CN107705563A - Continuous vehicle speed detection method based on laser radar - Google Patents
Continuous vehicle speed detection method based on laser radar Download PDFInfo
- Publication number
- CN107705563A CN107705563A CN201711219157.3A CN201711219157A CN107705563A CN 107705563 A CN107705563 A CN 107705563A CN 201711219157 A CN201711219157 A CN 201711219157A CN 107705563 A CN107705563 A CN 107705563A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- vehicle
- index
- msup
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- 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
- G08G1/0108—Measuring and analyzing of parameters relative to traffic conditions based on the source of data
- G08G1/0116—Measuring and analyzing of parameters relative to traffic conditions based on the source of data from roadside infrastructure, e.g. beacons
-
- 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
- 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
- G08G1/04—Detecting movement of traffic to be counted or controlled using optical or ultrasonic detectors
-
- 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
- G08G1/052—Detecting movement of traffic to be counted or controlled with provision for determining speed or overspeed
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Traffic Control Systems (AREA)
- Optical Radar Systems And Details Thereof (AREA)
Abstract
Continuous vehicle speed detection method based on laser radar, is related to laser radar velocity measuring technique, and vehicle speed detection technical field.Solve the problems, such as it is existing using laser radar realize present in the technology of vehicle speed detection using 2D laser radars detection exist the object of vehicle and non-vehicle can not be classified, and using high 64 line of cost 3D laser thunder costs it is too high the problem of.The present invention uses the roadway scene data of 16 line laser radar continuous acquisition runways, then realized using the associated method of the vehicle in more frame data and vehicle is associated, and then realize the detection to car speed, it completely avoid due to the problem of laser radar frame losing influences to measure, it also avoid due to the problem of vehicle mutually blocks influence measurement.The present invention realizes the measurement of vehicle continuous velocity change curve, significant to further analysis vehicle behavior.Present invention can apply to the place of the various measurement speeds in traffic engineering.
Description
Technical field
The present invention relates to laser radar velocity measuring technique, and vehicle speed detection technical field.
Background technology
Shown according to World Health Organization's WHO2010 statistics, the annual whole world is due to the people of Prevention of Road Traffic Fatalities
Number is about 1,250,000 people, has 2000 to 50,000,000 people injured.Traffic safety problem getting worse, and accurately extract road letter
It is obvious that breath acts on the date to road improvement traffic safety.
Important component of the vehicle as road information, accurately extracts vehicle characteristics, is solving road traffic safety
The premise of problem.In recent years, with the development of laser radar technique, new means are provided for vehicle detection.Pass through laser thunder
Up to 3D point cloud data, road traffic three-dimensional scenic can be obtained, object is participated in traffic and carries out accurately detection and classification.
The detection method of Current vehicle mainly have video detection, Data mining, microwave radar detection, infrared detection,
Magnetic Sensor detection etc..Wherein, video detection mode is extensive for commercial applications, and the detection range of video detection mode is big, energy
Enough detect various traffic events etc..But because video detection is very sensitive to light, change for shade and the unexpected of visible ray
Become, video detection error can be caused to increase, and video detection mode lacks range information.By contrast, laser radar conduct
A kind of active sensor, the information little interference by environment of return, precision are high, are easy to get to the depth information of object, not by visible
The interference of light.These advantages cause the application of laser radar to become more and more extensive, are provided to solve traffic engineering relevant issues
Effective means.With the development of laser radar manufacturing technology, its cost gradually lowers, and is provided for large-scale application
May.
Currently, many work on road traffic analysis carry out, wherein main problem is that being transported on road
The detection of moving-target and trajectory track.At present, path locus tracking uses different sensing modes, there is different detection methods.
Wherein, what is mainly used at present has the road vehicle detection method of view-based access control model and the road vehicle detection method based on distance,
Wherein, the road vehicle detection method of view-based access control model is easily by the interference of visible ray and disturbing for shade, the different shape of vehicle
Shape, size and color, and the direction of motion that vehicle is different on multilane so that vehicle attitude estimation, which becomes to have very much, chooses
War property；Road vehicle detection and trajectory track based on distance, it is different according to the sensor of selection, such as laser radar, microwave thunder
Up to etc..
For laser radar dynamic detection scene, car tracing is carried out using SLAMMOT algorithms, earliest work can chase after
Trace back by 2007.For Continuous Traffic data collection task, ITS seminar of University of Minnesota develops one and is based on laser thunder
Up to the traffic information acquisition system of network, for gathering traffic intersection road information.Detected using laser radar and carry out vehicle inspection
Survey and trajectory track, using 2D laser radars, vehicle detection and tracking are carried out in a horizontal plane.Because 2D laser radars detect
Mode limits, detection angles above water or following object information can not all obtain, cannot get object height
Information, therefore the object of vehicle and non-vehicle can not be classified.
Vehicle detection is carried out using 3D laser radars at present, is concentrated mainly on robot field, studies automatic driving vehicle
The vehicle behavioral problem on periphery.These research work, 64 line laser radar Velodyne HDL-64E LIDAR are all based on, are
Vehicle detection, the simply vehicle row in the smaller range of automatic driving vehicle periphery of research are carried out in the environment of a dynamic
For.Because 64 line laser radar costs are higher, it is impossible to carry out engineer applied on a large scale.
The content of the invention
The present invention is solved existing realized using laser radar and 2D laser is used present in the technology of vehicle speed detection
Be present the problem of can not classifying to the object of vehicle and non-vehicle in detections of radar, and swashed using the 3D of high 64 line of cost
The problem of light thunder cost is too high.
Continuous vehicle speed detection method of the present invention based on laser radar is continuously adopted using 16 line laser radars
Collect the roadway scene data of runway, the average speed of vehicle in current frame data, tool are then obtained according to the K frame data of collection
Body process is：
1), calculated according to adjacent two frame data and obtain the distance between adjacent all vehicles of frame data matrix：
Wherein, there is m in kth frame datakCar, SijRepresent class center and the frame of kth -1 of i-th car in kth frame data
The distance between class center of jth car in data, i ∈ (0, mk], j ∈ (0, mk-1],
Wherein,The abscissa at the class center of i-th car in kth frame data is represented,Represent in the frame data of kth -1
The abscissa at the class center of jth car,The ordinate at the class center of i-th car in kth frame data is represented,Represent the
The ordinate at the class center of jth car in k-1 frame data；
K=2,3L K, obtain the distance between multiple adjacent all vehicles of frame data matrix；
2), the vehicle incidence matrix obtained according to above-mentioned distance matrix：
The car of adjacent two frame data according to corresponding to obtaining the distance matrix of the vehicle of all adjacent two frame data of acquisition
Associated data, the vehicle associated data for collecting all adjacent two frame data obtain vehicle incidence matrix and are：
Wherein, akjThe vehicle class-mark being successfully associated in the frame data of kth -1 with jth in kth frame data is represented, J represents K frames
Containing the vehicle number in the most data frame of vehicle in data, j ∈ (0, J]；As the vehicle number m in kth frame datakDuring less than J,
Then have：It is 0；
3) all vehicles in kth frame data, are traveled through, obtain the vehicle set P={ V having detected thatId=1,VId=2,L,
VId=count, detailed process is：
Judge a corresponding to i-th carkiWhether it is 0,
If akiFor 0, then whether the attribute id for judging i-th car is 0, if 0, for the newly-built vehicle id of i-th car,
Id=count+1, while count=count+1 is updated, and create VId=countObject, and the object is added into and had detected that
Vehicle set P={ VId=1,VId=2,L,VId=countIn；Otherwise, the attribute id of i-th car is constant；
If akiBe not 0, then it is i-th car is associated with the vehicle in the vehicle set having detected that, if be successfully associated,
Then the id of vehicles of the id of this car with being successfully associated is corresponding, and is added into the vehicle set P ' being successfully associated, if
Association failure, then for i-th car newly-built vehicle id, the id=count+1, while update count=count+1, and create
Build object VId=count, and the object is added into the vehicle set P={ V having detected thatId=1,VId=2,L,VId=countIn；
Travel through in kth frame data after all vehicles, obtain the vehicle set P={ V detectedId=1,VId=2,L,
VId=count, and the vehicle set P '={ V ' being successfully associatedId=1,V′Id=2,L,V′Id=count ', count '≤count；
4) average speed of its vehicle, is calculated for each car in the above-mentioned vehicle set being successfully associated
Wherein, kindex′,sRepresent the sequence number for the data frame that the vehicle that vehicle id is index ' occurs first, kindex′,eRepresent
The sequence number for the data frame that vehicle id is occurred by index ' vehicle last time,It is index's ' to represent vehicle id
The characteristic point of vehicle is in kthindex′,eX-axis coordinate in frame data,Represent the feature for the vehicle that vehicle id is index '
Point is in kthindex′,sX-axis coordinate in frame data,Represent the characteristic point of the vehicle that vehicle id is index ' the
kindex′,eY-axis coordinate in frame data,Represent the characteristic point for the vehicle that vehicle id is index ' in kthindex′,sFrame number
According to middle y-axis coordinate, tindex′,sFor kthindex′,sThe timestamp of frame data frame, tindex′,eFor kthindex′,eThe time of frame data frame
Stamp.
The present invention realizes the detection of continuous car speed using 16 line laser radars, and then to realize that track of vehicle is followed the trail of
Provide effective means.
16 line laser radar cost of the present invention is relatively low, and with abundant 3D point cloud information.
During existing use consecutive frame DATA REASONING car speed, when the vehicle fleet size in adjacent two frame data is not communicated with,
Tachometric survey can be influenceed, such as：Have m chassis in current frame data, and have n chassis in previous frame data, then according to m with
N magnitude relationship, there are three kinds of situations：
Situation 1：When present frame vehicle number m is less than former frame vehicle number n, then illustrate now there is to leave a detection zone
Or disappeared due to reasons such as occlusions.Vehicle so now only less than or equal to present frame vehicle number can be associated to
Work(, former frame is remaining can only not to be associated by the vehicle that present frame associates by follow-up frame.
Situation 2：When present frame vehicle number m is equal to former frame vehicle number n, now two frame vehicle most probables are all interrelated
Success.Situations such as quantity of disappearance vehicle is with entering vehicle fleet size phase is also likely to be present, the car that now only consecutive frame all occurs
It can just be successfully associated.
Situation 3：When present frame vehicle number m is more than former frame vehicle number n, then illustrate now to have a car newly enter or
The vehicle to disappear before occurs.So now present frame has the situation that association failure occurs in vehicle,
Analyzed based on more than, due to occlusion and detecting distance farther out, cause that vehicle can be random in a certain frame number
According to the situation of middle loss, can cause to obtain vehicle association failure using adjacent frame data, and then velocity measuring can not be realized.
The present invention uses the vehicle correlating method based on more frame data, and then overcomes and lacked using existing for adjacent frame data
Fall into, the influence due to frame losing to measurement can be avoided completely, also overcome completely because vehicle mutually blocks the influence to measurement.
In traffic engineering, the continuous velocity change curve of vehicle is significant, of the invention from laser radar 3D points
Cloud data are set out, and by vehicle detection with associating, vehicle continuous velocity changing rule have been obtained, to further analysis vehicle behavior
It is significant.
Brief description of the drawings
Fig. 1 is two-way four that the Velodyne LiDAR VLP-16 laser radars described in embodiment one detect
The scene in track.
Fig. 2 is the box model (Box model of a vehicle) of the chassis described in embodiment.
Fig. 3 is True Data model of the foundation box model according to detections of radar one car of acquisition described in embodiment
(Model of a real vehicle)。
Fig. 4 is that the vehicle described in embodiment is located at vehicle form detected by laser radar diverse location
(Different status of a vehicle by relative posittion compared to the LIDAR)。
Fig. 5 to 8 is for the individually rate curve analysis of vehicle sample in four tracks respectively described in test experiments one
Result schematic diagram.
Fig. 9 is result of the vehicle of the first lane shown in Fig. 5 after 237 frame original images and grid filtering.
Figure 10 is that the true tailstock point algorithm of the vehicle found according to 0 ° of position of laser radar described in test experiments two obtains
The vehicle tail point P obtained position view.
Figure 11 is the revised result schematic diagram of the car speed positioned at first lane described in test experiments two.
Figure 12 is the revised result schematic diagram of the car speed positioned at second lane described in test experiments two.
Figure 13 is described in test experiments four, and 1-2500 frame data represent the detecting distance that vehicle has just enter into and finally left
Test result, the evaluation index 2-3 of the result is shown in Figure 14.
Figure 15 is described in test experiments four, 2500-5456 frame data represent vehicle have just enter into and finally leave detection away from
From test result, evaluation of result index 2-3 is shown in Figure 16.Wherein, the Left first detection thunders in figure
Vehicle is detected first up to left side, and Right first detection represent to detect vehicle first on the right side of radar.
Embodiment
Embodiment one：The continuous vehicle speed detection method based on laser radar described in present embodiment uses 16 lines
The roadway scene data of laser radar continuous acquisition runway, then obtain car in current frame data according to the K frame data of collection
Average speed, detailed process is：
1), calculated according to adjacent two frame data and obtain the distance between adjacent all vehicles of frame data matrix：
Wherein, there is m in kth frame datakCar, SijRepresent class center and the frame of kth -1 of i-th car in kth frame data
The distance between class center of jth car in data, i ∈ (0, mk], j ∈ (0, mk-1],
Wherein,The abscissa at the class center of i-th car in kth frame data is represented,Represent in the frame data of kth -1
The abscissa at the class center of jth car,The ordinate at the class center of i-th car in kth frame data is represented,Represent the
The ordinate at the class center of jth car in k-1 frame data；In the association of consecutive frame vehicle, a car is in 1 frame along x-axis and along y-axis
The distance of motion is all the section of a closure, it is impossible to is undergone mutation；
K=2,3L K, obtain the distance between multiple adjacent all vehicles of frame data matrix；
2), the vehicle incidence matrix obtained according to above-mentioned distance matrix：
According to a car in adjacent data frames along x-axis and the distance moved along y-axis be all one closure section, can not
The principle that can be undergone mutation, the vehicle in adjacent data frames is associated, obtains the vehicle associated data of adjacent two frame data,
Collecting the formation vehicle incidence matrix of the vehicle associated data in all adjacent two frame data is：
Wherein, akjThe vehicle class-mark being successfully associated in the frame data of kth -1 with jth in kth frame data is represented, J represents K frames
Containing the vehicle number in the most data frame of vehicle in data, j ∈ (0, J]；As the vehicle number m in kth frame datakDuring less than J,
Then have：It is 0；
3) all vehicles in kth frame data, are traveled through, obtain the vehicle set P={ V having detected thatId=1,VId=2,L,
VId=count, detailed process is：
Judge a corresponding to i-th carkiWhether it is 0,
If akiFor 0, then whether the attribute id for judging i-th car is 0, if 0, represents the vehicle to enter detection for the first time
Region, for i-th car newly-built vehicle id, the id=count+1, while count=count+1 is updated, and create object
VId=count, and the object is added into the vehicle set P={ V having detected thatId=1,VId=2,L,VId=countIn；Otherwise, it is described
The attribute id of i-th car is constant；
If akiBe not 0, then it is i-th car is associated with the vehicle in the vehicle set having detected that, if be successfully associated,
Then illustrate to occur before the vehicle in the detection area, the id of vehicles of the id of this car with being successfully associated is corresponding, and will
It is added in the vehicle set P ' being successfully associated, if association failure, it is to first enter into detection zone to illustrate the car, for institute
I-th car newly-built vehicle id, id=count+1 are stated, while updates count=count+1, and creates object VId=count, and will
The object is added into the vehicle set P={ V having detected thatId=1,VId=2,L,VId=countIn；
Travel through in kth frame data after all vehicles, obtain the vehicle set P={ V detectedId=1,VId=2,L,
VId=count, and the vehicle set P '={ V ' being successfully associatedId=1,V′Id=2,L,V′Id=count ', count '≤count；
4) average speed at its current time, is calculated for each car in the above-mentioned vehicle set being successfully associated：
Wherein, kindex′,sRepresent the sequence number for the data frame that the vehicle that vehicle id is index ' occurs first, kindex′,eRepresent
The sequence number for the data frame that vehicle id is occurred by index ' vehicle last time,It is index's ' to represent vehicle id
The characteristic point of vehicle is in kthindex′,eX-axis coordinate in frame data,Represent the feature for the vehicle that vehicle id is index '
Point is in kthindex′,sX-axis coordinate in frame data,Represent the characteristic point of the vehicle that vehicle id is index ' the
kindex′,eY-axis coordinate in frame data,Represent the characteristic point for the vehicle that vehicle id is index ' in kthindex′,sFrame number
According to middle y-axis coordinate, tindex′,sFor kthindex′,sThe timestamp of frame data frame, tindex′,eFor kthindex′,eThe time of frame data frame
Stamp.In the case where laser radar detects vehicle context, because same car is different in the class-mark that different pieces of information frame clusters, so having to
Vehicle is associated, so can just obtain speed of the same car in different pieces of information frame.For glitch-free carry out special bus
Detection, laser radar is placed in roadside.Due to the straightline propagation of laser, there is 1 frame or multiframe quilt in the vehicle in remote track
The situation of nearly track occlusion.
Under such detection mode, vehicle association needs solve problems with：
1) because vehicle blocks completely, cause the vehicle on former track to be wholly absent, cause adjacent two frames association failure；
2) because vehicle sections block, cause the vehicle form detected by former track irregular so that vehicle geometry is special
Levy disunity；
3) for compact car traveling in track farther out, laser radar detects that available point is less, causes corresponding vehicle face shaping
With real vehicles poor appearance away from larger.
For the presence of above-mentioned realistic problem, the multiframe vehicle that present embodiment is taken based on based on consecutive frame association closes
Linked method, completely solve above three problem.
Above-mentioned steps 2) described in the methods of vehicle associated data of adjacent two frame data of acquisition be：
The distance matrix obtained according to step 1), often capable minimum range S is chosen successivelyia, the minimum range is the frame of kth -1
The distance between i-th car in a car and kth frame data in data, then determines whether following two conditions：
Condition 1：Sia＜=Tmax_xy_move, wherein Tmax_xy_moveIt is the maximum that a car can travel between consecutive frame
Distance, the distance are decided by the maximal rate that the section limits；
Condition 2：yo(i)-yo(a) ＜=Tmax_y_move, Tmax_y_moveRepresent that the longitudinal axis of the car between consecutive frame is maximum
Offset, the offset represent the offset of vehicle and travel direction, actually represent the lane information of the car, select
Tmax_y_moveThe advantages of parameter：The vehicle match in other tracks can be prevented to current lane, be also possible to prevent the vehicle of lane change
Association failure；
If meet above-mentioned two condition simultaneously, it is believed that be successfully associated.
Above-mentioned steps 3) in, it is by i-th car method associated with the vehicle in the vehicle set having detected that：
Rule of judgment 1：Whether associated vehicle is near track where vehicle to be associated or its, i.e. works as satisfaction
Condition
When, then it is assumed that associated vehicle track where vehicle to be associated or near, wherein, yO,kiRepresent to be associated
The class center ordinate of vehicle, the vehicle to be associated are i-th car in kth frame data；For associated vehicle
Class center ordinate in the data frame that its last time occurs, the associated vehicle are the vehicle set having detected that
Middle vehicle id be index vehicle, Tmax_y_moveShow longitudinal axis maximum offset of the car between consecutive frame；
Rule of judgment 2：Whether associated vehicle is in nearer data frame before the data frame that vehicle to be associated occurs
Occur, when meeting condition
When, then it is assumed that occur in associated vehicle data frame nearer before the data frame that vehicle to be associated occurs
, wherein, KkiFor a sequence number for a data frame where waiting to cut-off,Represent the number that associated vehicle last time is appeared in
According to the sequence number of frame, Tmax_frame_loseRepresent the maximum loss frame number allowed；
Rule of judgment 3：Whether associated vehicle occurs occurring before position in vehicle to be associated, obtains vehicle fortune first
Dynamic direction：
Wherein, v_direction represents direction of vehicle movement, represents that direction of vehicle movement is along horizontal stroke when its value is+1
Axle (x-axis) positive direction, represent that direction of vehicle movement is along transverse axis (x-axis) negative direction, T when its value is -1v_change_laneTable
Show that direction of vehicle movement starts the track distance changed, yO,kiThe class center ordinate of vehicle to be associated is represented, then car to be associated
With associated vehicle location relation should meet condition：
Wherein, xO,kiThe abscissa at the class center of vehicle to be associated is represented,Associated vehicle occurs in last time
Data frame in its class center abscissa；
Meet above three condition simultaneously, then preliminary judgement vehicle is successfully associated；
Then, exist for each associated vehicle that above-mentioned preliminary judgement is successfully associated with vehicle to be associated, both reckonings
The range difference of the location of present frame, then select the associated vehicle corresponding to the minimum value in the range difference and closed with waiting
Connection vehicle is successfully associated.
Above-mentioned reckoning, which obtains, calculates that the two is in the process of the range difference of the location of present frame：
According to the average speed v of associated vehicleindex′Estimate that the vehicle is in the position of present frame：
Sindex′=vindex′(tki-tindex′,e),
K be present frame data frame number, tkiFor the timestamp of present frame, tindex′,eFor associated vehicle last time
The timestamp of the data frame of appearance；
The distance between vehicle to be associated and associated vehicle are：
Wherein, xForE,kiRepresent the x-axis coordinate of the assemblage characteristic point of vehicle to be associated in current frame data, yForE,kiRepresent
The y-axis coordinate of the assemblage characteristic point of vehicle to be associated in current frame data；
Obtain the range difference of associated vehicle and vehicle to be associated in the location of present frame | Sindex′,ki-Sindex′|。
What the associated vehicle corresponding to the above-mentioned minimum value selected in the range difference was successfully associated with vehicle to be associated
Process is：
Take the minimum value in the difference of the distance of vehicle to be associated and all associated vehicles：
ΔSmin_index′,ki=arg min | Sindex′,ki-Sindex′|
The vehicle id that min_index ' is in all associated vehicles and the difference of the distance of vehicle to be associated is minimum, when described
Minimum value meets condition Δ Smin_index′,ki≤Tmin_ΔSWhen, then judge to be successfully associated, the id of vehicle to be associated is entered as min_
Index ', Tmin_ΔSRepresent the error amount that range estimation allows.
It is that the vehicle-relevant data in the data for obtaining laser radar uses when vehicle distances are calculated in present embodiment
The mode of vehicle box model is expressed, shown in Figure 2, and A, B, C, D in the vehicle box model represent the top of vehicle respectively
Point, O points are the central points of vehicle, are the class centers of the characteristic point, also known as vehicle of vehicle.L in figure1、l2、l3、l4Respectively
Represent corresponding at the distance between 2 points, v represents the speed that vehicle is advanced, (xcenter, ycenter) represent vehicle center point O points seat
Mark, (xmin, ymin) represent vehicle summit A coordinate, (xmax, ymin) represent vehicle summit B coordinate.E points and F points are vehicle box
Two characteristic points of son, are the real points on the vehicle detected by laser radar, with timestamp information, E points and F points lead to
Crossing N number of point that chosen distance A, B is nearest respectively takes its x-axis coordinate and being averaged for y-axis coordinate to be worth to, and its timestamp information is to take
The average value of the timestamp of closest N number of point.Fig. 3 is the vehicle box model of the real vehicles of laser radar collection.
Vehicle box model also has assemblage characteristic point ForE, and it is to vehicle according to the relative position of vehicle and laser radar
E points and F points be combined acquisition.Because vehicle is different from the relative position of laser radar, cause to detect the form of vehicle
It is different.It is shown in Figure 4：
S1 represents vehicle on the laser radar left side, and vehicle only has A, B and C point to be detected；
S2 represents vehicle face laser radar laser radar, and only A and B points are detected；
S3 represents vehicle on the right of laser radar, and vehicle only has A, B and D point to be detected.And with the movement of vehicle, inspection
The AB sections measured can not represent real vehicle commander.
To sum up, in order to prevent the characteristic point due to selection from can not correspond to the same position of vehicle, cause to calculate distance
There is larger deviation so that car speed calculates inaccuracy, should be dynamically selected spy according to the position of vehicle and laser radar
Levy point, that is, assemblage characteristic point ForE.
Test experiments one：Analyzed for the independent vehicle sample rate curve in four tracks：
The data frame that continues through of an independent car in four tracks is chosen respectively, referring to shown in Fig. 5-Fig. 8, being position respectively
In the rate curve of the independent vehicle on first lane, second lane, third lane and Four-Lane Road.For the car in each track
, measure car speed with characteristic point O points, E points, F points and assemblage characteristic point ForE respectively.See on the whole, it is special using combination
The rate curve that point ForE detection speeds obtain is levied, no matter has just been initially entered in vehicle or vehicle leaves detection zone, obtained
Rate curve compared with using other three feature lists detection obtain rate curve it is all more steady.
Test experiments two, first lane and second lane error analysis：
Result of the first lane vehicle after the original image of 237 frames filters with grid is selected, as shown in Figure 9.#237
In 0 ° of position of laser radar.It is the beginning and end of the frame data of laser radar one in the position.So cause to detect
Vehicle minimum point, the point for the vehicle that actually #236 is scanned in a short period of time, be not real tailstock point.
And real tailstock point is actually the minimum point in min position.Therefore, the speed now calculated is laser radar at this
The linear velocity of position, rather than the movement velocity of vehicle.
The real tailstock point of vehicle is found according to 0 ° of position of laser radar, arthmetic statement is as follows：
R1:Judge whether to be in laser radar face position
R2:The point set C according to where laser radar data horizontal angle (azimuth) finds true tailstock point
R3:X_min_revise is looked in C point sets
R4:This point nearest N=5 point nearby is found, calculates x, y's, t is averaged, as the true tailstock point x_min_ of vehicle
revise_ave
R5:In long side the spot projection to car, i.e., (x_min_revise_ave, y_min)
R6:Speed is calculated using the point
The true tailstock point arithmetic result of vehicle found according to 0 ° of position of laser radar is as shown in Figure 10, and P points are true car
Tail point.Immediately there is minimum point to vehicle in a frame speed after 0 ° of face position, and the vehicle in front that reason is is in 0 ° of face
The vehicle tailstock point that position is got is untrue, after being modified to real tailstock point, then can cause the time difference mistake with its a later frame
It is small, it may appear that and the same problem that upper surface error occurs.Treating method now is will to skip 0 ° of face position of laser radar, profit
Car speed is calculated with two frames before and after the frame (time interval of this two frame is about 0.1s or so).
Shown in the revised vehicle speed curve Figure 11 being located at positioned at first lane of speed, and the car positioned at second lane
Rate curve is shown in Figure 12, and after speed amendment, the speed of originally abnormal frame is corrected, and rate curve becomes flat
Surely.
Test experiments three：Vehicle continuous velocity is analyzed and error analysis.
Obtain vehicle continuous velocity it is critical that the accuracy that vehicle detection associates with vehicle.Doing Algorithm Analysis
When, for 1-2500 frames when algorithm design is carried out, algorithm wants high point to the degree that it is fitted；Then 2500-5456 conducts
Checking collection, for testing algorithm service behaviour.When associating analysis of the accuracy to vehicle, following evaluation index is selected to carry out performance
Assess：
1) vehicle fleet size Vehicle_count.The quantity that the actual vehicle of sample set is passed through is certain, if algorithm passes through
Quantity and physical presence difference after vehicle association, then illustrate that vehicle association has error.
2) maximum detecting distance DR.The position that vehicle enters and leaves sample set in the sample set is more fixed, if vehicle
It is successfully associated, when vehicle enters detection zone from Far Left, detection zone is left from rightmost along x-axis positive movement, vehicle.Instead
It, vehicle is along x-axis negative movement, then conversely., can be as one of standard whether detection vehicle is successfully associated according to this.
3) the frame number Frames_num that vehicle is detected.Vehicle is from most initially entering detection zone, to leaving detection zone
Domain, the frame number passed through are an amounts with velocity correlation.When speed difference scope is little, then vehicle is detected frame number
And in the section of some fixation.
Tested on two separated data sets, vehicle fleet size result accuracy rate is more as shown in table 1.1-2500 frames
The test result for the detecting distance that vehicle has just been initially entered and finally left is shown in Figure 13, its evaluation index 2-3 result
As shown in figure 14.2500-5456 frames vehicle has just initially entered and as shown in figure 15, the evaluation index 2-3 that finally leaves detecting distance
Result it is as shown in figure 16.
In following table：Real represent vehicle traveling actual distance, Detection represent using this method measurement obtain away from
From Error Ratio represent error, and Accuracy represents precision.
Table 1
Claims (7)
1. the continuous vehicle speed detection method based on laser radar, it is characterised in that methods described uses 16 line laser radars
The roadway scene data of continuous acquisition runway, then obtain vehicle in current frame data according to the K frame data of collection and be averaged
Speed, detailed process are：
1), calculated according to adjacent two frame data and obtain the distance between adjacent all vehicles of frame data matrix：
<mrow>
<msub>
<mi>S</mi>
<mrow>
<msub>
<mi>m</mi>
<mi>k</mi>
</msub>
<mo>&times;</mo>
<msub>
<mi>m</mi>
<mrow>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
</mrow>
</msub>
<mo>=</mo>
<mfenced open = "{" close = "}">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>S</mi>
<mn>11</mn>
</msub>
<mo>,</mo>
<msub>
<mi>S</mi>
<mn>12</mn>
</msub>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<msub>
<mi>S</mi>
<mrow>
<mn>1</mn>
<msub>
<mi>m</mi>
<mrow>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>S</mi>
<mn>21</mn>
</msub>
<mo>,</mo>
<msub>
<mi>S</mi>
<mn>22</mn>
</msub>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<msub>
<mi>S</mi>
<mrow>
<mn>2</mn>
<msub>
<mi>m</mi>
<mrow>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>M</mi>
<mo>,</mo>
<mi>M</mi>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<mi>M</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>S</mi>
<mrow>
<msub>
<mi>m</mi>
<mi>k</mi>
</msub>
<mn>1</mn>
</mrow>
</msub>
<mo>,</mo>
<msub>
<mi>S</mi>
<mrow>
<msub>
<mi>m</mi>
<mi>k</mi>
</msub>
<mn>2</mn>
</mrow>
</msub>
<mo>,</mo>
<mo>...</mo>
<mo>,</mo>
<msub>
<mi>S</mi>
<mrow>
<msub>
<mi>m</mi>
<mi>k</mi>
</msub>
<msub>
<mi>m</mi>
<mrow>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>,</mo>
</mrow>
Wherein, there is m in kth frame datakCar, SijRepresent class center and the frame data of kth -1 of i-th car in kth frame data
In the distance between the class center of jth car, i ∈ (0, mk], j ∈ (0, mk-1],
<mrow>
<msub>
<mi>S</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<mo>=</mo>
<msqrt>
<mrow>
<msup>
<mrow>
<mo>(</mo>
<msub>
<msup>
<mi>x</mi>
<mi>i</mi>
</msup>
<mrow>
<mi>o</mi>
<mo>,</mo>
<mi>k</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<msup>
<mi>x</mi>
<mi>j</mi>
</msup>
<mrow>
<mi>o</mi>
<mo>,</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mrow>
<mo>(</mo>
<msub>
<msup>
<mi>y</mi>
<mi>i</mi>
</msup>
<mrow>
<mi>o</mi>
<mo>,</mo>
<mi>k</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<msup>
<mi>y</mi>
<mi>j</mi>
</msup>
<mrow>
<mi>o</mi>
<mo>,</mo>
<mi>k</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
<mo>,</mo>
</mrow>
Wherein,The abscissa at the class center of i-th car in kth frame data is represented,Represent jth in the frame data of kth -1
The abscissa at the class center of car,The ordinate at the class center of i-th car in kth frame data is represented,Represent the frame of kth -1
The ordinate at the class center of jth car in data；
K=2,3L K, obtain the distance between multiple adjacent all vehicles of frame data matrix；
2), the vehicle incidence matrix obtained according to above-mentioned distance matrix：
The vehicle of adjacent two frame data closes according to corresponding to obtaining the distance matrix of the vehicle of all adjacent two frame data of acquisition
Join data, the vehicle associated data acquisition vehicle incidence matrix for collecting all adjacent two frame data is：
<mrow>
<msub>
<mi>Match</mi>
<mrow>
<mi>K</mi>
<mo>&times;</mo>
<mi>J</mi>
</mrow>
</msub>
<mo>=</mo>
<msub>
<mi>a</mi>
<mrow>
<mi>k</mi>
<mi>j</mi>
</mrow>
</msub>
<mo>=</mo>
<mfenced open = "{" close = "}">
<mtable>
<mtr>
<mtd>
<mrow>
<mn>0</mn>
<mo>,</mo>
<mn>0</mn>
<mo>,</mo>
<mi>L</mi>
<mo>,</mo>
<mn>0</mn>
<mo>,</mo>
<mi>L</mi>
<mo>,</mo>
<mi>L</mi>
<mo>,</mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>a</mi>
<mn>21</mn>
</msub>
<mo>,</mo>
<msub>
<mi>a</mi>
<mn>22</mn>
</msub>
<mo>,</mo>
<mi>L</mi>
<mo>,</mo>
<msub>
<mi>a</mi>
<mrow>
<mn>2</mn>
<msub>
<mi>m</mi>
<mn>2</mn>
</msub>
</mrow>
</msub>
<mo>,</mo>
<mi>L</mi>
<mo>,</mo>
<mi>L</mi>
<mo>,</mo>
<msub>
<mi>a</mi>
<mrow>
<mn>2</mn>
<mi>J</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mi>M</mi>
<mi> </mi>
<mi>M</mi>
<mi> </mi>
<mi>M</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>a</mi>
<mrow>
<mi>K</mi>
<mn>1</mn>
</mrow>
</msub>
<mo>,</mo>
<msub>
<mi>a</mi>
<mrow>
<mi>K</mi>
<mn>2</mn>
</mrow>
</msub>
<mo>,</mo>
<mi>L</mi>
<mo>,</mo>
<mi>L</mi>
<mo>,</mo>
<msub>
<mi>a</mi>
<mrow>
<msub>
<mi>Km</mi>
<mi>K</mi>
</msub>
</mrow>
</msub>
<mo>,</mo>
<mi>L</mi>
<mo>,</mo>
<msub>
<mi>a</mi>
<mrow>
<mi>K</mi>
<mi>J</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mi>v</mi>
<mi>a</mi>
<mi>l</mi>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
<mi> </mi>
<mi>s</mi>
<mi>u</mi>
<mi>c</mi>
<mi>c</mi>
<mi>e</mi>
<mi>s</mi>
<mi>s</mi>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mn>0</mn>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>e</mi>
<mi>l</mi>
<mi>s</mi>
<mi>e</mi>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
<mo>,</mo>
</mrow>
Wherein, akjThe vehicle class-mark being successfully associated in the frame data of kth -1 with jth in kth frame data is represented, J represents K frame data
In containing the vehicle number in the most data frame of vehicle, j ∈ (0, J]；As the vehicle number m in kth frame datakDuring less than J, then
Have：It is 0；
3) all vehicles in kth frame data, are traveled through, obtain the vehicle set P={ V having detected thatId=1,VId=2,L,VId=count,
Detailed process is：
Judge a corresponding to i-th carkiWhether it is 0,
If akiFor 0, then whether the attribute id for judging i-th car is 0, if 0, for i-th car newly-built the vehicle id, id=
Count+1, while count=count+1 is updated, and create VId=countObject, and the object is added into the car having detected that
Set P={ VId=1,VId=2,L,VId=countIn；Otherwise, the attribute id of i-th car is constant；
If akiBe not 0, then it is i-th car is associated with the vehicle in the vehicle set having detected that, should if be successfully associated
The id of vehicles of the id of car with being successfully associated is corresponding, and is added into the vehicle set P ' being successfully associated, if association
Failure, then for i-th car newly-built vehicle id, the id=count+1, while update count=count+1, and establishment pair
As VId=count, and the object is added into the vehicle set P={ V having detected thatId=1,VId=2,L,VId=countIn；
Travel through in kth frame data after all vehicles, obtain the vehicle set P={ V detectedId=1,VId=2,L,VId=count,
And the vehicle set P '={ V ' being successfully associatedId=1,V′Id=2,L,V′Id=count ', count '≤count；
4) average speed of its vehicle, is calculated for each car in the above-mentioned vehicle set being successfully associated
<mrow>
<msub>
<mi>v</mi>
<mrow>
<msup>
<mi>index</mi>
<mo>&prime;</mo>
</msup>
</mrow>
</msub>
<mo>=</mo>
<mfrac>
<msqrt>
<mrow>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mrow>
<mi>F</mi>
<mi>o</mi>
<mi>r</mi>
<mi>E</mi>
<mo>,</mo>
<msub>
<mi>k</mi>
<mrow>
<msup>
<mi>index</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>e</mi>
</mrow>
</msub>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>x</mi>
<mrow>
<mi>F</mi>
<mi>o</mi>
<mi>r</mi>
<mi>E</mi>
<mo>,</mo>
<msub>
<mi>k</mi>
<mrow>
<msup>
<mi>index</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>s</mi>
</mrow>
</msub>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>y</mi>
<mrow>
<mi>F</mi>
<mi>o</mi>
<mi>r</mi>
<mi>E</mi>
<mo>,</mo>
<msub>
<mi>k</mi>
<mrow>
<msup>
<mi>index</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>e</mi>
</mrow>
</msub>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>y</mi>
<mrow>
<mi>F</mi>
<mi>o</mi>
<mi>r</mi>
<mi>E</mi>
<mo>,</mo>
<msub>
<mi>k</mi>
<mrow>
<msup>
<mi>index</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>s</mi>
</mrow>
</msub>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
<mrow>
<msub>
<mi>t</mi>
<mrow>
<msup>
<mi>index</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>e</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>t</mi>
<mrow>
<msup>
<mi>index</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>s</mi>
</mrow>
</msub>
</mrow>
</mfrac>
<mo>,</mo>
</mrow>
Wherein, kindex′,sRepresent the sequence number for the data frame that the vehicle that vehicle id is index ' occurs first, kindex′,eRepresent vehicle
The sequence number of the data frame occurred by index ' vehicle last time,Represent the vehicle that vehicle id is index '
Characteristic point ForE is in kthindex′,eX-axis coordinate in frame data,Represent the characteristic point for the vehicle that vehicle id is index '
ForE is in kthindex′,sX-axis coordinate in frame data,Represent that the characteristic point ForE for the vehicle that vehicle id is index ' exists
Kthindex′,eY-axis coordinate in frame data,Represent the characteristic point for the vehicle that vehicle id is index ' in kthindex′,sFrame
Y-axis coordinate in data, tindex′,sFor kthindex′,sThe timestamp of frame data frame, tindex′,eFor kthindex′,eFrame data frame when
Between stab.
2. the continuous vehicle speed detection method according to claim 1 based on laser radar, it is characterised in that step 2)
Described in the methods of vehicle associated data of adjacent two frame data of acquisition be：
The distance matrix obtained according to step 1), often capable minimum range S is chosen successivelyia, the minimum range is the frame data of kth -1
In a car and kth frame data in the distance between i-th car, then determine whether following two conditions：
Condition 1：Sia＜=Tmax_xy_move, wherein Tmax_xy_moveIt is the ultimate range that a car can travel between consecutive frame；
Condition 2：yo(i)-yo(a) ＜=Tmax_y_move, Tmax_y_moveRepresent longitudinal axis peak excursion of the car between consecutive frame
Amount, the offset represent the offset of vehicle and travel direction；
If meet above-mentioned two condition simultaneously, it is believed that be successfully associated.
3. the continuous vehicle speed detection method according to claim 1 based on laser radar, it is characterised in that step 3)
Described in, if akiIt is not 0, then is by i-th car method associated with the vehicle in the vehicle set having detected that：
Rule of judgment 1：Associated vehicle whether where the vehicle to be associated track or its near, i.e. when meeting condition
<mrow>
<msub>
<mi>y</mi>
<mrow>
<mi>O</mi>
<mo>,</mo>
<mi>k</mi>
<mi>i</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>y</mi>
<mrow>
<mi>O</mi>
<mo>,</mo>
<msub>
<mi>V</mi>
<mrow>
<mi>i</mi>
<mi>d</mi>
<mo>=</mo>
<mi>i</mi>
<mi>n</mi>
<mi>d</mi>
<mi>e</mi>
<mi>x</mi>
</mrow>
</msub>
</mrow>
</msub>
<mo>&le;</mo>
<msub>
<mi>T</mi>
<mrow>
<mi>m</mi>
<mi>a</mi>
<mi>x</mi>
<mo>_</mo>
<mi>y</mi>
<mo>_</mo>
<mi>m</mi>
<mi>o</mi>
<mi>v</mi>
<mi>e</mi>
</mrow>
</msub>
<mo>,</mo>
<mi>i</mi>
<mi>n</mi>
<mi>d</mi>
<mi>e</mi>
<mi>x</mi>
<mo>&Element;</mo>
<mo>&lsqb;</mo>
<mn>1</mn>
<mo>,</mo>
<mi>c</mi>
<mi>o</mi>
<mi>u</mi>
<mi>n</mi>
<mi>t</mi>
<mo>&rsqb;</mo>
</mrow>
When, then it is assumed that associated vehicle track where vehicle to be associated or near, wherein, yO,kiRepresent vehicle to be associated
Class center ordinate, the vehicle to be associated be kth frame data in i-th car；It is associated vehicle at it
Class center ordinate in the data frame that last time occurs, the associated vehicle is car in the vehicle set having detected that
Id is index vehicle, Tmax_y_moveShow longitudinal axis maximum offset of the car between consecutive frame；
Rule of judgment 2：Whether associated vehicle occurs in nearer data frame before the data frame that vehicle to be associated occurs,
When meeting condition
<mrow>
<mn>0</mn>
<mo><</mo>
<msub>
<mi>K</mi>
<mrow>
<mi>k</mi>
<mi>i</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>K</mi>
<msub>
<mi>V</mi>
<mrow>
<mi>i</mi>
<mi>d</mi>
<mo>=</mo>
<mi>i</mi>
<mi>n</mi>
<mi>d</mi>
<mi>e</mi>
<mi>x</mi>
</mrow>
</msub>
</msub>
<mo>&le;</mo>
<msub>
<mi>T</mi>
<mrow>
<mi>max</mi>
<mo>_</mo>
<mi>f</mi>
<mi>r</mi>
<mi>a</mi>
<mi>m</mi>
<mi>e</mi>
<mo>_</mo>
<mi>l</mi>
<mi>o</mi>
<mi>s</mi>
<mi>e</mi>
</mrow>
</msub>
<mo>,</mo>
<mi>i</mi>
<mi>n</mi>
<mi>d</mi>
<mi>e</mi>
<mi>x</mi>
<mo>&Element;</mo>
<mo>&lsqb;</mo>
<mn>1</mn>
<mo>,</mo>
<mi>c</mi>
<mi>o</mi>
<mi>u</mi>
<mi>n</mi>
<mi>t</mi>
<mo>&rsqb;</mo>
</mrow>
When, then it is assumed that occur in associated vehicle data frame nearer before the data frame that vehicle to be associated occurs, its
In, KkiA sequence number for place data frame of cut-offfing is waited,Represent the data frame that associated vehicle last time is appeared in
Sequence number, Tmax_frame_loseRepresent the maximum loss frame number allowed；
Rule of judgment 3：Whether associated vehicle occurs occurring before position in vehicle to be associated, obtains vehicle movement side first
To：
<mrow>
<mi>v</mi>
<mo>_</mo>
<mi>d</mi>
<mi>i</mi>
<mi>r</mi>
<mi>e</mi>
<mi>c</mi>
<mi>t</mi>
<mi>i</mi>
<mi>o</mi>
<mi>n</mi>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<mo>+</mo>
<mn>1</mn>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
<mi> </mi>
<msub>
<mi>y</mi>
<mrow>
<mi>O</mi>
<mo>,</mo>
<mi>k</mi>
<mi>i</mi>
</mrow>
</msub>
<mo>&le;</mo>
<msub>
<mi>T</mi>
<mrow>
<mi>v</mi>
<mo>_</mo>
<mi>c</mi>
<mi>h</mi>
<mi>a</mi>
<mi>n</mi>
<mi>g</mi>
<mi>e</mi>
<mo>_</mo>
<mi>l</mi>
<mi>a</mi>
<mi>n</mi>
<mi>e</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>-</mo>
<mn>1</mn>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
<mi> </mi>
<msub>
<mi>y</mi>
<mrow>
<mi>O</mi>
<mo>,</mo>
<mi>k</mi>
<mi>i</mi>
</mrow>
</msub>
<mo>></mo>
<msub>
<mi>T</mi>
<mrow>
<mi>v</mi>
<mo>_</mo>
<mi>c</mi>
<mi>h</mi>
<mi>a</mi>
<mi>n</mi>
<mi>g</mi>
<mi>e</mi>
<mo>_</mo>
<mi>l</mi>
<mi>a</mi>
<mi>n</mi>
<mi>e</mi>
</mrow>
</msub>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein, v_direction represent direction of vehicle movement, when its value be+1 expression direction of vehicle movement be along transverse axis just
Direction, represent that direction of vehicle movement is along transverse axis negative direction, T when its value is -1v_change_laneRepresent direction of vehicle movement
Start the track distance changed, yO,kiThe class center ordinate of vehicle to be associated is represented, then vehicle to be associated and associated vehicle
Position relationship should meet condition：
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msub>
<mi>x</mi>
<mrow>
<mi>O</mi>
<mo>,</mo>
<mi>k</mi>
<mi>i</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>x</mi>
<msub>
<mi>V</mi>
<mrow>
<mi>i</mi>
<mi>d</mi>
<mo>=</mo>
<mi>i</mi>
<mi>n</mi>
<mi>d</mi>
<mi>e</mi>
<mi>x</mi>
</mrow>
</msub>
</msub>
<mo>></mo>
<mn>0</mn>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
<mi> </mi>
<mi>v</mi>
<mo>_</mo>
<mi>d</mi>
<mi>i</mi>
<mi>r</mi>
<mi>e</mi>
<mi>c</mi>
<mi>t</mi>
<mi>i</mi>
<mi>o</mi>
<mi>n</mi>
<mo>></mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<msub>
<mi>x</mi>
<mrow>
<mi>O</mi>
<mo>,</mo>
<mi>k</mi>
<mi>i</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>x</mi>
<msub>
<mi>V</mi>
<mrow>
<mi>i</mi>
<mi>d</mi>
<mo>=</mo>
<mi>i</mi>
<mi>n</mi>
<mi>d</mi>
<mi>e</mi>
<mi>x</mi>
</mrow>
</msub>
</msub>
<mo><</mo>
<mn>0</mn>
<mo>,</mo>
</mrow>
</mtd>
<mtd>
<mrow>
<mi>i</mi>
<mi>f</mi>
<mi> </mi>
<mi>v</mi>
<mo>_</mo>
<mi>d</mi>
<mi>i</mi>
<mi>r</mi>
<mi>e</mi>
<mi>c</mi>
<mi>t</mi>
<mi>i</mi>
<mi>o</mi>
<mi>n</mi>
<mo><</mo>
<mn>0</mn>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
Wherein, xO,kiThe abscissa at the class center of vehicle to be associated is represented,The number that associated vehicle occurs in last time
According to the abscissa at its class center in frame；
Meet above three condition simultaneously, then preliminary judgement vehicle is successfully associated；
Then, each associated vehicle and vehicle to be associated being successfully associated for above-mentioned preliminary judgement, calculate the two current
The range difference of the location of frame, then select the associated vehicle corresponding to the minimum value in the range difference and car to be associated
It is successfully associated.
4. the continuous vehicle speed detection method according to claim 3 based on laser radar, it is characterised in that the pin
Each associated vehicle and the vehicle to be associated being successfully associated to above-mentioned preliminary judgement, calculate the two in the location of present frame
The process of range difference be：
According to the average speed v of associated vehicleindex′Estimate that the vehicle is in the position of present frame：
Sindex′=vindex′(tki-tindex′,e),
K be present frame data frame number, tkiFor the timestamp of present frame, tindex′,eOccur to be associated vehicle last time
The timestamp of data frame；
The distance between vehicle to be associated and associated vehicle are：
<mrow>
<msub>
<mi>S</mi>
<mrow>
<msup>
<mi>index</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>k</mi>
<mi>i</mi>
</mrow>
</msub>
<mo>=</mo>
<msqrt>
<mrow>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>x</mi>
<mrow>
<mi>F</mi>
<mi>o</mi>
<mi>r</mi>
<mi>E</mi>
<mo>,</mo>
<mi>k</mi>
<mi>i</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>x</mi>
<mrow>
<mi>F</mi>
<mi>o</mi>
<mi>r</mi>
<mi>E</mi>
<mo>,</mo>
<msub>
<mi>k</mi>
<mrow>
<msup>
<mi>index</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>e</mi>
</mrow>
</msub>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
<mo>+</mo>
<msup>
<mrow>
<mo>(</mo>
<msub>
<mi>y</mi>
<mrow>
<mi>F</mi>
<mi>o</mi>
<mi>r</mi>
<mi>E</mi>
<mo>,</mo>
<mi>k</mi>
<mi>i</mi>
</mrow>
</msub>
<mo>-</mo>
<msub>
<mi>y</mi>
<mrow>
<mi>F</mi>
<mi>o</mi>
<mi>r</mi>
<mi>E</mi>
<mo>,</mo>
<msub>
<mi>k</mi>
<mrow>
<msup>
<mi>index</mi>
<mo>&prime;</mo>
</msup>
<mo>,</mo>
<mi>e</mi>
</mrow>
</msub>
</mrow>
</msub>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
</msqrt>
</mrow>
Wherein, xForE,kiRepresent the x-axis coordinate of the characteristic point ForE of vehicle to be associated in current frame data, yForE,kiRepresent current
The characteristic point ForE of vehicle to be associated y-axis coordinate in frame data；
Obtain the range difference of associated vehicle and vehicle to be associated in the location of present frame | Sindex′,ki-Sindex′|。
5. the continuous vehicle speed detection method according to claim 3 based on laser radar, it is characterised in that the choosing
Determining the process that the associated vehicle corresponding to the minimum value in the range difference is successfully associated with vehicle to be associated is：
Take the minimum value in the difference of the distance of vehicle to be associated and all associated vehicles：
ΔSmin_index′,ki=arg min | Sindex′,ki-Sindex′|
The vehicle id that min_index ' is in all associated vehicles and the difference of the distance of vehicle to be associated is minimum, when the minimum
Value meets condition Δ Smin_index′,ki≤Tmin_ΔSWhen, then judge to be successfully associated, the id of vehicle to be associated is entered as min_
Index ', Tmin_ΔSRepresent the error amount that range estimation allows.
6. the continuous vehicle speed detection method according to claim 1 based on laser radar, it is characterised in that the car
Class center refer to, by laser radar obtain data in vehicle data expressed in the form of vehicle box model, institute
Stating the vehicle characteristics point included in vehicle box model has E points, F points, O points and assemblage characteristic point ForE, wherein O point to represent vehicle
Central point, also known as vehicle class center.
7. the continuous vehicle speed detection method according to claim 6 based on laser radar, it is characterised in that the E
Point and F points take its x-axis coordinate and being averaged for y-axis coordinate to be worth to by the nearest N number of points of chosen distance A, B respectively, its time
Stamp information is the average value for the timestamp for taking closest N number of point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711219157.3A CN107705563B (en) | 2017-11-28 | 2017-11-28 | Laser radar-based continuous vehicle speed detection method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711219157.3A CN107705563B (en) | 2017-11-28 | 2017-11-28 | Laser radar-based continuous vehicle speed detection method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107705563A true CN107705563A (en) | 2018-02-16 |
| CN107705563B CN107705563B (en) | 2020-01-31 |
Family
ID=61181116
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711219157.3A Active CN107705563B (en) | 2017-11-28 | 2017-11-28 | Laser radar-based continuous vehicle speed detection method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107705563B (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108828608A (en) * | 2018-03-29 | 2018-11-16 | 苏州大学张家港工业技术研究院 | Laser radar background data filtering method in vehicle checking method |
| CN109598947A (en) * | 2018-12-26 | 2019-04-09 | 武汉万集信息技术有限公司 | A kind of vehicle identification method and system |
| CN109814102A (en) * | 2019-01-31 | 2019-05-28 | 厦门精益远达智能科技有限公司 | A kind of one-lane superelevation monitoring method, device, equipment and storage medium |
| CN110378178A (en) * | 2018-09-30 | 2019-10-25 | 长城汽车股份有限公司 | Method for tracking target and device |
| CN110648538A (en) * | 2019-10-29 | 2020-01-03 | 苏州大学 | Traffic information sensing system and method based on laser radar network |
| EP3621052A1 (en) * | 2018-09-05 | 2020-03-11 | VITRONIC Dr.-Ing. Stein Bildverarbeitungssysteme GmbH | Method for analysing the driving behaviour of motor vehicles, including autonomous vehicles |
| CN111540201A (en) * | 2020-04-23 | 2020-08-14 | 山东大学 | Vehicle queuing length real-time estimation method and system based on roadside laser radar |
| CN111815981A (en) * | 2019-04-10 | 2020-10-23 | 黑芝麻智能科技(重庆)有限公司 | System and method for detecting objects on long distance roads |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101604448A (en) * | 2009-03-16 | 2009-12-16 | 北京中星微电子有限公司 | A kind of speed-measuring method of moving target and system |
| CN102722886A (en) * | 2012-05-21 | 2012-10-10 | 浙江捷尚视觉科技有限公司 | Video speed measurement method based on three-dimensional calibration and feature point matching |
| CN104318782A (en) * | 2014-10-31 | 2015-01-28 | 浙江力石科技股份有限公司 | Expressway video speed measuring method and system for zone overlapping |
| CN106781537A (en) * | 2016-11-22 | 2017-05-31 | 武汉万集信息技术有限公司 | A kind of overspeed of vehicle grasp shoot method and system |
| CN107341819A (en) * | 2017-05-09 | 2017-11-10 | 深圳市速腾聚创科技有限公司 | Method for tracking target and storage medium |
-
2017
- 2017-11-28 CN CN201711219157.3A patent/CN107705563B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101604448A (en) * | 2009-03-16 | 2009-12-16 | 北京中星微电子有限公司 | A kind of speed-measuring method of moving target and system |
| CN102722886A (en) * | 2012-05-21 | 2012-10-10 | 浙江捷尚视觉科技有限公司 | Video speed measurement method based on three-dimensional calibration and feature point matching |
| CN104318782A (en) * | 2014-10-31 | 2015-01-28 | 浙江力石科技股份有限公司 | Expressway video speed measuring method and system for zone overlapping |
| CN106781537A (en) * | 2016-11-22 | 2017-05-31 | 武汉万集信息技术有限公司 | A kind of overspeed of vehicle grasp shoot method and system |
| CN107341819A (en) * | 2017-05-09 | 2017-11-10 | 深圳市速腾聚创科技有限公司 | Method for tracking target and storage medium |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108828608A (en) * | 2018-03-29 | 2018-11-16 | 苏州大学张家港工业技术研究院 | Laser radar background data filtering method in vehicle checking method |
| CN108828608B (en) * | 2018-03-29 | 2021-08-17 | 苏州大学张家港工业技术研究院 | Laser radar background data filtering method in vehicle detection method |
| EP3621052A1 (en) * | 2018-09-05 | 2020-03-11 | VITRONIC Dr.-Ing. Stein Bildverarbeitungssysteme GmbH | Method for analysing the driving behaviour of motor vehicles, including autonomous vehicles |
| CN110378178A (en) * | 2018-09-30 | 2019-10-25 | 长城汽车股份有限公司 | Method for tracking target and device |
| CN110378178B (en) * | 2018-09-30 | 2022-01-28 | 毫末智行科技有限公司 | Target tracking method and device |
| CN109598947A (en) * | 2018-12-26 | 2019-04-09 | 武汉万集信息技术有限公司 | A kind of vehicle identification method and system |
| CN109814102A (en) * | 2019-01-31 | 2019-05-28 | 厦门精益远达智能科技有限公司 | A kind of one-lane superelevation monitoring method, device, equipment and storage medium |
| CN109814102B (en) * | 2019-01-31 | 2020-10-27 | 厦门精益远达智能科技有限公司 | Single lane superelevation monitoring method, device, equipment and storage medium |
| CN111815981A (en) * | 2019-04-10 | 2020-10-23 | 黑芝麻智能科技(重庆)有限公司 | System and method for detecting objects on long distance roads |
| CN110648538A (en) * | 2019-10-29 | 2020-01-03 | 苏州大学 | Traffic information sensing system and method based on laser radar network |
| CN111540201A (en) * | 2020-04-23 | 2020-08-14 | 山东大学 | Vehicle queuing length real-time estimation method and system based on roadside laser radar |
| CN111540201B (en) * | 2020-04-23 | 2021-03-30 | 山东大学 | Vehicle queuing length real-time estimation method and system based on roadside laser radar |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107705563B (en) | 2020-01-31 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107705563A (en) | Continuous vehicle speed detection method based on laser radar | |
| CN109084786B (en) | Map data processing method | |
| US20220126871A1 (en) | Lane Change Monitoring Method and Lane Change Monitoring System for Autonomous Vehicle | |
| CN107894767A (en) | A kind of automatic driving vehicle transverse movement control object system of selection | |
| Alhajyaseen | The integration of conflict probability and severity for the safety assessment of intersections | |
| WO2023124383A1 (en) | Vehicle speed measurement method, collision early-warning method, and electronic device | |
| Feng et al. | An augmented reality environment for connected and automated vehicle testing and evaluation | |
| CN108596081A (en) | A kind of traffic detection method merged based on radar and video camera | |
| CN110414803B (en) | Method and device for evaluating intelligent level of automatic driving system under different internet connection degrees | |
| CN106291736A (en) | Pilotless automobile track dynamic disorder object detecting method | |
| CN106184220B (en) | Abnormal driving detection method in a kind of track based on vehicle location track | |
| CN105844222A (en) | System and method for front vehicle collision early warning based on visual sense | |
| CN113340615B (en) | Automobile automatic driving function simulation test system and method | |
| CN103456172A (en) | Traffic parameter measuring method based on videos | |
| CN106093951A (en) | Object tracking methods based on array of ultrasonic sensors | |
| CN110884502A (en) | Automatic driving path planning quality evaluation method, device and system | |
| CN108458746A (en) | One kind being based on sensor method for self-adaption amalgamation | |
| Zaki et al. | Automated analysis of pedestrians’ nonconforming behavior and data collection at an urban crossing | |
| CN110458080A (en) | The pre-judging method and system of front pit-hole in a kind of running car | |
| Liu et al. | A car-following data collecting method based on binocular stereo vision | |
| CN103927437B (en) | The method measuring space headway in non-rectilinear section | |
| Othman et al. | Safety analysis of horizontal curves using real traffic data | |
| CN117198057A (en) | Experimental method and system for road side perception track data quality inspection | |
| CN109195849A (en) | Photographic device | |
| Pelletier et al. | In-vehicle validation of heavy-duty vehicle fuel savings via a hierarchical predictive online controller |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |