible states, be respectively unimpeded, walk or drive slowly and block up, be specially:
(1) unimpeded, represent that road vehicle is sparse, travel speed is not affected by front vehicles;
(2) jogging, represents that roadway occupancy is higher, and Vehicle Speed is subject to impact to a certain degree;
(3) block up, represent that roadway occupancy is high, the travel speed of vehicle is had a strong impact on;
In the state conversion of finite state machine, according to wagon flow length and wagon flow speed definition state transition function; If detected, current wagon flow length reaches the upper limit and estimated speed is 0, judges that whole track is in congestion status; If current wagon flow length reaches the upper limit, speed is not 0, and wagon flow is in jogging state; If do not arrive limit and detect wagon flow length, judge that now track is in unimpeded state;
step 2, estimate wagon flow length
(1) in the road image of being taken by high-definition camera, by certain region
coloured image be transformed into gray space; The formula of conversion is:
Wherein,
,
,
represent respectively the channel strength value of corresponding each pixel,
intensity level corresponding to gray space after conversion;
(2) gray level image is extracted to edge strength,
Define a kind of
boundary operator:
（2）
Wherein,
,
respectively to calculate
,
direction gradient operator; Definition image is
, wherein
,
be pixel coordinate, the gradient table of all directions is shown:
（3）
（4）
Gradient magnitude is expressed as:
（5）
(3) by the edge size of image to being parallel to the cumulative projection of way direction, establish road direction perpendicular to
axle, utilizes formula (6) vertically to add up:
（6）
Wherein width represents the width of road;
(4) obtain the length of wagon flow on road,
Set a threshold value
, solve:
（7）
Choose and separate concentrated minimum value:
（8）
be exactly the afterbody ordinate of current road wagon flow, calculate wagon flow measure initial ordinate with
poor, can obtain wagon flow length;
step 3, estimate wagon flow speed
Here utilize the variation of consecutive frame image pixel to carry out estimation region motion feature, first define some reference block patch, the area image pixel that in the previous frame that prestores, patch demarcates; Utilize image template matching technology, in next frame image, search for the position of patch; The movement of patch utilizes the motion feature in region to represent, the size of movement velocity represents the travel speed of this region vehicle, and direction of motion represents the moving direction of this region vehicle; Concrete steps are:
(1), the position of reference block is set, reference block is designated as patch, at a large amount of patch of track region division, for preserving the image in previous frame patch region;
(2), the movement that utilizes the mode of template matches to calculate patch, each patch is set to the maximum magnitude that it moves, be referred to as to search for forms
, adopt SSD formula:
（9）
Wherein,
,
,
,
the height, the search forms that represent respectively wide, the patch of patch
image pixel and the pixel of patch;
(3), add up the integrated moving of a series of patch, the computer memory cosine assessment goodness of fit:
（10）
Wherein
,
represent respectively the direction of road and the direction that patch moves, filter out noisy patch;
(4), calculate wagon flow translational speed, the mean value that calculating patch moves:
（11）
Here,
represent region integrated moving vector,
quantity for effective patch;
the motion-vector that represents patch; Wagon flow translational speed N frame per second
represent;
step 4, differentiate current road condition;
Due to frame per second
represent wagon flow translational speed, therefore utilize the road wagon flow length of previous calculations
with region integrated moving vector
, in the finite state machine model of having set up, carry out the differentiation of current road condition; Set two threshold values: (1)
, represent the wagon flow length upper limit that road can hold; (2)
, represent that the region integrated moving of road tolerance is vectorial,
:
The parameter going out according to current detection
, M, and the state of previous frame, finite state machine carries out state to be maintained or changes; If road is in jogging state, and now detects before
,
, trigger " length is very big, and speed is minimum " condition, state is changed and changes into congestion in road; And if
,
, triggering " length is very big, speed " condition, state maintains; If
, " queue length the is shorter " condition that triggers occurs, and state is changed and changes into the coast is clear; If road is in unimpeded state, and now detects before
,
, trigger " length is very big, speed " condition, state is changed and changes into road jogging, and work as
time, triggering " fleet's length is shorter " condition, road will remain unimpeded state.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210240036.8A CN102768802B (en) | 2012-07-12 | 2012-07-12 | Method for judging road vehicle jam based on finite-state machine (FSM) |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210240036.8A CN102768802B (en) | 2012-07-12 | 2012-07-12 | Method for judging road vehicle jam based on finite-state machine (FSM) |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102768802A CN102768802A (en) | 2012-11-07 |
| CN102768802B true CN102768802B (en) | 2014-08-06 |
Family
ID=47096188
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210240036.8A Active CN102768802B (en) | 2012-07-12 | 2012-07-12 | Method for judging road vehicle jam based on finite-state machine (FSM) |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102768802B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108732385B (en) * | 2018-04-10 | 2020-07-17 | 厦门大学 | Drift correction device and automatic correction method of scanning probe microscopic imaging system |
| CN110110590A (en) * | 2019-03-25 | 2019-08-09 | 费今歌 | The determination method and Related product in congestion lane |
| CN109949581A (en) * | 2019-04-11 | 2019-06-28 | 浙江无极互联科技有限公司 | A kind of road condition judgment method based on computer vision |
| CN110192232B (en) * | 2019-04-18 | 2022-11-01 | 京东方科技集团股份有限公司 | Traffic information processing apparatus, system and method |
| CN114429710B (en) * | 2021-12-17 | 2023-12-15 | 华人运通(上海)自动驾驶科技有限公司 | Traffic flow analysis method and system based on V2X vehicle Lu Yun cooperation |
| CN116386344B (en) * | 2023-06-07 | 2023-08-25 | 江苏群力技术有限公司 | Road traffic monitoring system based on camera module |
| CN116913108B (en) * | 2023-09-13 | 2023-12-19 | 深圳市新城市规划建筑设计股份有限公司 | Urban traffic monitoring and scheduling method and system |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101271622A (en) * | 2007-03-22 | 2008-09-24 | 上海经达实业发展有限公司 | Instant condition discrimination and inducing system of urban highway traffic |
| CN101469985A (en) * | 2007-12-26 | 2009-07-01 | 河海大学常州校区 | Single-frame image detection apparatus for vehicle queue length at road junction and its working method |
| CN201203419Y (en) * | 2007-12-26 | 2009-03-04 | 河海大学常州校区 | Single frame image detection device of city road crossing vehicle queue length |
| CN101807345B (en) * | 2010-03-26 | 2012-07-04 | 重庆大学 | Traffic jam judging method based on video detection technology |
| CN101833863B (en) * | 2010-05-17 | 2013-09-11 | 瑞斯康达科技发展股份有限公司 | Method and device for detecting vehicle flow speed, as well as method and system for controlling traffic lights |
-
2012
- 2012-07-12 CN CN201210240036.8A patent/CN102768802B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN102768802A (en) | 2012-11-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102768802B (en) | Method for judging road vehicle jam based on finite-state machine (FSM) | |
| US9704060B2 (en) | Method for detecting traffic violation | |
| CN106373430B (en) | Intersection traffic early warning method based on computer vision | |
| CN102496281B (en) | Vehicle red-light violation detection method based on combination of tracking and virtual loop | |
| JP4973008B2 (en) | Vehicle discrimination device and program thereof | |
| CN102005120A (en) | Traffic intersection monitoring technology and system based on video image analysis | |
| CN103714325A (en) | Left object and lost object real-time detection method based on embedded system | |
| CN101872546A (en) | Video-based method for rapidly detecting transit vehicles | |
| CN102622884A (en) | Vehicle illegal turning behavior detection method based on tracking | |
| CN103617410A (en) | Highway tunnel parking detection method based on video detection technology | |
| CN101364347A (en) | Detection method for vehicle delay control on crossing based on video | |
| CN103903465A (en) | Method and system for releasing reason for road congestion in real time | |
| CN105354857B (en) | A kind of track of vehicle matching process for thering is viaduct to block | |
| CN105632170A (en) | Mean shift tracking algorithm-based traffic flow detection method | |
| CN111951576A (en) | Traffic light control system based on vehicle identification and method thereof | |
| CN106327880A (en) | Vehicle speed identification method and system based on monitored video | |
| Chen et al. | Traffic congestion classification for nighttime surveillance videos | |
| Klubsuwan et al. | Traffic violation detection using multiple trajectories evaluation of vehicles | |
| CN105427615B (en) | Make a dash across the red light capturing system and the method for robust under a kind of low-light (level) | |
| CN109409264B (en) | Video queuing overflow detection method based on local tracking | |
| Al Okaishi et al. | Vehicular queue length measurement based on edge detection and vehicle feature extraction | |
| Ardestani et al. | Signal timing detection based on spatial–temporal map generated from CCTV surveillance video | |
| Guo et al. | Vehicle detection and tracking based on optical field | |
| Al-Kadi et al. | Road scene analysis for determination of road traffic density | |
| CN105069407A (en) | Video-based traffic flow acquisition method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |