CN106570454B - Pedestrian traffic parameter extraction method based on mobile laser scanning - Google Patents
Pedestrian traffic parameter extraction method based on mobile laser scanning Download PDFInfo
- Publication number
- CN106570454B CN106570454B CN201610883070.5A CN201610883070A CN106570454B CN 106570454 B CN106570454 B CN 106570454B CN 201610883070 A CN201610883070 A CN 201610883070A CN 106570454 B CN106570454 B CN 106570454B
- Authority
- CN
- China
- Prior art keywords
- pedestrian
- target
- data
- traffic parameter
- pedestrians
- 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.)
- Expired - Fee Related
Links
- 238000000605 extraction Methods 0.000 title claims description 6
- 238000001514 detection method Methods 0.000 claims abstract description 41
- 238000000034 method Methods 0.000 claims abstract description 26
- 239000000284 extract Substances 0.000 claims abstract description 5
- 238000012216 screening Methods 0.000 claims description 9
- 230000008859 change Effects 0.000 claims description 2
- 238000002372 labelling Methods 0.000 claims description 2
- 230000000877 morphologic effect Effects 0.000 claims description 2
- 238000013480 data collection Methods 0.000 claims 1
- VMXUWOKSQNHOCA-UKTHLTGXSA-N ranitidine Chemical compound [O-][N+](=O)\C=C(/NC)NCCSCC1=CC=C(CN(C)C)O1 VMXUWOKSQNHOCA-UKTHLTGXSA-N 0.000 abstract description 7
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000008901 benefit Effects 0.000 abstract description 2
- 230000011218 segmentation Effects 0.000 abstract description 2
- 230000004888 barrier function Effects 0.000 description 21
- 238000011160 research Methods 0.000 description 6
- 238000001914 filtration Methods 0.000 description 4
- 238000007689 inspection Methods 0.000 description 4
- 206010039203 Road traffic accident Diseases 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 2
- 239000003550 marker Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000000717 retained effect Effects 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 244000170916 Paeonia officinalis Species 0.000 description 1
- 235000006484 Paeonia officinalis Nutrition 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000005352 clarification Methods 0.000 description 1
- 230000007123 defense Effects 0.000 description 1
- 230000010339 dilation Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000013532 laser treatment Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 238000013517 stratification Methods 0.000 description 1
- 239000000725 suspension Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Human Computer Interaction (AREA)
- Multimedia (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a kind of pedestrian traffic parameter extracting methods based on mobile laser scanning, belong to microcosmic traffic parameter acquiring technology field, the present invention utilizes the three-dimensional laser scanner acquisition vehicle periphery point cloud being installed on vehicle, pass through data segmentation, cluster and detection, respectively obtain different classes of data object, it identifies pedestrian, extracts the fixation atural object on road.Fixation atural object in two frame data is matched, the distance that Autonomous Vehicles are mobile in two frames is calculated, then calculates the speed of vehicle, the final traffic parameter for counting the stream of people.Specific advantage is as follows: the dynamic data of pedestrian and ambient enviroment is acquired on road using the VLP-16 Compact Laser Radar of Velodyne company, it is cheap；Association and traffic parameter for target data are extracted, and in order to obtain correct match information in multiframe data, the present invention has well solved object matching problem using the nearest neighbor method that reflective information is added.
Description
Technical field
The present invention relates to microcosmic traffic parameter acquiring technology fields, show in particular a kind of row based on mobile laser scanning
People's traffic parameter extracting method.
Background technique
Because sense of traffic is thin, traffic accident caused by non-motor vehicle, pedestrian's reason has become China's traffic accident
Pith.Therefore, carrying out for driver seems especially heavy to the initiative recognition of non-motor vehicle, pedestrian and active safety early warning
It wants.Laser scanner technique has obtained research and application in many aspects such as City Modeling, vegetational analysis.The present invention is based on
16 line laser scanner of Velodyne detects a variety of inhomogeneities such as motor vehicle, non-motor vehicle, pedestrian, atural object of vehicle periphery
The data of type respectively obtain different classes of data object by data segmentation, classification.And by taking pedestrian's object as an example, in conjunction with road
Circuit-switched data analyzes the traffic behavior of the class object, obtains corresponding microscopic traffic flow parameter.
It obtains microscopic traffic flow parameter the decisions of Autonomous Vehicles and traffic accident are avoided having great significance.And it is effective
Ground identifies pedestrian and non motorized vehicle using the data that sensor obtains, and the microscopic traffic flow parameter for extracting target is research
One of the obstruction that difficult point and unmanned Autonomous Vehicles are popularized.One reason for this is that sensor needed for obtaining precise information is very high
It is expensive, and need Multi-sensor Fusion.
The present invention utilize 16 relatively inexpensive line laser radars, under single sensor complete pedestrian target identification and
The identification of passageway tree, and then the speed of vehicle and the speed of pedestrian are calculated, and then analyze the traffic behavior of pedestrian.Felt using patrilineal line of descent with only one son in each generation
It is difficult point and significance of the invention that device, which can complete the target,.
In terms of Target detection and identification, according to different sensor used, there are many different methods.Mainstream now
Sensor has monocular camera, more mesh cameras and laser radar.According to all sensors, there is the target inspection based on monocular vision
Survey, the target detection based on stereoscopic vision and the target detection combined based on laser radar and vision.Due to of the invention main
It is the target detection based on laser radar, so main below introduce the target detection based on laser radar.
In the target detection based on laser radar, the radar used has single line radar, multi-thread radar and three-dimensional omnidirectional thunder
It reaches.Wherein single line radar and multi-thread radar can only do the simple functions such as detection of obstacles, and three-dimensional omnidirectional radar can obtain it is richer
Rich, more comprehensively and more accurate environmental information, is widely used in now in the research of unmanned vehicle.
Institutes Of Technology Of Tianjin Sha Depeng et al. has studied vehicle laser velocity-measuring system, realizes the survey to automobile using laser
Speed.Since laser radar plays important function in obstacle detection and identification, Central South University Zhou Zhi et al. has been done based on sharp
The intelligent vehicle active safety algorithm and model of optical radar is studied, and identifies front vehicles using laser radar data, and obtain vehicle
Speed information, judge whether vehicle in a safe condition in conjunction with the knowledge in terms of kinematics.Liu university, the National University of Defense technology
Et al. done the research of multi-line laser radar country obstacle detection, adopted in research using four line laser radars on Autonomous Vehicles
Collect data.Liu great Xue et al. realizes the classification of data in scene；The factor for influencing radar ranging accuracy is analyzed, is given
The filtering method of radar data is gone out；With relative altitude, the gradient and dot density complete in country and hinder as judgment condition
Hinder the identification of object.
In real-time detection and identification, the Cheng Jian of Zhejiang University realizes real-time mesh using 64 line laser radar of velodyne
Mark detection.For target classification, pedestrian is judged according to simple geometry feature in the barrier of non-vehicle.In pedestrian detection side
Face, Navarro-Serment et al. proposes to be based on high-precision three-dimensional laser radar, by target point cloud mass according to two legs and body point
At three parts, point Yun Tezheng, such as covariance matrix, inertial tensor matrix etc. are then extracted in each section, reach quasi-
True pedestrian detection.Luciano Spinello et al. proposes with 3D Lidar come pedestrian detection, according to cluster block highl stratification,
Geometrical characteristic is extracted in each layer, statistical nature obtains classifier finally by machine learning method.Kiyosumi kidono
Et al. propose two kinds of novel features: a kind of length and width for the hierarchy slicing projection being characterized in cluster block indicate the wheel mausoleum of pedestrian,
The pedestrian body of different height level is of different size, and the outline of pedestrian can be preferably described with the description of the feature of hierarchy slicing；
Another kind is characterized in counting the distribution of the point cloud reflected intensity of tested cluster block, come finally by SVM training classifier stable
Pedestrian detection.
Scholar has done the much research about pedestrian detection and vehicle driving safety based on laser radar, proposes many symbols
The algorithm for closing usage scenario, has many reference values to the present invention.But many laser treatment data methods be all based on it is specific
Laser radar data, the present invention needs to propose in specific field the VLP-16 laser radar data based on Velodyne company
The solution that scape uses.
Summary of the invention
Goal of the invention
The main purpose of the present invention is to provide a kind of pedestrian traffic parameter extracting methods based on mobile laser scanning, adopt
The traffic flow parameter analysis that pedestrian is realized with low-cost VLP, solve currently without realize between adjacent two frame shade tree it
Between association and the further method of the traffic flow parameters such as calculating speed.
Technical solution
A kind of pedestrian traffic parameter extracting method based on mobile laser scanning characterized by comprising target detection,
VELOCITY EXTRACTION and traffic parameter statistics, wherein target detection, which is divided into, establishes grating map, target label and object filtering, specific to walk
It is rapid as follows:
S1. it establishes grating map: based on collected three-dimensional laser data, determining that method is true by the barrier based on distance
The range and size of fixed grid lattice；
S2. target label: target label is divided into pedestrian and trees, using pedestrian for the relation mark row of ground level difference
People marks trees according to target height difference using the characteristics of trees are straight and upright and highly significant；
S3. object filtering: after completing grating map mark, pass through the cluster side of morphological dilation and zone marker
Method expands to target grid, after completing expansion and cluster, is screened herein to target object, rejects the object for not meeting feature, this
In screening mainly to pedestrian's Object Operations；
S4. after completing target acquisition screening, data correlation is carried out to target, letter is reflected using being added in data correlation part
The nearest neighbor method of breath completes data correlation, then extracts the speed of data acquisition vehicle；
S5. data acquisition vehicle and pedestrian have been obtained after target detection and target data association this sequence of operations
Speed, traffic parameter of the statistics pedestrian relative to acquisition vehicle.
Preferably, S2 is specific as follows: after dyspoiesis object grating map, obstacle tag being come out, after having marked, first
The data for being are marked in traversal barrier grating map, these data are calculated and judged, if meeting pedestrian
Feature be just labeled as pedestrian, trees are just labeled as if it is the feature for meeting shade tree, at two when marking pedestrian and trees
It is marked on barrier grating map.
Preferably, S3 is specific as follows: utilizing the ratio of width to height constraint condition, identify pedestrian from the point cloud of label, avoid
The interference of other barriers of the roadsides such as street lamp, dustbin.
Preferably, S4 data correlation specific steps: (1) extracting the spy that the trees target (2) that detection is completed calculates target
It levies point (3) and carries out next frame association using the nearest neighbor algorithm that reflective information is added.
Preferably, S5 is specific as follows: using the principle that shade tree is motionless, according to coordinate origin with respect to shade tree, pedestrian
Variation calculate car speed and pedestrian's speed, to calculate the microscopic traffic flow parameter of pedestrian.
Specific advantage is as follows:
1) pedestrian and ambient enviroment are acquired in campus road using the VLP-16 Compact Laser Radar of Velodyne company
Dynamic data, it is cheap；
2) detection of pedestrian and shade tree are marked in order to facilitate the processing of laser point cloud data using grating map
Barrier marks pedestrian and shade tree on the basis of barrier.Finally by expansion and cluster, complete pedestrian and row are extracted
Road tree laser point cloud data；
3) association of target data and traffic parameter are extracted, in order to obtain correctly matching letter in multiframe data
Breath, the present invention have well solved object matching problem using the nearest neighbor method that reflective information is added.Using kinematics formula,
The information such as the speed of vehicle and pedestrian are obtained, microcosmic traffic parameter is counted.
Detailed description of the invention
Fig. 1 is that the trees and pedestrian based on grating map mark flow chart；
Fig. 2 is pedestrian's laser point cloud shape graph at 10 meters of distances of VLP16 sensor acquisition；
Fig. 3 is pedestrian's laser point cloud shape graph at 5 meters of distances of VLP16 sensor acquisition；
Fig. 4 is pedestrian's typical shape of VLP16 acquisition；
Fig. 5 is the shade tree typical shape of VLP16 acquisition.
Specific embodiment
Below in conjunction with the attached drawing in the embodiment of the present application, technical solutions in the embodiments of the present application carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of embodiments of the present application, instead of all the embodiments.It is based on
Embodiment in the application, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall in the protection scope of this application.
Laser radar data has the features such as precision height and big data volume.VLP-16 generates more than 120,000 laser radar each second
Point cloud.Therefore unsuitable directly to be handled in initial data, in terms of target detection, it is fine that data are divided using grating map
Solution this problem.Grid detection has simple and quick, and stability is preferable.As shown in Figure 1, the present invention, which uses, is based on grid
The object detection method of lattice map establishes a grating map for each object, and subsequent identification and relevant parameter is facilitated to extract.
Object detection method based on grating map divides environment using discrete two-dimensional grid.With based on feature
Target detection is compared, and grating map does not need to assume the geometrical model of target, can detect a plurality of types of targets, such as pedestrian, certainly
Driving, automobile etc..The label of the present invention with will be described in detail barrier grid map generalization and target object, with primarily determining grid
The object properties of cell in figure.The Complete three-dimensional point cloud data of target object is obtained followed by expansion and cluster, finally
Carry out the screening of target object.
S1. grating map is established
Grating map expresses ambient enviroment by calculating the probability value in each cell comprising object.Often
The state of a cell is " blank " or " occupying " one of both, or is fallen between, and is unknown state.Establish grating map
Key problem, that is, the posterior probability using data-oriented calculating map.How to utilize data-oriented estimation unit lattice
Quantity of state, different grating maps can set different estimation scheme, tell different target objects.
Continuous space is divided into equally distributed grid by grating map.Most common grating map is horizontal layout
Figure, that is, three-dimensional world is indicated with two-dimensional section figure.The present invention projects to three-dimensional point cloud in two-dimensional grid map, reduces
The quantity and calculation amount of cell, it is single accuracy can be influenced simultaneously.If generating three-dimensional grating map, grid list
The quantity of member can increase several times.And the laser radar that the present invention uses is 16 line radars, when target compares from coordinate origin
When remote, point cloud is very sparse, generates three-dimensional grating map and does not improve a lot to accuracy is improved.Therefore using two-dimensional
Grid map.
In view of the visual range of laser radar, 20 meters after setting grating map size as Chinese herbaceous peony and vehicle, 10 meters of both sides, grid
Lattice size is 20cm*20cm, that is, establishes the grating map of a 200*100.
Example: barrier grating map
The angle of vehicle is set out, and the stream of people, non-motor vehicle and trees are all one kind of barrier.Barrier map is first generated,
Then according to the signature grid of target object.Grating map is by three-dimensional point discretization, according to the data of each cell
Estimation unit trellis state.But when target is dispersed to multiple cells, since data lose integrality, target identification will
It can be very inaccurate.The invention proposes the zone marker methods a kind of grating map by barrier, improve the complete of cluster
Property.
Grating map is generated to barrier, it is necessary first to determine that is only barrier, it then could determination unit lattice category
Property.One of barrier is mainly characterized by the protrusion for having certain altitude than ground, therefore can be with the point cloud in Traversal Unit lattice, such as
The density and height of fruit dot cloud meet certain requirements, so that it may think that there are barriers in this Unit Cell.In Traversal Unit lattice
Three-dimensional point cloud, statistics obtain Max_Z, Min_Z and number of scanning lines n in cell, obtain relative maximum height Δ Z.If Δ Z is greater than
Threshold xi and number of scanning lines are greater than threshold value λ, are labeled as barrier point with regard to cell.It is this that obstacle is determined according to relative altitude and density
The method of point, can effectively reduce erroneous judgement, but it is also possible that increasing the risk failed to judge.Because of angle interval between scan line
Be it is fixed, in the remoter target of laser origin scanning property it is more sparse, possible distant place barrier only has seldom several to sweep
Line scanning is retouched to arrive.
S2. grid object attribute marks
Using label pedestrian and shade tree in multi-thread laser scanning data, and provide the experience ginseng of pedestrian, shade tree label
Number.
Example:
Pedestrian's label
Before marking pedestrian's cell attribute, it is to be understood that the three-dimensional point cloud geometric shape of pedestrian.Select respectively pedestrian with
Point cloud data when lasing central horizontal distance is 10m and 5m is analyzed.Pedestrian as seen from Figure 2, when pedestrian from
When launching centre horizontal distance 10m.Only about 6 scan lines with pedestrian.It is humanoid for capable of finding out dimly from point cloud data,
Therefore we only detect the pedestrian within 10m.As shown in figure 3, the shape of people can be clearly seen that when horizontal distance 5m, on the person
There are about 10 scan lines.Similarly pedestrian can be marked according to the item number of scan line and the geometrical characteristic of pedestrian.
Since laser radar is erected on roof, and laser radar vertical field of view angle only has 30 °, so laser radar is close
Place can have a data and acquire blind area.Pedestrian's body scan cannot be arrived when pedestrian is too close from vehicle, as shown in Figure 3.
From fig. 4 it can be seen that pedestrian's only more than half body can be swashed when pedestrian is 3.5 meters from far point horizontal distance
Photoscanner scanning is arrived, and scan line compares comparatively dense at this time.Pedestrian's geometrical characteristic of the geometrical characteristic of pedestrian and 10m and 5m at this time
It is different, the feature judgment criteria of front cannot be continued to use.According to the erection situation of laser radar, when horizontal distance is less than 4 meters,
Complete pedestrian's point cloud cannot be scanned.
Laser radar blind area is avoided, the complete pedestrian of scanning in policy tag 4m a to 10m is first worked out.According to front institute
The barrier grating map generating principle stated establishes a tightened up specific threshold standard for the generation of pedestrian's grating map.Tool
Body way is traversal barrier grating map, then all the points cloud data traversal in the cell for being to label is primary,
Its height difference and number of scanning lines are counted, cell attribute is re-flagged according to the threshold value of reset.Into the comparison of excessive frame data
Test, provided in experimental data here and show good threshold value standard: maximum height difference Δ Z be greater than 0.9 meter less than 2 meters, scan line
Number is more than or equal to 5 less than 12.
Incomplete pedestrian is scanned in order to be tagged in laser radar blind area, another strategy detection trip need to be specified
People's target.It can be seen in figure 3 that scanning the pedestrian's shape come only has half of body, therefore height difference standard can be relaxed, if
It is scheduled in 4 meters of horizontal distance, target maximum height difference Δ Z is greater than 0.4 meter less than 1.5 meters, and scan line is more than or equal to 6 less than 11.
Shade tree label
According to the method for pedestrian detection, it is equally applicable to detection shade tree, only needs to change threshold value.In given threshold
Before it should be understood that the point cloud data of trees.The point cloud data for extracting trees by hand first, uses Matlab Software on Drawing point
Cloud, as shown in Figure 5.
From fig. 5, it is seen that trunk height is both greater than 2 meters, and there is multi-strip scanning line to sweep to.Similarly according to trees
Geometrical characteristic changes marking-threshold.The labelling strategies of shade tree are: maximum height difference Δ Z is greater than 2 meters, and scan line is greater than 9.Due to
Abundanter apart from the nearlyr tree information of vehicle in order to calculate data acquisition vehicle speed when shade tree detects, setting is compared here
Stringent threshold value retains missing inspection remote with a distance from the woods permission close from vehicle.
1 obstacle tag parameter of table
2 pedestrian's flag parameters of table
3 shade tree flag parameters of table
S3. object filtering
Using the ratio of width to height constraint condition, identify pedestrian from the point cloud of label, avoid the roadsides such as street lamp, dustbin other
The interference of barrier.
Pedestrian target screening
By the sequence of operations of front, although target object is tentatively marked, there are a large amount of erroneous detection,
Middle major part erroneous detection object is trees, it is therefore desirable to which target is further screened, and the object for meeting feature is retained, and rejects erroneous detection
Object.
The present invention is based on depth-width ratio features to shine the pedestrian target selected again.According to " Chinese adult human body
Size (GB10000-88) ", normal person's depth-width ratio is about between 4 to 5.It is set in view of eclipse phenomena, therefore by depth-width ratio threshold value
It sets between 3 to 7.After depth-width ratio constraint is added, only a other erroneous detection point and missing inspection point.
According to experimental data, the depth-width ratio of trees is generally greater than 9 in data, therefore can pick in pedestrian target well
Except trees target.And in trees detection, tree features are fairly obvious, and false detection rate is relatively low.
Pedestrian and surface constraints
Work as pedestrian in 4m and 10m according to front analysis is available, be generally possible to scanning and arrive complete pedestrian's shape data,
Pedestrian always walks on the ground.Can be rejected according to this feature leads to target similar with pedestrian's geometrical characteristic because blocking.
Target in 4m cannot use this screening constraint condition, can only collect pedestrian's half body when distance acquisition vehicle is close
Picture, so " suspension " skyborne target should be retained at this time.
S4. target data association and traffic parameter extract
Data acquisition vehicle and pedestrian are movable bodies, and vehicle and pedestrian can be mobile relative to fixed atural object.In order to obtain vehicle
And pedestrian's velocity information, it is necessary to before and after frames association is carried out to fixed atural object and pedestrian's these two types data, calculates vehicle and row
The traffic parameters such as the speed of people.The processing of these two types of moving targets includes the steps that the data correlation and fortune of two crucial before and after frames
The estimation of dynamic rail mark.For moving object, sensor can not can obtain its motion state according to an only frame data, number
Moving object and the judgement that does well are identified according to continuous multiple frames sensing data according to processing module needs.
By that can be tracked whithin a period of time to target to target data association, then extracted according to kinematics
The speed consecutive variations figure of data acquisition vehicle, the traffic row of the stream of people is counted by the relativeness of data acquisition vehicle and pedestrian
For.
Data correlation specific steps:
(1) trees target (2) calculating clarification of objective point (3) for extracting detection completion, which uses, is added the most adjacent of reflective information
Nearly method carries out next frame association
S5 obtains stream of people's traffic parameter
Using the principle that shade tree is motionless, the variation according to coordinate origin with respect to shade tree, pedestrian calculate car speed and
Pedestrian's speed, to calculate the microscopic traffic flow parameter of pedestrian.
S5.1 coordinate system and kinematic principle
Laser data is collected after conversion, is the Z axis Vertical Launch plane using launching centre as coordinate origin.Coordinate
The origin of system is the launching centre of laser radar always, as the kinetic coordinate system origin of vehicle also can be with movement.
Only a frame data cannot obtain the motion information of this moving target of coordinate origin, need and two frame data of front and back
It is associated the direction of motion for calculating coordinate origin.Using ground as motion reference system, coordinate origin is still relative to fixedly
Object movement.Using instrument as motion reference system, fixed atural object is moved relative to instrument.Therefore fixed atural object is in two frame of front and back
The variation of coordinate is exactly the variation of coordinate origin movement.
S5.2 calculates car speed
After all frames all carry out data correlation, the instantaneous velocity at each moment is obtained using kinematic principle.
Can see according to velocity variations are calculated, automobile is to have been started up when start recording data, have one section plus
In the stage of speed, then period, speed had several minor swings, and the stage reduces speed.Extremely meet with the case where acquisition data, into reality
Automobile starts to accelerate behind road inspection road, and road midway encounters the stream of people so being slowed down, and has finally arrived straight way end automobile and has started
Slow down.
S5.3 pedestrian traffic parameter
According to previously described target detection, pedestrian target can detecte, calculate the data correlation according to two frame of front and back, together
Sample can also calculate the speed of pedestrian.After completing these operations, the traffic parameter of the stream of people can be counted with that.Every 10 frames
The microcosmic traffic parameter of the stream of people is counted, parameter list is as shown in table 4 below.
4 stream of people's traffic parameter of table
Claims (5)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610883070.5A CN106570454B (en) | 2016-10-10 | 2016-10-10 | Pedestrian traffic parameter extraction method based on mobile laser scanning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610883070.5A CN106570454B (en) | 2016-10-10 | 2016-10-10 | Pedestrian traffic parameter extraction method based on mobile laser scanning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106570454A CN106570454A (en) | 2017-04-19 |
| CN106570454B true CN106570454B (en) | 2019-06-11 |
Family
ID=58532696
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610883070.5A Expired - Fee Related CN106570454B (en) | 2016-10-10 | 2016-10-10 | Pedestrian traffic parameter extraction method based on mobile laser scanning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106570454B (en) |
Families Citing this family (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10139833B1 (en) * | 2017-05-17 | 2018-11-27 | GM Global Technology Operations LLC | Six-dimensional point cloud system for a vehicle |
| EP3671279B1 (en) * | 2017-08-18 | 2024-08-14 | Koito Manufacturing Co., Ltd. | Recognition sensor, control method therefor, automobile, vehicle light fixture, object identification system, and object identification method |
| JP6689006B2 (en) | 2017-08-25 | 2020-04-28 | ベイジン・ボイジャー・テクノロジー・カンパニー・リミテッド | Method and system for detecting environmental information of a vehicle |
| CN109521756B (en) * | 2017-09-18 | 2022-03-08 | 阿波罗智能技术(北京)有限公司 | Obstacle motion information generation method and apparatus for unmanned vehicle |
| CN109753982B (en) * | 2017-11-07 | 2021-09-03 | 北京京东乾石科技有限公司 | Obstacle point detection method, obstacle point detection device, and computer-readable storage medium |
| CN110309240A (en) * | 2018-03-14 | 2019-10-08 | 北京京东尚科信息技术有限公司 | The method and apparatus for removing dynamic object |
| CN108501954A (en) * | 2018-04-03 | 2018-09-07 | 北京瑞特森传感科技有限公司 | A kind of gesture identification method, device, automobile and storage medium |
| CN108763730B (en) * | 2018-05-24 | 2022-08-12 | 浙江农林大学 | Street tree screening method, system, terminal and medium based on thermal comfort index |
| CN108803602B (en) * | 2018-06-01 | 2021-07-13 | 浙江亚特电器有限公司 | Obstacle self-learning method and new obstacle self-learning method |
| CN108985254A (en) * | 2018-08-01 | 2018-12-11 | 上海主线科技有限公司 | A kind of band based on laser hangs tag vehicle tracking |
| CN108717540B (en) * | 2018-08-03 | 2024-02-06 | 浙江梧斯源通信科技股份有限公司 | Method and device for distinguishing pedestrians and vehicles based on 2D laser radar |
| WO2020110802A1 (en) * | 2018-11-30 | 2020-06-04 | 株式会社小糸製作所 | On-vehicle object identification system, automobile, vehicle lamp, classifier learning method, and arithmetic operation device |
| CN109407115B (en) * | 2018-12-25 | 2022-12-27 | 中山大学 | Laser radar-based pavement extraction system and extraction method thereof |
| CN109785632B (en) * | 2019-03-14 | 2021-05-04 | 浪潮集团有限公司 | Traffic flow statistical method and device |
| CN110161526A (en) * | 2019-05-24 | 2019-08-23 | 河南辉煌科技股份有限公司 | A kind of circuitry obstacle object recognition methods based on three-dimensional imaging |
| CN110470308B (en) * | 2019-09-06 | 2021-09-24 | 北京云迹科技有限公司 | Obstacle avoidance system and method |
| CN111007534A (en) * | 2019-11-19 | 2020-04-14 | 武汉光庭科技有限公司 | Obstacle detection method and system using sixteen-line laser radar |
| CN113030890B (en) * | 2019-12-24 | 2023-11-21 | 大富科技(安徽)股份有限公司 | Target identification method and device based on vehicle-mounted radar |
| CN113859228B (en) * | 2020-06-30 | 2023-07-25 | 上海商汤智能科技有限公司 | Vehicle control method and device, electronic equipment and storage medium |
| CN113971694A (en) * | 2020-07-22 | 2022-01-25 | 商汤集团有限公司 | Method and device for processing point cloud data |
| CN112102151B (en) * | 2020-07-27 | 2024-05-14 | 广州视源电子科技股份有限公司 | Grid map generation method and device, mobile intelligent device and storage medium |
| CN113343835B (en) * | 2021-06-02 | 2022-04-15 | 合肥泰瑞数创科技有限公司 | Object identification method and system suitable for emergency rescue and storage medium |
| CN113358112B (en) * | 2021-06-03 | 2023-01-17 | 北京超星未来科技有限公司 | Map construction method and laser inertia odometer |
| CN118478908B (en) * | 2024-07-15 | 2024-10-01 | 上海友道智途科技有限公司 | Vehicle drivable boundary planning method based on height layering and vehicle |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010157118A (en) * | 2008-12-26 | 2010-07-15 | Denso It Laboratory Inc | Pattern identification device and learning method for the same and computer program |
| CN102779280A (en) * | 2012-06-19 | 2012-11-14 | 武汉大学 | Traffic information extraction method based on laser sensor |
| CN103871234A (en) * | 2012-12-10 | 2014-06-18 | 中兴通讯股份有限公司 | Grid mapping growth-based traffic network division method and configuration server |
| CN104281746A (en) * | 2014-09-28 | 2015-01-14 | 同济大学 | Method for generating traffic safety road characteristic graphs on basis of point-cloud |
| CN106022460A (en) * | 2016-05-25 | 2016-10-12 | 重庆市勘测院 | Crowd density real-time monitoring method based on laser radar |
-
2016
- 2016-10-10 CN CN201610883070.5A patent/CN106570454B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2010157118A (en) * | 2008-12-26 | 2010-07-15 | Denso It Laboratory Inc | Pattern identification device and learning method for the same and computer program |
| CN102779280A (en) * | 2012-06-19 | 2012-11-14 | 武汉大学 | Traffic information extraction method based on laser sensor |
| CN103871234A (en) * | 2012-12-10 | 2014-06-18 | 中兴通讯股份有限公司 | Grid mapping growth-based traffic network division method and configuration server |
| CN104281746A (en) * | 2014-09-28 | 2015-01-14 | 同济大学 | Method for generating traffic safety road characteristic graphs on basis of point-cloud |
| CN106022460A (en) * | 2016-05-25 | 2016-10-12 | 重庆市勘测院 | Crowd density real-time monitoring method based on laser radar |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106570454A (en) | 2017-04-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106570454B (en) | Pedestrian traffic parameter extraction method based on mobile laser scanning | |
| CN106842231B (en) | A kind of road edge identification and tracking | |
| Serna et al. | Paris-rue-Madame database: a 3D mobile laser scanner dataset for benchmarking urban detection, segmentation and classification methods | |
| Manandhar et al. | Auto-extraction of urban features from vehicle-borne laser data | |
| JP6621445B2 (en) | Feature extraction device, object detection device, method, and program | |
| CN109144072A (en) | A kind of intelligent robot barrier-avoiding method based on three-dimensional laser | |
| CN106199558A (en) | Barrier method for quick | |
| CN112379674B (en) | Automatic driving equipment and system | |
| Abuhadrous et al. | Digitizing and 3D modeling of urban environments and roads using vehicle-borne laser scanner system | |
| CN109716060A (en) | The vehicle location of earth's surface is utilized by event camera | |
| WO2023060632A1 (en) | Street view ground object multi-dimensional extraction method and system based on point cloud data | |
| CN103902976A (en) | Pedestrian detection method based on infrared image | |
| CN106683530A (en) | Computerized judging system and method based on three-dimensional laser vision and high-precision lane model | |
| CN108717540A (en) | The method and device of pedestrian and vehicle are distinguished based on 2D laser radars | |
| Mueller et al. | GIS-based topological robot localization through LIDAR crossroad detection | |
| CN103625649A (en) | Aircraft autonomous landing region judging method based on self adaptive region division and window communication | |
| Zhou et al. | ASL-SLAM: A LiDAR SLAM with activity semantics-based loop closure | |
| Lehtomäki et al. | Performance analysis of a pole and tree trunk detection method for mobile laser scanning data | |
| Schlichting et al. | Vehicle localization by lidar point correlation improved by change detection | |
| Habermann et al. | Artificial neural nets object recognition for 3D point clouds | |
| Wu et al. | A variable dimension-based method for roadside LiDAR background filtering | |
| Tarko et al. | Tscan: Stationary lidar for traffic and safety studies—object detection and tracking | |
| Tang et al. | 3D object detection from roadside data using laser scanners | |
| Jiang et al. | Effective and robust corrugated beam guardrail detection based on mobile laser scanning data | |
| Wolf et al. | Techniques for Automated Classification and Segregation of Mobile Mapping 3D Point Clouds. |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190611 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |