CN104504897B - A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data - Google Patents
A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data Download PDFInfo
- Publication number
- CN104504897B CN104504897B CN201410505847.5A CN201410505847A CN104504897B CN 104504897 B CN104504897 B CN 104504897B CN 201410505847 A CN201410505847 A CN 201410505847A CN 104504897 B CN104504897 B CN 104504897B
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- track
- vehicle
- intersection
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/231—Hierarchical techniques, i.e. dividing or merging pattern sets so as to obtain a dendrogram
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Chemical & Material Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Analytical Chemistry (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a kind of intersection traffic properties of flow analysis based on track data and vehicle movement Forecasting Methodology, belong to intelligent transportation system and traffic flow parameter acquisition technique field.The present invention starts with from the space transient analysis of vehicle initial trace, describe and analyze from different perspectives the local geometric features of track, the multi-level spectral clustering processing framework based on the original coarse movement track of vehicle is formed, a variety of traffic direction patterns in intersection that track data is included are automatically extracted and analyze.Based on this, the present invention can obtain intersection split-phase position (signalized crossing) magnitude of traffic flow and each direction of motion vehicle by the detailed traffic characterisitic parameter such as journey time of intersection, in this, as the important supplement of conventional traffic data.The present invention is by the travel path for all moving vehicles for tracking current time, and the method matched using traffic direction trajectory model is predicted the next step behavior of vehicle, contributes to real-time early warning intersection security risk that may be present.
Description
Technical field
The invention belongs to intelligent transportation system (machine vision and image procossing) and traffic flow parameter acquisition technique field, tool
Body is related to a kind of intersection traffic properties of flow analysis based on the original rough track data of vehicle and vehicle movement Forecasting Methodology.
Background technology
Intersection is the important component of urban road system.For the actual measurement of intersection traffic stream operation characteristic
Undoubtedly there is important theory and practical significance with analysis, it will be analyzed for intersection capacity, delay and service level, be handed over
Prong canalized design and traffic organization optimize, and intersection control pipe ought to etc. provide important theoretical foundation (X.Li,
X.Li,D.Tang,X.Xu.Deriving features of traffic flow around an intersection
from trajectories of vehicles[C].18th International Conference on
Geoinformatics,Beijing,2010:1-5).Meanwhile, with being significantly increased for the Current City Road volume of traffic, traffic is disobeyed
Judicial act problem becomes increasingly conspicuous, and intersection order is often very chaotic, as road traffic accident multiplely.Therefore it is directed to
Intersection Exploration on Train Operation Safety, it is necessary to strengthen the research to urban road vehicle motion forecast method, in case can be further complete
Into vehicle safety early warning task.
Currently, the need for the growing and traffic control of transport need, multiple sensors are widely used in handing over
Logical state-detection.Compared to the conventional traffic stream of the registration of vehicle in an indirect way such as live manual testing and ground induction coil detector
Acquisition technique, video encoder server carrys out the mobility of registration of vehicle with monitoring device in a straightforward manner, and friendship can be recorded in detail
The running of the numerous vehicles of prong and influence each other.The vehicle movement initial trace gathered by traffic video treatment technology
Data, undoubtedly important basic data source (Z.Fu, W.Hu, T.Tan.Similarity based a vehicle
trajectory clustering and anomaly detection[C].IEEE International Conference
on Image Processing.2005,2:602-605)(X.Li,W.Hu,W.Hu.A coarse-to-fine strategy
for vehicle motion trajectory clustering[C].Proceedings ofthe 18th
International Conference on Pattern Recognition.2006,1:591-594).For specific friendship
Logical environment, traditional method of trajectory clustering assumes existed or be readily apparent error free and unremitting moving object rail
Mark (B.T.Morris, M.M.Trivedi.A survey ofvision-based trajectory learning and
analysis for surveillance[J].IEEE Trans.on Circuits and Systems for Video
Technology.2008,18(8):1114-1127)(S.Atev,G.Miller,
N.P.Papanikolopoulos.Clustering of vehicle trajectories[J].IEEE Trans.on
Intelligent Transportation Systems.2010,11(3):674-657).Due to the complexity of traffic environment in itself
Property, during real video stream is handled, the reliability of vehicle detection and track algorithm is relatively low, and this will cause vehicle to be transported
There are a series of serious problems in dynamic rail mark result, such as fragment, tracking interruption and error hiding.Therefore, people often through
Manual synchronizing improves track quality.To become impossible by manual synchronizing yet with following 2 reasons:(1) with
Traffic video data are sharply increased, and manual synchronizing expends the time very much, ensure using manual synchronizing quality of data general completely
It can become impossible;(2) it is difficult to avoid that the undesirable artificial deviation of introducing by manually operating.In summary, current work
Make more by very time-consuming manual synchronizing, so as to it is difficult to obtain extensive high-quality intersection vehicles movement locus number
According to ultimately resulting in the condition for not possessing and carrying out intersection traffic stream operation characteristic site-test analysis and vehicle movement prediction work.
The content of the invention
Vehicle movement track is applied to the research of DETECTION OF TRAFFIC PARAMETERS by the present invention, and root problem to be solved is that do not having
In the case of having manual synchronizing, directly intersection is robustly found from the Vehicle tracing data of original coarse (low quality)
Inherent traffic flow pattern.The traffic direction pattern obtained is analyzed based on video frequency vehicle actual measurement track data clusters, can be to road
Intersection traffic environment is identified and strengthens understanding, finally clearly describes the true of intersection vehicles motor pattern and vehicle pass-through
Carry out journey.In order to solve to be difficult that the present invention is with part the problem of obtaining extensive high-quality intersection vehicles motion trace data
Robust features extract it is theoretical based on, it is proposed that it is a kind of using the analysis microcosmic geometric properties in track analysis method (S.Huet,
E.Karatekin,V.S.Tran,I.Fanget,S.Cribier,J.Henry.Analysis of transient
behavior in complex trajectories:application to secretory vesicle dynamics
[J].Biophysical Journal.2006,91(9):3542-3559)(J.A.Helmuth,C.J.Burckhardt,
P.Koumoutsakos,U.F.Greber,I.F.Sbalzarini.A novel supervised trajectory
segmentation algorithm identifies distinct types ofhuman adenovirus motion in
host cells[J].Journal ofStructural Biology.2007,159(3):347-358), retouch from different perspectives
The local geometric features of track are stated and analyzed, are handled directly against coarse initial trace data, it is proposed that a set of general
Multi-level spectral clustering processing framework based on the original coarse movement track of vehicle, automatically extracts and analyzes (data mining) track number
According to a variety of traffic direction patterns in the intersection included.
A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data, it is characterised in that
Comprise the following steps:
Step 1:Intersection vehicles based on motion tracking move original rough track and obtained, and set up the extensive car in intersection
Motion trace data set;
Step 1.1:Intersection vehicles motion initial trace collection
Using the tracking based on image block in OpenCV, automatically extract intersection vehicles and move original track data,
And it is expressed as vehicle movement point sequence T:
T={ t1,t2,…,ti,…,tn}
={ (x1,y1),(x2,y2),…,(xi,yi),…,(xn,yn)}
And track step sequence S:
S={ s1,s2,…,si,…,sn-1}
={ (δ x1,δy1),…,(δxi,δyi),…,(δxn-1,δyn-1)}
Wherein, ti=(xi,yi) represent the position of moving vehicle ith sample dot center, si=(δ xi,δyi)=(xi+1-
xi,yi+1-yi) deviation of neighbouring sample dot center is represented, n represents the sum that track of vehicle includes sampled point;
Step 1.2:Vehicle movement initial trace is pre-processed
Following smoothing processing is carried out for every track:(1) sample noise is considered, if the distance between sampled point
It is sufficiently small, just continuous sampled point is merged and replaced by first sampled point;(2) complete deletion length is less than pre-
Define the short track of threshold value;(3) it is smoothed using mean filter, to retain prototype structure substantive characteristics, wherein
For the position of the moving vehicle ith sample point after smoothing processing, tkFor the position of k-th of sampled point of moving vehicle before smoothing processing
Put, w is the smooth step number selected in experiment:
Step 2:Intersection vehicles motor pattern study based on the original rough track multilayer spectral clustering of vehicle
Step 2.1:Multi-level track characteristic is extracted
Carry out multi-level consolidated statement using linearity and flexibility, three kinds of features of course bearing histogram and track centers respectively
Up to the local feature of track;
Linearity and flexibility refer to the measurement of vehicle traffic direction mean change, and linearity is defined as follows:
Flexibility is defined as follows:
Wherein, αiRepresent siWith si+1Direction change between step, positive direction is turned to while setting and becoming to the left;N represents car
Track includes the sum of sampled point;
Course bearing histogram (Trajectory Directional Histogram, abbreviation TDH) is a kind of description rail
The expression of mark directional statistics feature, first, calculates the deflection β of ith sample pointi:
Wherein, (δ xi)2+(δyi)2≠ 0 and βi∈[-π,+π);Then, by it is interval [- π ,+π) be evenly dividing as N number of sub-district
Between, and by all sampled points according to the different mappings of deflection into corresponding subinterval;Finally, adopted according in all subintervals
The sum M of sampling point normalizes the number M of sampled point in j-th of subintervaljFor
rj=Mj/ M, j=1,2 ..., N.
Then course bearing histogram TDH is expressed as:
p3=(r1,…,rj,…,rN)
Track centers refer to center position of the every track after smoothing processing:
Wherein,For the position of the moving vehicle ith sample point after smoothing processing;
Step 2.2:Spectral clustering and multi-level distance metric
Using the spectral clustering implementation method based on Random Walk, data-oriented collection Z=(z1,…,zu,…,zv,…,
zm), then similarity matrix W each composition element definition is:
Wuv=exp (- dist (zu,zv)/2σ2)
Wherein, dist (zu,zv) it is distance metric, σ is standard variance;
Incidence matrix P is drawn by similarity matrix W and diagonal matrix D conversions:
P=D-1/2WD-1/2
Wherein, diagonal matrix
D=diag (D11,…,Duu,…,Dmm)
In element DuuRepresent the summation of u column elements in similarity matrix WSolved for P corresponding to it
Tag system characteristic value and characteristic vector can just complete spectral clustering;
The need for for different levels spectral clustering, the distance metric in above formula is calculated according to different track characteristics
dist(zu,zv), use Euclidean distance in first layer and third layer:
E (u, v)=| | zu-zv||2
Wherein, zuAnd zvThe feature of u articles and the v articles track is represented respectively;
In the second layer using calculating Pasteur's distance:
Wherein TDHubAnd TDHvbB-th of element in the corresponding TDH of the u articles and the v articles track is represented respectively;
Step 3:Intersection traffic properties of flow is analyzed and motion prediction
Step 3.1:Sub-trajectory is represented
Using tapped delay cable architecture, the different vehicle movement track of every length is expressed as multiple regular length ξ
Sub-trajectory, ξ is a predefined parameter, it is considered to which one between the stable description of initial trace and low operand is traded off, should
Value is defined as follows:
Wherein, L is the average length of track data collection, lminIt is the minimum length of track data collection;
Step 3.2:Traffic stream characteristics are analyzed
Different vehicle sport modes shows true stroke of the vehicle of approaching intersection when by intersection, bag
Intersection upstream section and turn particularly to that they are reached are included, turns right, keep straight on, turn left;
Intersection shunt to the magnitude of traffic flow clustered using every kind of vehicle movement in the sum of track of vehicle represent, and pin
The histogram of the average turning time of vehicle each track arc length in every kind of vehicle movement cluster of different turn types is represented;
Step 3.3:Motion prediction
Given componental movement track, is included to most universal in 12 track class set using k nearest neighbor algorithms (k-NN)
Motor pattern, the similarity measurement in k-NN defines according to Euclidean distance:
Wherein, Tp(i) Current vehicle driving trace T is representedpI-th point, and Kc(i) phase of various motor patterns is represented
Answer track subclass i-th point;
By Current vehicle driving trace TpTrack subclass K corresponding to various motor patternscIt is compared, calculating vehicle will
Carry out possible movement locus probability Pc:
Predict that the motion that current kinetic vehicle is likely to occur in the future becomes by selecting the type of gesture with maximum probability
Gesture, while completing the prediction of the current all vehicle movement trend in intersection, realizes intersection vehicles sports safety early warning.
Compared with prior art, the present invention has following clear superiority:
(1) present invention proposes that a kind of multilayer subslot Spectral Clustering is used to recognize different vehicle sport modes, is used for
Analyze intersection traffic properties of flow and vehicle movement safe prediction early warning.
(2) root problem to be solved by this invention is in the case of no manual synchronizing, directly from original low-quality
Robustly found in the track of vehicle data of amount in vehicle motor pattern.
(3) present invention is based on local robust features extract theory, directly according to original low-quality track of vehicle
Data, analyse in depth the local geometric features near every in movement locus.In track, Data processing isolated point would generally be produced
Raw many errors, the interference of these single point tolerances can be avoided using our bright institute's application method.
(4) present invention can obtain intersection split-phase position (signalized crossing) magnitude of traffic flow and each direction of motion car
By the detailed traffic characterisitic parameter such as journey time of intersection, in this, as the important supplement of conventional traffic data.
(5) present invention is by the travel path for all moving vehicles for tracking current time, using traffic direction track mould
The method of formula matching predicts the next step behavior of vehicle, real-time early warning intersection security risk that may be present.
Brief description of the drawings
The general frame of Fig. 1 methods involved in the present invention;
The multi-level moving vehicle trajectory clustering result in Fig. 2 intersections;
Fig. 3 a-3f intersection vehicles motion tracking results;
Fig. 4 a-4b initial trace preprocessing process;
Fig. 5 tapped delay lines;
Vehicle sport mode recognition results of Fig. 6 a-6w based on multilayer subslot spectral clustering framework;
The multi-level spectral clustering performance comparisions of Fig. 7;
Fig. 8 a-8f intersection traffics properties of flow is analyzed;
Fig. 9 intersection vehicles motion predictions.
Embodiment
The present invention will be further described with reference to the accompanying drawings and examples.
The embodiment of the present invention is realized on the PC for installing VC2008 and OpenCV2.4.5.
The flow chart of present invention method is as shown in figure 1, comprise the following steps:
Step 1:Intersection vehicles based on motion tracking move original rough track and obtained, and set up the extensive car in intersection
Motion trace data set.
Step 1.1:Intersection vehicles motion initial trace collection.
The true traffic video that the present invention is obtained by the video camera set up on the high building near the intersection of Beijing, is surveyed
The performance of multi-level spectral clustering framework is tried, as shown in Figures 2 and 3.The object of which movement track algorithm realized using OpenCV, to complete
The frame traffic video sequence of portion 17387 is handled, and symbiosis is into 1123 tracks, as shown in white portion in Fig. 3 d.It is each in Fig. 3
Figure particular content is expressed as follows:A. sequence of video images;B. vehicle movement tracking result;C. background image;D. be added to background
All movement locus in figure;E.3 trajectory diagram is tieed up;F.2 trajectory diagram is tieed up.
Step 1.2:Original vehicle movement locus is pre-processed.
After pretreatment, 997 tracks are still suffered from, as shown in fig. 4 a.Each figure particular content is expressed as follows in Fig. 4:A. fold
It is added to the intersection initial trace pre-processed results in Background;B. trajectory diagrams are tieed up in the 2 of intersection initial trace pre-processed results.
Step 2:Intersection vehicles motor pattern study based on the original rough track multilayer spectral clustering of vehicle.
Step 2.1:Multi-level track characteristic is extracted.
Step 2.2:As indicated by traffic environment and signal control strategy, 12 are co-existed near the intersection
Plant typical vehicle sport mode.This vehicle sport mode cluster result recognized with the present invention using multi-level spectral clustering framework
Be it is consistent, as shown in Figure 6.In order to more effectively represent each track cluster, the present invention constructs template track so as to more effective
Each track cluster of representative.Actual template track is exactly the distance of other all tracks and the minimum cluster heart into same cluster.
Given initial trace data as shown in fig. 4 a, the present invention is represented by multi-level Spectral Clustering Multi-layer technology should
The different tracks cluster of intersection vehicles unique motion pattern.Each track cluster all (is superimposed by template track in initial trace
The thick segment with arrow) represent.Fig. 6 a, Fig. 6 b, Fig. 6 c are followed successively by the cluster result of different levels, there is 4,8,12 rails respectively
Mark cluster, wherein representing template track with different gray scales.4 track clusters in first layer may particularly denote as Fig. 6 d-6g, and this 4
Cluster track has been further divided into 8 clusters in the second layer, as shown in Fig. 6 h-6o.There are 4 clusters (Fig. 6 h-6k) in the cluster of the above 8 again the
8 clusters are further divided into three layers, as shown in Fig. 6 p-6w.There are 12 different sections in tree structure shown in final Fig. 2
Point.Each figure particular content is expressed as follows in Fig. 6:A. first layer cluster result (4 track clusters);B. second layer cluster result (8
Track cluster);C. third layer cluster result (12 track clusters);D. first layer turning 1;E. first layer turning 2;F. first layer is kept straight on
1;G. first layer straight trip 2;H. the second layer turn 1 direction 1;I. the second layer turn 1 direction 2;J. the second layer turn 2 directions 1;K.
Two layers of 2 direction 2 of turning;L. the second layer keep straight on 1 direction 1;M. the second layer keep straight on 1 direction 2;N. the second layer keep straight on 2 directions 1;O.
Two layers of 2 direction 2 of straight trip;P. 1 direction of third layer turning 1 is right;Q. 1 direction of third layer turning 1 is left;R. third layer turn 1 direction 2
It is left;S. 1 direction of third layer turning 2 is right;T. 2 direction of third layer turning 1 is left;U. 2 direction of third layer turning 1 is right;V. third layer turns
Curved 2 direction 2 is right;W. 2 direction of third layer turning 2 is left.
For qualitative assessment typical vehicle motor pattern cluster result, the present invention uses as follows closely with separating criterion
(TSC) the trajectory clustering effect of each layer of inspection:
Wherein cjIt is the template track of jth cluster, TSC measures in cluster separating degree between tight ness rating and cluster simultaneously.TSC numerical value is smaller
Represent that systematic function is better.Fig. 7 (abscissa represents ever-increasing cluster level, and ordinate represents TSC value) clearly show with
The increase of the number of plies, motor pattern is assembled better.
Step 3:Intersection traffic properties of flow is analyzed and motion prediction.
Step 3.1:Sub-trajectory is represented.
Step 3.2:Traffic stream characteristics are analyzed, during present invention primarily contemplates point garage's direction magnitude of traffic flow with average stroke
Between.
There are 12 kinds of typical vehicle motor patterns near selected intersection.According to the difference in reached section, these allusion quotations
Pattern formula can be classified as 4 kinds of arrival types, and every kind of arrival type further respectively has 3 kinds of turn types, including turn right, directly
Row, left-hand bend, as shown in Figure 8 a-8d.Fig. 8 e and Fig. 8 f are expressed as the magnitude of traffic flow of given vehicle operation stroke and are averaged
Journey time.Ordinate y illustrates y in the percentage that a kind of magnitude of traffic flow of pattern accounts for total flow, Fig. 8 f and illustrated in Fig. 8 e
The average and variance of turn inside diameter time in per cluster vehicle operation stroke.Each figure particular content is expressed as follows in Fig. 8:A. the 1st
3 kinds of traffic directions of part of path;B. 3 kinds of traffic directions of the 2nd part of path;C. 3 kinds of traffic directions of the 3rd part of path;D. the 4th line
3 kinds of traffic directions in section;E. (abscissa represents that 12 kinds of different vehicle sport modes gather to the magnitude of traffic flow that intersection is respectively flowed to
Class result);F. (abscissa represents 12 kinds of different vehicle sport mode cluster knots to the average travel time that intersection is respectively flowed to
Really).These numerical value have been set out in detail in table 1.
The traffic stream characteristics of table 1 are analyzed
Step 3.3:Intersection vehicles motion prediction.
Fig. 9 illustrates specific motion prediction example, wherein thick black solid line, which is represented, is detected vehicle (using grey ellipse mark
Note) current kinetic track, thin black dotted line represents prediction locus.Percentage whiteness number beside thin black dotted line represents tested measuring car
In intersection may motor pattern probability.Vehicle enters the traffic scene from right side section in fig .9, only remains general
Prediction locus before rate value corresponding to 3.When vehicle turns left, only a kind of probable value of prediction locus increases to
99.42%.It is similar, the prediction of the current all vehicle movement trend in intersection can be completed, and then realize intersection vehicles
Sports safety early warning.
Finally it should be noted that:Above example only not limits technical side described in the invention to illustrate the present invention
Case;Therefore, although this specification with reference to above-mentioned example to present invention has been detailed description, this area it is common
It will be appreciated by the skilled person that still can be modified to the present invention or equivalent substitution;And all do not depart from invention spirit and
The technical scheme of scope and its improvement, it all should cover among scope of the presently claimed invention.
Claims (1)
1. a kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data, it is characterised in that bag
Include following steps:
Step 1:Intersection vehicles based on motion tracking move original rough track and obtained, and set up the extensive vehicle fortune in intersection
Dynamic rail mark data acquisition system;
Step 1.1:Intersection vehicles motion initial trace collection
Using the tracking based on image block in OpenCV, automatically extract intersection vehicles and move original track data, and table
It is shown as vehicle movement point sequence T:
T={ t1,t2,…,ti,…,tn}
={ (x1,y1),(x2,y2),…,(xi,yi),…,(xn,yn)}
And track step sequence S:
S={ s1,s2,…,si,…,sn-1}
={ (δ x1,δy1),…,(δxi,δyi),…,(δxn-1,δyn-1)}
Wherein, ti=(xi,yi) represent moving vehicle ith sample point position, si=(δ xi,δyi)=(xi+1-xi,yi+1-yi)
The deviation of neighbouring sample point is represented, n represents the sum that track of vehicle includes sampled point;
Step 1.2:Vehicle movement initial trace is pre-processed
Following smoothing processing is carried out for every track:(1) sample noise is considered, if the distance between sampled point is enough
It is small, just continuous sampled point is merged and replaced by first sampled point;(2) complete deletion length is less than predefined
The short track of threshold value;(3) it is smoothed using mean filter, to retain prototype structure substantive characteristics, whereinIt is flat
The position of moving vehicle ith sample point after sliding processing, tkFor the position of k-th of sampled point of moving vehicle before smoothing processing, w
For the smooth step number selected in experiment:
<mrow>
<mover>
<msub>
<mi>t</mi>
<mi>i</mi>
</msub>
<mo>&OverBar;</mo>
</mover>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>w</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>k</mi>
<mo>=</mo>
<mi>i</mi>
<mo>-</mo>
<mi>w</mi>
<mo>/</mo>
<mn>2</mn>
</mrow>
<mrow>
<mi>i</mi>
<mo>+</mo>
<mi>w</mi>
<mo>/</mo>
<mn>2</mn>
</mrow>
</munderover>
<msub>
<mi>t</mi>
<mi>k</mi>
</msub>
</mrow>
Step 2:Intersection vehicles motor pattern study based on the original rough track multilayer spectral clustering of vehicle
Step 2.1:Multi-level track characteristic is extracted
Carry out multi-level Integrative expression rail using linearity and flexibility, three kinds of features of course bearing histogram and track centers respectively
The local feature of mark;
Linearity and flexibility refer to the measurement of vehicle traffic direction mean change, and linearity is defined as follows:
<mrow>
<msub>
<mi>p</mi>
<mn>1</mn>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mi>n</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>n</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</munderover>
<msub>
<mi>cos&alpha;</mi>
<mi>i</mi>
</msub>
</mrow>
Flexibility is defined as follows:
<mrow>
<msub>
<mi>p</mi>
<mn>2</mn>
</msub>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mi>n</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>n</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</munderover>
<msub>
<mi>sin&alpha;</mi>
<mi>i</mi>
</msub>
</mrow>
Wherein, αiRepresent siWith si+1Direction change between step, positive direction is turned to while setting and becoming to the left;N represents vehicle rail
Mark includes the sum of sampled point;
Course bearing histogram (Trajectory Directional Histogram, abbreviation TDH) is a kind of description track side
To the expression of statistics feature, first, the deflection β of ith sample point is calculatedi:
<mrow>
<msub>
<mi>&beta;</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mi>arctan</mi>
<mrow>
<mo>(</mo>
<mfrac>
<mrow>
<msub>
<mi>&delta;y</mi>
<mi>i</mi>
</msub>
</mrow>
<mrow>
<msub>
<mi>&delta;x</mi>
<mi>i</mi>
</msub>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>&delta;x</mi>
<mi>i</mi>
</msub>
<mo>></mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>arctan</mi>
<mrow>
<mo>(</mo>
<mfrac>
<mrow>
<msub>
<mi>&delta;y</mi>
<mi>i</mi>
</msub>
</mrow>
<mrow>
<msub>
<mi>&delta;x</mi>
<mi>i</mi>
</msub>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mo>+</mo>
<mi>&pi;</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>&delta;x</mi>
<mi>i</mi>
</msub>
<mo><</mo>
<mn>0</mn>
<mo>,</mo>
<msub>
<mi>&delta;y</mi>
<mi>i</mi>
</msub>
<mo>></mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>arctan</mi>
<mrow>
<mo>(</mo>
<mfrac>
<mrow>
<msub>
<mi>&delta;y</mi>
<mi>i</mi>
</msub>
</mrow>
<mrow>
<msub>
<mi>&delta;x</mi>
<mi>i</mi>
</msub>
</mrow>
</mfrac>
<mo>)</mo>
</mrow>
<mo>-</mo>
<mi>&pi;</mi>
</mrow>
</mtd>
<mtd>
<mrow>
<msub>
<mi>&delta;x</mi>
<mi>i</mi>
</msub>
<mo><</mo>
<mn>0</mn>
<mo>,</mo>
<msub>
<mi>&delta;y</mi>
<mi>i</mi>
</msub>
<mo><</mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein, (δ xi)2+(δyi)2≠ 0 and βi∈[-π,+π);Then, by it is interval [- π ,+π) be evenly dividing as N number of subinterval, and
By all sampled points according to the different mappings of deflection into corresponding subinterval;Finally, according to sampled point in all subintervals
Sum M normalize the number M of sampled point in j-th of subintervaljFor
rj=Mj/ M, j=1,2 ..., N
Then course bearing histogram TDH is expressed as:
p3=(r1,…,rj,…,rN)
Track centers refer to center position of the every track after smoothing processing:
<mrow>
<msub>
<mi>p</mi>
<mn>4</mn>
</msub>
<mo>=</mo>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</msubsup>
<msub>
<mover>
<mi>t</mi>
<mo>&OverBar;</mo>
</mover>
<mi>i</mi>
</msub>
<mo>/</mo>
<mi>n</mi>
</mrow>
Wherein,For the position of the moving vehicle ith sample point after smoothing processing;
Step 2.2:Spectral clustering and multi-level distance metric
Using the spectral clustering implementation method based on Random Walk, data-oriented collection Z=(z1,…,zu,…,zv,…,zm), then
Similarity matrix W each composition element definition is:
Wuv=exp (- dist (zu,zv)/2σ2)
Wherein, dist (zu,zv) it is distance metric, σ is standard variance;
Incidence matrix P is drawn by similarity matrix W and diagonal matrix D conversions:
P=D-1/2WD-1/2
Wherein, diagonal matrix
D=diag (D11,…,Duu,…,Dmm)
In element DuuRepresent the summation of u column elements in similarity matrix WThe feature corresponding to it is solved for P
System features value and characteristic vector can just complete spectral clustering;
The need for for different levels spectral clustering, the distance metric dist in above formula is calculated according to different track characteristics
(zu,zv), use Euclidean distance in first layer and third layer:
E (u, v)=| | zu-zv||2
Wherein, zuAnd zvThe feature of u articles and the v articles track is represented respectively;
Pasteur's distance is used in the second layer:
<mrow>
<mi>H</mi>
<mrow>
<mo>(</mo>
<mi>u</mi>
<mo>,</mo>
<mi>v</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msup>
<mrow>
<mo>&lsqb;</mo>
<mn>1</mn>
<mo>-</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>b</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>m</mi>
</munderover>
<msqrt>
<mrow>
<msub>
<mi>TDH</mi>
<mrow>
<mi>u</mi>
<mi>b</mi>
</mrow>
</msub>
<msub>
<mi>TDH</mi>
<mrow>
<mi>v</mi>
<mi>b</mi>
</mrow>
</msub>
</mrow>
</msqrt>
<mo>&rsqb;</mo>
</mrow>
<mrow>
<mn>1</mn>
<mo>/</mo>
<mn>2</mn>
</mrow>
</msup>
<mo>&Element;</mo>
<mo>&lsqb;</mo>
<mn>0</mn>
<mo>,</mo>
<mn>1</mn>
<mo>&rsqb;</mo>
</mrow>
Wherein TDHubAnd TDHvbB-th of element in the corresponding TDH of the u articles and the v articles track is represented respectively;
Step 3:Intersection traffic properties of flow is analyzed and motion prediction
Step 3.1:Sub-trajectory is represented
Using tapped delay cable architecture, the different vehicle movement track of every length is expressed as to multiple regular length ξ
Track, ξ is a predefined parameter, it is considered to which one between the stable description of initial trace and low operand is traded off, and the value is determined
Justice is as follows:
<mrow>
<mi>&xi;</mi>
<mo>=</mo>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
<mrow>
<mo>(</mo>
<msqrt>
<mi>L</mi>
</msqrt>
<mo>,</mo>
<msub>
<mi>l</mi>
<mi>min</mi>
</msub>
<mo>)</mo>
</mrow>
</mrow>
Wherein, L is the average length of track data collection, lminIt is the minimum length of track data collection;
Step 3.2:Traffic stream characteristics are analyzed
Different vehicle sport modes shows true stroke of the vehicle of approaching intersection when by intersection, including it
The intersection upstream section that is reached and turn particularly to, turn right, keep straight on, turn left;
Intersection shunt to the magnitude of traffic flow clustered using every kind of vehicle movement in the sum of track of vehicle represent, and for not
Represented with the histogram of the average turning time of vehicle each track arc length in every kind of vehicle movement cluster of turn type;
Step 3.3:Motion prediction
Given componental movement track, most common fortune in 12 track class set is included to using k nearest neighbor algorithms (k-NN)
Similarity measurement in dynamic model formula, k-NN is defined according to Euclidean distance:
<mrow>
<msub>
<mi>d</mi>
<mi>c</mi>
</msub>
<mrow>
<mo>(</mo>
<msub>
<mi>T</mi>
<mi>p</mi>
</msub>
<mo>,</mo>
<msub>
<mi>K</mi>
<mi>c</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>=</mo>
<msqrt>
<mrow>
<msub>
<mi>&Sigma;</mi>
<mrow>
<mi>i</mi>
<mo>&Element;</mo>
<mi>&xi;</mi>
</mrow>
</msub>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>T</mi>
<mi>p</mi>
</msub>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
<mo>-</mo>
<msub>
<mi>K</mi>
<mi>c</mi>
</msub>
<mo>(</mo>
<mi>i</mi>
<mo>)</mo>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
</mrow>
Wherein, Tp(i) Current vehicle driving trace T is representedpIth sample point, and Kc(i) phase of various motor patterns is represented
The ith sample point of track subclass is answered,
By Current vehicle driving trace TpTrack subclass K corresponding to various motor patternscIt is compared, calculating vehicle in the future can
The movement locus probability P of energyc:
<mrow>
<msub>
<mi>P</mi>
<mi>c</mi>
</msub>
<mo>=</mo>
<mfrac>
<mrow>
<mn>1</mn>
<mo>/</mo>
<msub>
<mi>d</mi>
<mi>c</mi>
</msub>
</mrow>
<mrow>
<msubsup>
<mi>&Sigma;</mi>
<mrow>
<mi>c</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mn>12</mn>
</msubsup>
<mn>1</mn>
<mo>/</mo>
<msub>
<mi>d</mi>
<mi>c</mi>
</msub>
</mrow>
</mfrac>
</mrow>
The movement tendency that current kinetic vehicle is likely to occur in the future is predicted by selecting the type of gesture with maximum probability, together
When complete the predictions of the current all vehicle movement trend in intersection, realize intersection vehicles sports safety early warning.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410505847.5A CN104504897B (en) | 2014-09-28 | 2014-09-28 | A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410505847.5A CN104504897B (en) | 2014-09-28 | 2014-09-28 | A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104504897A CN104504897A (en) | 2015-04-08 |
CN104504897B true CN104504897B (en) | 2017-10-31 |
Family
ID=52946470
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410505847.5A Active CN104504897B (en) | 2014-09-28 | 2014-09-28 | A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104504897B (en) |
Families Citing this family (32)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105117737A (en) * | 2015-10-08 | 2015-12-02 | 北京邮电大学 | Method and apparatus for determining real direction of vehicle on basis of locus vector of vehicle |
CN105374209B (en) * | 2015-11-05 | 2018-02-06 | 北京航空航天大学 | A kind of urban area road network running status characteristics information extraction method |
CN106372619B (en) * | 2016-09-20 | 2019-08-09 | 北京工业大学 | A kind of detection of vehicle robust and divided lane reach summation curve estimation method |
CN106530698A (en) * | 2016-11-22 | 2017-03-22 | 青岛理工大学 | Estimation method for city road travel time taking exit turning into consideration |
CN107315994B (en) * | 2017-05-12 | 2020-08-18 | 长安大学 | Clustering method based on Spectral Clustering space trajectory |
CN107830865B (en) * | 2017-10-16 | 2020-04-17 | 东软集团股份有限公司 | Vehicle target classification method, device, system and computer program product |
CN108399741B (en) * | 2017-10-17 | 2020-11-27 | 同济大学 | Intersection flow estimation method based on real-time vehicle track data |
CN109840660B (en) * | 2017-11-29 | 2021-07-30 | 北京四维图新科技股份有限公司 | Vehicle characteristic data processing method and vehicle risk prediction model training method |
CN108111173B (en) * | 2017-12-27 | 2021-03-26 | 东软集团股份有限公司 | Track compression method and device, storage medium and electronic equipment |
CN108648445B (en) * | 2018-04-19 | 2020-02-21 | 浙江浙大中控信息技术有限公司 | Dynamic traffic situation prediction method based on traffic big data |
CN110444011B (en) * | 2018-05-02 | 2020-11-03 | 杭州海康威视***技术有限公司 | Traffic flow peak identification method and device, electronic equipment and storage medium |
CN108711288B (en) * | 2018-06-07 | 2020-11-20 | 郑州大学 | Quantitative evaluation method for right turn connection safety problem of non-motor vehicles at connection intersection |
CN109255315B (en) * | 2018-08-30 | 2021-04-06 | 跨越速运集团有限公司 | People and vehicle separation judgment method and device during vehicle leaving |
WO2020079074A2 (en) * | 2018-10-16 | 2020-04-23 | Five AI Limited | Autonomous vehicle planning |
CN109359690B (en) * | 2018-10-19 | 2021-10-22 | 江苏智通交通科技有限公司 | Vehicle travel track identification method based on checkpoint data |
CN109767615B (en) * | 2018-10-19 | 2021-05-18 | 江苏智通交通科技有限公司 | Method for analyzing key flow direction and key path of road network traffic flow |
CN109448370B (en) * | 2018-10-29 | 2021-09-28 | 东南大学 | Traffic control subarea division method based on vehicle track data |
CN109523807B (en) * | 2018-11-28 | 2020-06-05 | 湖南大学 | Traffic network vehicle distributed control method |
CN109636040A (en) * | 2018-12-13 | 2019-04-16 | 杭州杰富睿科技有限公司 | A kind of drinking-water point distributed intelligence system |
DE102019002790B4 (en) * | 2019-04-16 | 2023-05-04 | Mercedes-Benz Group AG | Method for predicting a traffic situation for a vehicle |
CN110400461B (en) * | 2019-07-22 | 2021-01-12 | 福建工程学院 | Road network change detection method |
US11443184B2 (en) | 2019-08-19 | 2022-09-13 | Toyota Research Institute, Inc. | Methods and systems for predicting a trajectory of a road agent based on an intermediate space |
CN110728842B (en) * | 2019-10-23 | 2021-10-08 | 江苏智通交通科技有限公司 | Abnormal driving early warning method based on reasonable driving range of vehicles at intersection |
CN110570658B (en) * | 2019-10-23 | 2022-02-01 | 江苏智通交通科技有限公司 | Method for identifying and analyzing abnormal vehicle track at intersection based on hierarchical clustering |
CN110782667A (en) * | 2019-10-30 | 2020-02-11 | 北京百度网讯科技有限公司 | Signal lamp time-sharing timing method and device, electronic equipment and storage medium |
CN110827540B (en) * | 2019-11-04 | 2021-03-12 | 黄传明 | Motor vehicle movement mode recognition method and system based on multi-mode data fusion |
CN110675632B (en) * | 2019-11-11 | 2021-11-30 | 重庆邮电大学 | Vehicle short-time trajectory prediction control method aiming at multi-feature space and data sparseness |
CN111292533B (en) * | 2020-02-11 | 2021-07-30 | 北京交通大学 | Method for estimating flow of arbitrary section of highway at any time period based on multi-source data |
TWI775123B (en) * | 2020-07-29 | 2022-08-21 | 財團法人資訊工業策進會 | Traffic condition prediction system and traffic condition prediction method |
CN112614336B (en) * | 2020-11-19 | 2021-12-07 | 南京师范大学 | Traffic flow modal fitting method based on quantum random walk |
CN115083162B (en) * | 2022-06-15 | 2023-11-17 | 北京三快在线科技有限公司 | Road condition prediction method, device, equipment and storage medium |
CN117994987B (en) * | 2024-04-07 | 2024-06-11 | 东南大学 | Traffic parameter extraction method and related device based on target detection technology |
Family Cites Families (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2005038381A (en) * | 2003-06-30 | 2005-02-10 | Toshiba Corp | Data analyzing apparatus, data analyzing program, and mobile terminal |
JP2008097251A (en) * | 2006-10-11 | 2008-04-24 | Toyota Central R&D Labs Inc | Congestion prediction apparatus, and driving support apparatus and support system |
CN101334845B (en) * | 2007-06-27 | 2010-12-22 | 中国科学院自动化研究所 | Video frequency behaviors recognition method based on track sequence analysis and rule induction |
JP2009245297A (en) * | 2008-03-31 | 2009-10-22 | Kddi Corp | Probe data collection system, probe data collection method, and program |
JP2010061335A (en) * | 2008-09-03 | 2010-03-18 | Kyosan Electric Mfg Co Ltd | Traffic information providing system and local traffic information providing method |
CN103456192B (en) * | 2013-09-01 | 2015-07-15 | 中国民航大学 | Terminal area prevailing traffic flow recognizing method based on track spectral clusters |
CN103714555B (en) * | 2013-12-13 | 2017-06-13 | 中国科学院深圳先进技术研究院 | A kind of four-dimensional motion point cloud segmentation and method for reconstructing based on movement locus |
CN103903019B (en) * | 2014-04-11 | 2017-12-15 | 北京工业大学 | A kind of automatic generation method of multilane track of vehicle space-time diagram |
-
2014
- 2014-09-28 CN CN201410505847.5A patent/CN104504897B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN104504897A (en) | 2015-04-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104504897B (en) | A kind of analysis of intersection traffic properties of flow and vehicle movement Forecasting Methodology based on track data | |
CN111310583B (en) | Vehicle abnormal behavior identification method based on improved long-term and short-term memory network | |
Kang et al. | Deep architecture for high-speed railway insulator surface defect detection: Denoising autoencoder with multitask learning | |
CN107133974B (en) | Gaussian Background models the vehicle type classification method combined with Recognition with Recurrent Neural Network | |
Fu et al. | Similarity based vehicle trajectory clustering and anomaly detection | |
CN105184271A (en) | Automatic vehicle detection method based on deep learning | |
Qi et al. | Efficient railway tracks detection and turnouts recognition method using HOG features | |
Wu et al. | Detection of bird nests in overhead catenary system images for high-speed rail | |
CN107292870A (en) | Track plug pin fault detection method and system based on image alignment with detection network model | |
CN105424330A (en) | Lorry shaft end bolt fault detecting method and lorry shaft end bolt fault detecting device | |
CN104020751A (en) | Campus safety monitoring system and method based on Internet of Things | |
CN111127499A (en) | Security inspection image cutter detection segmentation method based on semantic contour information | |
CN105513349A (en) | Double-perspective learning-based mountainous area highway vehicle event detection method | |
CN107730889B (en) | Target vehicle retrieval method based on traffic video | |
CN102254394A (en) | Antitheft monitoring method for poles and towers in power transmission line based on video difference analysis | |
Babaei | Vehicles tracking and classification using traffic zones in a hybrid scheme for intersection traffic management by smart cameras | |
Katsamenis et al. | A few-shot attention recurrent residual U-Net for crack segmentation | |
Ren et al. | Automatic measurement of traffic state parameters based on computer vision for intelligent transportation surveillance | |
Dagvasumberel et al. | Railroad near-miss occurrence detection and risk estimation system with data from camera using deep learning | |
Ma et al. | Vessel motion pattern recognition based on one-way distance and spectral clustering algorithm | |
CN107240111B (en) | Edge communication segmentation passenger flow statistical method | |
Batapati et al. | Video analysis for traffic anomaly detection using support vector machines | |
Yang et al. | Defect detection of axle box cover device fixing bolts in metro based on convolutional neural network | |
Ou et al. | [Retracted] Analysis of People Flow Image Detection System Based on Computer Vision Sensor | |
Kinattukara et al. | Clustering based neural network approach for classification of road images |
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 | ||
TR01 | Transfer of patent right |
Effective date of registration: 20201012 Address after: Room 265, 2 / F, office building, no.6, lianguizhong Road, Lianhe Industrial Zone, Shishan town, Nanhai District, Foshan City, Guangdong Province 528200 Patentee after: Zhongtian Sichuang information technology (Guangdong) Co., Ltd Address before: 100124 Chaoyang District, Beijing Ping Park, No. 100 Patentee before: Beijing University of Technology |
|
TR01 | Transfer of patent right |