CN104680787B - A kind of congestion in road detection method - Google Patents
A kind of congestion in road detection method Download PDFInfo
- Publication number
- CN104680787B CN104680787B CN201510058449.8A CN201510058449A CN104680787B CN 104680787 B CN104680787 B CN 104680787B CN 201510058449 A CN201510058449 A CN 201510058449A CN 104680787 B CN104680787 B CN 104680787B
- Authority
- CN
- China
- Prior art keywords
- pixel
- prospect
- road
- gauss model
- background
- 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
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
- G08G1/0133—Traffic data processing for classifying traffic situation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
- G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
- G06V10/758—Involving statistics of pixels or of feature values, e.g. histogram matching
Abstract
The present invention relates to a kind of congestion in road detection method, comprise the following steps:1) video image of the present road of acquisition is handled, obtains the background and prospect of present road, be specially:Each pixel of present frame is modeled using mixed Gauss model;Vehicle and road surface are classified;The connected region of calculating prospect, judges whether the size of each connected region is less than setting value, if so, corresponding pixel then is included into background;To single pixel, illumination variation or prospect change are judged according to the change extraction feature of pixel in its neighborhood;Judge whether prospect is in non-vehicle region, if so, corresponding pixel then is included into background;Using judging into the pixel re -training mixed Gauss model of background, and update grader;2) roadway occupancy and vehicle average speed of present road are calculated according to obtained foreground pixel.Compared with prior art, the present invention has the advantages that speed is fast, robust is strong.
Description
Technical field
The present invention relates to Vehicle Detection field, more particularly, to a kind of congestion in road detection method.
Background technology
In city, as motor vehicle possesses increasing rapidly for quantity, the jam situation of road is increasingly serious, especially exists
The peak times such as upper and lower class, jam is more universal.At present, the method for discrimination of road traffic congestion is mainly passed through respectively
The traffic state data required for Vehicle Detection technology is obtained is planted, and by being analyzed these traffic state datas, being handled,
And then judge the jam situation of road.
Existing traffic jam detection method mainly includes:
1) Coil Detector method:Vehicle disturbs the earth magnetism magnetic by coil by the lead loop being embedded under road surface
Logical line simultaneously produces a voltage in coil, and this voltage is amplified by high-gain amplifier, the relay of detector is worked, from
And realize testing goal.But Coil Detector method does not have directionality, detection zone indefinite, repairing or installation need to suspend traffic,
Pavement life is influenceed, it is fragile.
2) Coil Detector method is answered:As the working method of single induction coil detector, but it can provide more accurate
Speed parameter.Answer Coil Detector method to be suspended traffic in repairing or installation, and easily influence pavement life, it is fragile, and
And it is more expensive than single induction coil.
3) video detecting method:Video camera absorbs the image of test point near zone, and image is carried out by computer program
Handle, recognize, so as to detect vehicle.The costly and accuracy of detection of video detecting method is by bright around weather, detection zone
Degree influence.
4) microwave detection method:By launching microwave signal, vehicle reflection radar microwave (Doppler effect) returns to antenna
Make the detection actuating of relay to reach testing goal.Microwave detection method can not detect vehicle that is static or running at a low speed, easily
It is affected by the external environment, is easily influenceed when the thresholding selection of transmitted wave is incorrect by people or thing, causes to know by mistake.
In the above-mentioned methods, video detection technology due to install do not damage road surface, do not result in traffic it is temporary transient in
The advantages of information break, obtained is more, in the time in recent years, have obtained development at full speed and universal application.Such as China specially
Sharp CN102867415A discloses a kind of congestion in road method of discrimination based on video detection technology, to regarding for the present road of acquisition
Frequency image is handled, and obtains the background image and foreground target of present road;According to the foreground target, calculate described current
The average lane occupation rate and time occupancy of road;According to the average lane occupation rate and time occupancy, according to foundation
Road traffic congestion criterion judge the jam situation of the present road.But prior art can not handle low resolution/
The video of complex scene/weather condition.
The content of the invention
It is an object of the present invention to overcome the above-mentioned drawbacks of the prior art and provide a kind of speed is fast, robust is strong
Congestion in road detection method.
The purpose of the present invention can be achieved through the following technical solutions:
A kind of congestion in road detection method, comprises the following steps:
1) video image of the present road of acquisition is handled, obtains the background and prospect of present road, be specially:
A) to the present frame of video image, each pixel of present frame is modeled using mixed Gauss model, judged
Whether each pixel belongs to prospect, if so, step b) is then performed, if it is not, then performing step f);
B) texture features and color characteristics of image are extracted, using RBF Kernel support vector machine classifier to vehicle
Classified with road surface;
C) connected region of prospect is calculated, judges whether the size of each connected region is less than setting value, if so, then will correspondence
Pixel be included into background, if it is not, then performing step d);
D) to single pixel, judge that illumination variation or prospect become according to the change extraction feature of pixel in its neighborhood
Change;
E) judge whether prospect is in non-vehicle region, if so, corresponding pixel then is included into background, perform step
F), if it is not, directly performing step f);
F) the pixel re -training mixed Gauss model for judging into background is used, and updates RBF Kernel supporting vector
Machine grader;
2) roadway occupancy and vehicle average speed of present road are calculated according to obtained foreground pixel.
Gauss model number is 5 in the mixed Gauss model.
In the step a), for each pixel, judge that the k-th belonged in mixed Gauss model is high according to its color value
This model, whether be in mixed Gauss model energy most strong, if so, respective pixel then is included into the back of the body if judging the Gauss model
Scape, if respective pixel otherwise is included into prospect.
In the step f), RBF Kernel SVMs is updated using the more new algorithm in Online-SVM algorithms
Grader.
In the step f), the frequency of re -training mixed Gauss model for 15 minutes once.
Compared with prior art, the present invention has advantages below:
1st, reliability is high, and the present invention is modeled using mixed Gauss model to each pixel of present frame, when judging
The influence of illumination variation is effectively rejected, so as to obtain higher Detection accuracy;
2nd, RBF Kernel of the invention support vector machine classifier and training mixed Gauss model are that change updates,
Real-time is good;
3rd, speed is fast, meets the demand of real-time;
4th, strong robustness, can be used under different scenes, can handle complex scene change.
Brief description of the drawings
Fig. 1 is schematic flow sheet of the invention.
Embodiment
The present invention is described in detail with specific embodiment below in conjunction with the accompanying drawings.The present embodiment is with technical solution of the present invention
Premised on implemented, give detailed embodiment and specific operating process, but protection scope of the present invention is not limited to
Following embodiments.
As shown in figure 1, the present embodiment provides a kind of congestion in road detection method, current time is obtained according to traffic video stream
Roadway occupancy and vehicle average speed, idiographic flow is as follows:
Step S1, is handled the video image of the present road of acquisition, obtains the background and prospect of present road, tool
Body is:
Step S101, to the present frame of video image, is built using mixed Gauss model to each pixel of present frame
Mould, judges whether each pixel belongs to prospect, if so, step b) is then performed, if it is not, then performing step f).
Average in mixed Gauss model is set to 0 when initial, variance is set to definite value, and Gauss model number is set as 5.
For each pixel, the k-th Gauss model for judging to belong in mixed Gauss model according to its color value, judging should
Whether Gauss model is that energy is most strong in mixed Gauss model, if so, respective pixel then is included into background, if otherwise will correspondence
Pixel is included into prospect.
Assuming that being that within a period of time, background (road surface) occurrence number is much larger than prospect (vehicle), while background (road surface)
Color is relatively fixed, therefore can form the maximum Gauss model of energy;And prospect (vehicle) etc., due to belonging to Different Individual, color is not
Unanimously, it is hardly formed stable Gauss model.Now single pixel judges into the accuracy of background more than 99%.But judge into
The pixel of prospect also has maximum probability to be not belonging to prospect (vehicle), and possible factor has:1) illumination variation;2) road debris is moved;
3) raindrop movement under special climate etc., therefore post-processed.
Step S102, extracts the texture features and color characteristics of image, using RBF Kernel support vector cassification
Device (SVMs based on RBF kernel functions) is classified to vehicle and road surface.Grader uses two kinds of spies of LBP and Color
Levy, main cause is:1) LBP describes texture features, and surface of vehicle is smoother, and lower frequency region spy's frame is stronger, and there is more make an uproar on road surface
Point, high frequency spy's frame is stronger;2) Color describes color characteristics, and for specific crossing, road surface color is fixed and vehicle color gap
Greatly.
Step S103, calculates the connected region of prospect, judges whether the size of each connected region is less than setting value, if so,
Corresponding pixel is then included into background, if it is not, then performing step d).The size of vehicle is fixed.It can be gone after so handling
Except debris such as raindrop, the scraps of paper.
Step S104, to single pixel, judged according to the change extraction feature of pixel in its neighborhood be illumination variation also
It is prospect change.The characteristic of illumination variation is global change, and change is uniform;And prospect changes, changed greatly in neighborhood, this
Sample processing can eliminate the influence of illumination variation.
Step S105, the track that vehicle occurs is fixed, judges whether prospect is in non-vehicle region, if so,
Corresponding pixel is then included into background, step S106 is performed, if it is not, directly performing step S106.
Step S106, using judging into the pixel re -training mixed Gauss model of background, and updates RBF Kernel's
Support vector machine classifier, updates RBF Kernel's using the more new algorithm (on-line learning algorithm) in Online-SVM algorithms
Support vector machine classifier, the frequency of re -training mixed Gauss model for 15 minutes once.
Step S2, the roadway occupancy and vehicle average speed of present road are calculated according to obtained foreground pixel.
Claims (5)
1. a kind of congestion in road detection method, it is characterised in that comprise the following steps:
1) video image of the present road of acquisition is handled, obtains the background and prospect of present road, be specially:
A) to the present frame of video image, each pixel of present frame is modeled using mixed Gauss model, judges each
Whether pixel belongs to prospect, if so, step b) is then performed, if it is not, then performing step f);
B) texture features and color characteristics of image are extracted, using RBF Kernel support vector machine classifier to vehicle and road
Classified in face;
C) connected region of prospect is calculated, judges whether the size of each connected region is less than setting value, if so, then by corresponding picture
Element is included into background, if it is not, then performing step d);
D) to single pixel, illumination variation or prospect change are judged according to the change extraction feature of pixel in its neighborhood,
The influence of illumination variation is rejected, wherein, the characteristic of the illumination variation is global change, and change is uniform;
E) judge whether prospect is in non-vehicle region, if so, corresponding pixel then is included into background, perform step f),
If it is not, directly performing step f);
F) the pixel re -training mixed Gauss model for judging into background is used, and updates RBF Kernel SVMs point
Class device;
2) roadway occupancy and vehicle average speed of present road are calculated according to obtained foreground pixel.
2. congestion in road detection method according to claim 1, it is characterised in that Gaussian mode in the mixed Gauss model
Type number is 5.
3. congestion in road detection method according to claim 1, it is characterised in that in the step a), for each picture
Element, the k-th Gauss model for judging to belong in mixed Gauss model according to its color value, whether judge the Gauss model is mixing
Energy is most strong in Gauss model, if so, respective pixel then is included into background, if respective pixel otherwise is included into prospect.
4. congestion in road detection method according to claim 1, it is characterised in that in the step f), using Online-
More new algorithm in SVM algorithm updates RBF Kernel support vector machine classifier.
5. congestion in road detection method according to claim 1, it is characterised in that in the step f), re -training is mixed
Close Gauss model frequency for 15 minutes once.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510058449.8A CN104680787B (en) | 2015-02-04 | 2015-02-04 | A kind of congestion in road detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510058449.8A CN104680787B (en) | 2015-02-04 | 2015-02-04 | A kind of congestion in road detection method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104680787A CN104680787A (en) | 2015-06-03 |
CN104680787B true CN104680787B (en) | 2017-08-01 |
Family
ID=53315769
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510058449.8A Active CN104680787B (en) | 2015-02-04 | 2015-02-04 | A kind of congestion in road detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104680787B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104933868B (en) * | 2015-06-17 | 2017-07-07 | 苏州大学 | A kind of real-time online Traffic State Detection Method based on Traffic Surveillance Video |
WO2018068311A1 (en) * | 2016-10-14 | 2018-04-19 | 富士通株式会社 | Background model extraction device, and method and device for detecting traffic congestion |
CN111402579A (en) * | 2020-02-29 | 2020-07-10 | 深圳壹账通智能科技有限公司 | Road congestion degree prediction method, electronic device and readable storage medium |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
TWI391876B (en) * | 2009-02-16 | 2013-04-01 | Inst Information Industry | Method and system for foreground detection using multi-modality fusion graph cut and computer program product |
CN101807345B (en) * | 2010-03-26 | 2012-07-04 | 重庆大学 | Traffic jam judging method based on video detection technology |
CN202563526U (en) * | 2012-03-22 | 2012-11-28 | 北京尚易德科技有限公司 | Transportation vehicle detection and recognition system based on video |
TWI602156B (en) * | 2013-06-14 | 2017-10-11 | 亞洲大學 | Moving foreground detection system and method thereof |
-
2015
- 2015-02-04 CN CN201510058449.8A patent/CN104680787B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN104680787A (en) | 2015-06-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102768804B (en) | Video-based traffic information acquisition method | |
CN102542289B (en) | Pedestrian volume statistical method based on plurality of Gaussian counting models | |
CN102819952B (en) | Method for detecting illegal lane change of vehicle based on video detection technique | |
CN103218816B (en) | A kind of crowd density estimation method and people flow rate statistical method based on video analysis | |
CN100502463C (en) | Method for collecting characteristics in telecommunication flow information video detection | |
CN102436644B (en) | Unstructured road detection method based on adaptive edge registration | |
CN103605953A (en) | Vehicle interest target detection method based on sliding window search | |
CN103117005A (en) | Lane deviation warning method and system | |
CN102867415B (en) | Video detection technology-based road jam judgement method | |
CN105336169A (en) | Method and system for judging traffic jams based on videos | |
CN105608431A (en) | Vehicle number and traffic flow speed based highway congestion detection method | |
CN103208190B (en) | Traffic flow detection method based on object detection | |
CN103617410A (en) | Highway tunnel parking detection method based on video detection technology | |
Liu et al. | Vision-based real-time lane marking detection and tracking | |
CN104183142A (en) | Traffic flow statistics method based on image visual processing technology | |
CN104680787B (en) | A kind of congestion in road detection method | |
CN105868734A (en) | Power transmission line large-scale construction vehicle recognition method based on BOW image representation model | |
CN103745216A (en) | Radar image clutter suppression method based on airspace feature | |
CN105069441A (en) | Moving vehicle detection method based on background updating and particle swarm optimization algorithm | |
CN102542805A (en) | Device for judging traffic jam based on videos | |
CN109993134A (en) | A kind of intersection vehicle checking method based on HOG and SVM classifier | |
CN104168462B (en) | Camera scene change detection method based on image angle point set feature | |
Bu et al. | A UAV photography–based detection method for defective road marking | |
CN104866844B (en) | A kind of crowd massing detection method towards monitor video | |
CN113537170A (en) | Intelligent traffic road condition monitoring method and computer readable storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |