CN104050473B - A kind of road data extracting method based on rectangular neighborhood analysis - Google Patents
A kind of road data extracting method based on rectangular neighborhood analysis Download PDFInfo
- Publication number
- CN104050473B CN104050473B CN201410213465.5A CN201410213465A CN104050473B CN 104050473 B CN104050473 B CN 104050473B CN 201410213465 A CN201410213465 A CN 201410213465A CN 104050473 B CN104050473 B CN 104050473B
- Authority
- CN
- China
- Prior art keywords
- point
- grid
- road
- value
- height value
- 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
Landscapes
- Traffic Control Systems (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of road data extracting methods based on rectangular neighborhood analysis comprising data prediction step, screening road candidate point step extract road axis and reject flat region point set step, road Connection Step.Present invention mainly solves in no reflected intensity, perhaps reflected intensity cannot preferably react terrain object attribute or not have situations such as aviation image, according to the progress road data extraction different from ambient enviroment of road topology design feature.
Description
Technical field
The present invention relates to a kind of road data extracting methods based on rectangular neighborhood analysis.
Background technique
Building and road extraction are always airborne LiDAR (Light Detection And Ranging) data characteristics
The key points and difficulties of extraction, it is more mainly around the extraction research of downtown roads net in terms of road extraction, in the region of field
Method for extracting roads research is less.Wherein airborne LiDAR is that a kind of collection laser ranging, global positioning system (GPS) and inertia are led
Three kinds of technologies of boat system (INS) are used to obtain data and generate accurate three-dimensional landform (DEM) in the system of one；It is airborne
LiDAR (or point cloud data) is the body surface point data acquired by three-dimensional laser scanner, and each point has space three-dimensional seat
Mark, some also have the information such as color and inverse intensity；Point cloud filtering refers to that noise, trees, house in removal point cloud etc. are non-
Ground point；Point cloud tissue, which refers to, is managed a cloud according to certain data format, convenient for the inquiry and display of data；Road
Extraction refers to automatically extracts waypoint from point cloud data.Existing method for extracting roads mainly has based on the road for reflecting strong information
Road extracting method, method for extracting roads based on on-board LiDAR data and remote sensing image etc. are common strong based on point cloud, reflection
The road data extracting method of degree and image is as follows:
(1) the structured road extraction method of Vehicle-borne Laser Scanning data
Vehicle-borne Laser Scanning, which refers to, is erected at the enterprising Mobile state vehicle of the mobile platforms such as automobile for laser scanner, and airborne
Laser scanning difference be the range of airborne lidar is bigger, speed faster, and the data that Vehicle-borne Laser Scanning obtains are all
It is therefore to extract and be relatively easy to along road.
Structured road refers to downtown roads or highway etc., and there is the spy for being higher by waypoint in the both sides of these roads
Sign, automatically extracted by scanning line drawing, the filtering of ground point cloud, road boundary, road boundary tracking and etc. complete road
It extracts.Specific method is detailed in " mapping journal ", and 2 phases of volume 42 in 2013, " structured road of Vehicle-borne Laser Scanning data automatically extracted
Method ".
(2) road extraction based on LiDAR echo information
It extracts using a kind of method (such as: successively encryption TIN method) first or generates digital terrain model (DTM), root
Strong point cloud echo information extracts road information from DTM, deletes therein make an uproar finally by the filtering algorithm of search isolated point
Sound point.Specific method is detailed in " Surveying and mapping " 2 phases " road extraction based on LiDAR echo information " of volume 36 in 2011.
(3) road extraction based on airborne LiDAR roughness index and echo strength
Firstly, by the derivative normalization digital surface model of point cloud data and roughness index, then to the multi-source after registration
Data carry out multi-resolution segmentation, and then are classified using the features such as roughness index and echo strength, the road description factor,
Finally, removal road noise, and obtain accurate road frame net.It is detailed in " Surveying and mapping Technology " 1 phase " bases of volume 30 in 2013
In the road extraction of airborne LiDAR roughness index and echo strength ".
(4) it is extracted based on the urban road network of airborne LiDAR and high-resolution remote sensing image
First then the two progress accuracy registration by vegetation information and is built respectively using the method for pseudo- road information removal
It builds the removals such as object information and obtains basic road profile, recycle morpho logical thinning algorithm to extract the center line of road, finally exist
Improved road Pruning Algorithm (IRT) is realized under ArcGIS and Matlab programmed environment to trim using algorithm progress road
Smooth and coherent urban road network is arrived.Be detailed in " remote sensing technology and application " 2013 4 phases of volume 28 " based on airborne LiDAR and
The urban road network of high-resolution remote sensing image extracts ".
However the reflectivity of return laser beam intensity and ground object target is proportional, i.e., the bigger laser intensity value of reflectivity is more
Height, but the echo strength of laser radar system record is not yet corrected value, it can be by atmospheric attenuation, laser incident angle
The influence of equal many factors, and the correction of return laser beam intensity is so far or a relatively difficult project, therefore laser returns
Intensity of wave is difficult to be truly reflected the reflectivity information of atural object.The extraction effect of road extraction method based on remote sensing image mainly takes
Certainly in the readability of image, but the remote sensing image acquired in the case where cloudy or visibility is inadequate often can not be very
Terrain object attribute is reacted well, therefore it is also more difficult to classify.
Summary of the invention
Object of the present invention is to: a kind of road data extracting method based on rectangular neighborhood analysis is provided, reflection is not being used
Under conditions of intensity and remote sensing image, by analyzing the structure feature of mound forest land domain road, using rectangular neighborhood judgement
Carry out the extraction of road.
The technical scheme is that a kind of road data extracting method based on rectangular neighborhood analysis comprising:
Step 1: data prediction, calculates point cloud spacing, establishes regular grid tissue, then use rule-based grid
Fast math morphologic filtering method be filtered, reject non-ground points；
Step 2: screening road candidate point, primarily determines waypoint set, determination range is reduced, it is adjacent to each point search
Point within the scope of domain, and compare its height value, if being used as road candidate point all in threshold range；
Step 3: extracting road axis, flat region point set is rejected, to each road candidate point, calculates different direction
Whether elevation variance yields in the rectangular extent at angle is road center point according to the diversity judgement current point between variance yields；
Step 4: road connects, intermittent road is attached, smoothly.
Based on the above technical solution, it further includes the following affiliated technical solutions:
The first step includes:
(1) point cloud data range (xmin, xmax, ymin, ymax) is read first, counts quantity pt_Num a little；
(2) point cloud spacing is calculated:
(3) according to cloud spacing pt_dis and calculating needs, covering point cloud range (xmin, xmax, ymin, a ymax) is established
Regular grid；
(4) on-board LiDAR data is assigned in each grid according to the coordinate of point,
(formula 2)
GridNum=XAxisNum*YAxisNum (formula 3)
XAxisNum, YAxisNum x, the direction y grid number, GridNum be grid sum；
Any point p in on-board LiDAR datai(xi,yi) where grid are as follows:
(formula 4)
Wherein piFor piX-direction grid number, the Yi at place be Pi where Y-direction grid number, Gi be Pi where a dimension
The grid number of group record；
Non- ground filtering is carried out to on-board LiDAR data, is retained comprising the ground point data including road waypoint, with rule mesh
Lattice are the mathematical morphology filter of unit, and grid Corrosion results are the minimum height value in w × w grid, and expansion results are w × w
Highest elevation value in grid, wherein w represents the size of window, and carries out the operation of formula 5:
(formula 5)
Wherein A represents target collection, B representative structure element, and A Θ B indicates corrosion,Indicate expansion, z in formulai' and
zi' ' is respectively the height value after grid cell corrosion and expansion, Gridi(zmin) it is elevation minimum value in grid.
The first step further comprises:
(1) by grid distance computation number of grid, and it is grid storage allocation space, point is pressed into plane coordinates storage one by one
It is compared into corresponding grid cell, while with the minimum height value saved in the grid cell, after recording relatively
Minimum height value；
(2) each grid is traversed, minimum height value in w × w grid of surrounding is compared, using first formula of formula 5,
The height value being minimized after corroding for current grid, and the grid highest elevation value is set as the height value after corrosion；
(3) each grid is traversed, w × w grid internal corrosion height value of surrounding is compared, using second formula of formula 5,
The height value being maximized after being expanded for current grid, and the grid minimum height value is set as the height value after expansion；
(4) each grid is traversed, the difference of the height value after seeking each point height value in grid and expanding, when its absolute value
When less than given threshold, otherwise it is non-ground points which, which is ground point,.
The second step includes:
(1) it is directed to each point pi(xi,yi) by its place grid of the calculating of formula 4；
(2) search radius is calculated,
(3) set up an office pi(xi,yi) rectangular neighborhood in point set be
Pi_near={ pj(xj,yj)|xj∈(xi-r,xi+r),yj∈(yi-r,yi+ r) } (wherein j be point point number in other words
It is subscript, and is distinguished with current point pi), thus centered on current point pi, search for the point set P in surrounding gridi_near；
(4) P is calculatedi_nearIn each point and pi(xi,yi) the sum of depth displacement absolute value Σ HiIf Σ HiLess than threshold value, then
Current point is road candidate point；
(formula 6)
Zj is Pi_nearIn j-th point of height value, zi is pi(xi,yi) height value, n Pi_nearTotal points；
(5) above step is repeated, until completing the judgement to all the points, and is P by road candidate's point set of reservationc。
The third step includes:
(1) to each road candidate point pi, rectangular extent point p is calculated according to formula 7i,r1, pi,r2, pi,r3, pi,r4, i.e. rectangle
Four vertex, pi,r1It is overlapped with current point, remaining presses arranged counterclockwise, and wherein d θ is the rotation angle of rectangle, by 360 degree
It is divided into n interval；
K=0,1,2 ... 18 (formula 7)
Wherein pi,r1It (x) is pi,r1X coordinate value, pi,r1It (y) is pi,r1Y-coordinate value, remaining similarly, l be rectangle it is long
Degree, w are rectangle width, and d θ is the rotation angle interval of rectangle, and k is the number of rotation or is the number of iterations；
(2) according to the neighborhood judgment method of step 4, p is searched foriThe neighbor point P of pointi_near, then judge Pi_nearWhether
pi,r1, pi,r2, pi,r3, pi,r4In the rectangular extent of composition, point set P will be saved as in rangei_r,k, and calculate its elevation side
Difference
Elevation mean value:Wherein zjFor point set Pi_r,kIn j-th point of height value, m is point set Pi_r,kPoint
Number；
Elevation variance yields:
(3) according to above two step, point p is calculatediAll rectangles (k=0,1,2 ... 18) elevation variance yields neighborhood in put,
Obtain elevation variance value setThe elevation variance yields being respectively compared in set Ψ records elevation
Variance yields is minimum, second small and three values that third is smallWithWith corresponding rectangle number k1、k2、k3If
Meet following two equation then current point p simultaneouslyiIt is denoted as road center point；
ψ is that user sets the coarse threshold value of road in formula, and μ is that user sets Road turnings threshold value；
(4) above step is repeated, until completing the judgement to all the points, and remembers that road-center point set is Pk。
4th step includes:
According to the road candidate's point set P being calculatedcWith road-center point set Pk, road candidate point is concentrated may be containing big
The point for measuring the flat regions such as parking lot, sports ground, roof is rejected these points by the following method:
(1) by point set PcAnd PkCarry out specification grid dividing, and with encode the type that 1 and 2 indicate a little be road candidate point also
It is road center point；
(2) to point set PcIn each point, search for around it point that attribute is 2 within the scope of r/2, the point retained if having,
Otherwise it is judged as non-rice habitats point, parameter r/2 indicates the half of road width and the ratio of grid spacing.
The invention has the advantages that:
Present invention mainly solves in no reflected intensity, perhaps reflected intensity cannot preferably react terrain object attribute or not have
Situations such as aviation image, according to road topology design feature progress road extraction different from ambient enviroment.Only with a cloud number
According to the road extraction for carrying out mound forest land domain, the less dependence to initial data.
Detailed description of the invention
The invention will be further described with reference to the accompanying drawings and embodiments:
Fig. 1 is the division schematic diagram that on-board LiDAR data is assigned to each grid by the present invention；
Fig. 2 is w × w grid schematic diagram of the invention；
Fig. 3 is that regular grid of the invention divides schematic diagram；
Fig. 4 is neighborhood point set schematic diagram of the invention；
Fig. 5 is rectangular extent point schematic diagram of the invention；
Fig. 6 is rotation rectangle schematic diagram of the invention.
Specific embodiment
Embodiment: with reference to shown in Fig. 1-6, the present invention provides a kind of road data extraction sides based on rectangular neighborhood analysis
Method, the especially a kind of airborne mound LiDAR forest land domain method for extracting roads based on rectangular neighborhood analysis, overall framework are as follows:
Step 1: data prediction, first calculates point cloud spacing, establishes regular grid tissue, then use rule-based net
The fast math morphologic filtering method of lattice is filtered, and rejects non-ground points；
Step 2: screening road candidate point, primarily determines waypoint set, determination range is reduced, it is adjacent to each point search
Point within the scope of domain, and compare its height value, if being used as road candidate point all in threshold range.
Step 3: extracting road axis calculates elevation in the rectangular extent of different orientations to each road candidate point
Whether variance yields is road center point according to the diversity judgement current point between variance yields.
Step 4: non-rice habitats point is rejected.
Specific detailed step is as follows:
1, according to on-board LiDAR data range and quantity survey (surveying) point cloud spacing pt_dis, wherein on-board LiDAR data range
It is the range for giving directions cloud to be covered, the general coordinate representation with four angles；
(1) on-board LiDAR data range (xmin, xmax, ymin, ymax) is read by airborne LiDAR first, counts point
Quantity pt_num；
(2) point cloud spacing pt_dis is calculated:
(formula 1)
2, regular grid division is carried out to on-board LiDAR data, facilitates the search for neighborhood point, specific steps and be described as follows:
(1) according to cloud spacing pt_dis and calculating needs, covering point cloud range (xmin, xmax, ymin, a ymax) is established
Regular grid, it is preferably point 2~3 times of cloud spacing pt_dis that the present invention, which takes grid spacing d,.
(2) on-board LiDAR data is assigned in each grid according to the coordinate of point.
(formula 2)
GridNum=XAxisNum*YAxisNum (formula 3)
XAxisNum, YAxisNum x, the direction y grid number, GridNum be grid sum.
Any point p in on-board LiDAR datai(xi,yi) where grid are as follows:
(formula 4)
X-direction grid number, Yi where wherein Xi is Pi be Pi where Y-direction grid number, Gi be Pi where it is one-dimensional
The grid number of array record, as shown in Figure 1, the data in figure are calculated as XAxisNum=11, YAxisNum=7, GridNum=
77, Xi=7, Yi=4, Gi=4*11+7=51 indicate Pi point in the 51st grid.
3, non-ground filtering is carried out to on-board LiDAR data by the basic operation of mathematical morphology and its extended arithmetic,
Retain comprising the ground point data including road waypoint.
Using regular grid as the mathematical morphology filter of unit, grid Corrosion results are the minimum elevation in w × w grid
Value, expansion results are highest elevation value in w × w grid, and wherein w represents the size of window, that is, sizing grid, such as Fig. 2 institute
Show, is the schematic diagram of w=2.
(formula 5)
Wherein formula 5 is typical regular mathematical morphology formula, and A represents target collection, B representative structure element, A Θ B
Indicate corrosion,Indicate expansion, z in formulai' and zi' ' is respectively the height value after grid cell corrosion and expansion, Gridi
(zmin) it is elevation minimum value in grid, w is the size of window.
Specific step is as follows:
(1) by grid distance computation number of grid, and it is grid storage allocation space, point is pressed into plane coordinates storage one by one
Into corresponding grid cell, while (initial default minimum is compared with the minimum height value saved in the grid cell
Height value is a very big value), the minimum height value after recording relatively；
(2) each grid is traversed, minimum height value in w × w grid of surrounding is compared, using first formula of formula 5,
The height value being minimized after corroding for current grid, and the grid highest elevation value is set as the height value after corrosion, specifically
It is to be compared the minimum height value in a certain range around in grid, retains the smallest height value, and be assigned to and worked as
This variable of the highest elevation value of preceding grid；
(3) each grid is traversed, w × w grid internal corrosion height value of surrounding is compared, using second formula of formula 5,
The height value being maximized after being expanded for current grid, and the grid minimum height value is set as the height value after expansion, specifically
It is to be compared the corrosion height value (result that previous step calculates) in a certain range around in grid, retains the smallest elevation
Value, and it is assigned to this variable of the minimum height value of current grid；
(4) each grid is traversed, the height value after seeking each point height value in grid and expanding is (minimum i.e. in the grid
Height value) difference, when its absolute value be less than given threshold when, the point be ground point, be otherwise non-ground points.
It is handled by above four step, the filtering processing to on-board LiDAR data can be rapidly completed.
4, estimate the width road_width of road, choose road candidate point in neighborhood.
(1) it is directed to each point pi(xi,yi) by its place grid of the calculating of formula 4；
(2) search radius is calculated,
(3) set up an office pi(xi,yi) rectangular neighborhood in point set be
Pi_near={ pj(xj,yj)|xj∈(xi-r,xi+r),yj∈(yi-r,yi+ r) } (wherein j be point point number in other words
It is subscript, in order to be different from the subscript i) of current point pi, thus centered on current point pi, searches for the point set in surrounding grid
Pi_near, as shown in figure 4, being p with the point that arrow is directed towardi(xi,yi), when R=3, Pi_nearPoint set is the point marked.
(4) P is calculatedi_nearIn each point and pi(xi,yi) the sum of depth displacement absolute value Σ HiIf Σ HiLess than threshold value, then
Current point is road candidate point.
(formula 6)
Zj is Pi_nearIn j-th point of height value, zi is pi(xi,yi) height value, n Pi_nearTotal points.
(5) above step is repeated, until completing the judgement to all the points, and road candidate's point set of reservation is denoted as Pc。
5, road axis is searched for
To each cloud, rectangular search neighborhood is established, the length l of rectangle is preferably 2~3 times of width w.
(1) to each road candidate point pi, rectangular extent point p is calculated according to formula 7i,r1, pi,r2, pi,r3, pi,r4, i.e. rectangle
Four vertex, pi,r1It is overlapped with current point, remaining presses arranged counterclockwise, and wherein d θ is the rotation angle of rectangle, by 360 degree
It is divided into n interval, the present invention takes θ=20 degree d, i.e. n=18, as shown in Figure 4.
K=0,1,2 ... 18 (formula 7)
Wherein pi,r1It (x) is pi,r1X coordinate value, pi,r1It (y) is pi,r1Y-coordinate value, remaining similarly, l be rectangle it is long
Degree, w are rectangle width, and d θ is the rotation angle interval of rectangle, and k is the number of rotation or is the number of iterations.
(2) according to the neighborhood judgment method of step 4, p is searched foriThe neighbor point P of pointi_near, then judge Pi_nearWhether
pi,r1, pi,r2, pi,r3, pi,r4In the rectangular extent of composition, point set P will be saved as in rangei_r,k, and calculate its elevation side
Difference
Elevation mean value:Wherein zjFor point set Pi_r,kIn j-th point of height value, m is point set Pi_r,kPoint
Number；
Elevation variance yields:
(3) according to above two step, point p is calculatediAll rectangles (k=0,1,2 ... 18) elevation variance yields neighborhood in put,
Obtain elevation variance value setThe elevation variance yields being respectively compared in set Ψ records elevation
Variance yields is minimum, second small and three values that third is smallWithWith corresponding rectangle number k1、k2、k3If
Meet following two equation then current point p simultaneouslyiIt is denoted as road center point.
ψ is that user sets the coarse threshold value of road in formula, and it is that user sets Road turnings threshold value that the present invention, which is taken as 0.5, μ, this
Invention is taken as 1, i.e. angle of turn is up to 20 degree.
(4) above step is repeated, until completing the judgement to all the points, and remembers that road-center point set is Pk。
6, non-rice habitats point is rejected
The road candidate's point set P calculated separately according to the 4th, 5 stepscWith road-center point set Pk, road candidate point concentration
These points may be rejected by the following method containing the point of the flat regions such as a large amount of parking lots, sports ground, roof.
(1) method similar with step 2 is used, by point set PcAnd PkSpecification grid dividing is carried out, and is indicated with coding 1 and 2
The type of point is road candidate point or road center point.
(2) to point set point set PcIn each point, search for around it point that attribute is 2 within the scope of r/2, retaining if having should
Otherwise point is judged as non-rice habitats point.Parameter r/2 indicates the half of road width and the ratio of grid spacing, the present invention
Middle r/2 is taken as 2.
Certainly the above embodiments merely illustrate the technical concept and features of the present invention, and its object is to allow be familiar with technique
People can understand the content of the present invention and implement it accordingly, it is not intended to limit the scope of the present invention.It is all according to this hair
The equivalent transformation or modification that the Spirit Essence of bright main technical schemes is done, should be covered by the protection scope of the present invention.
Claims (3)
1. a kind of road data extracting method based on rectangular neighborhood analysis, it is characterised in that comprising:
Step 1: data prediction, calculates point cloud spacing, regular grid tissue is established, then using the fast of rule-based grid
Fast mathematical morphology filter method is filtered, and rejects non-ground points；
Step 2: screening road candidate point, primarily determines waypoint set, determination range is reduced, to its neighborhood model of each point search
Interior point is enclosed, and compares its height value, if being used as road candidate point all in threshold range；
Step 3: extracting road axis, flat region point set is rejected, to each road candidate point, calculates different orientations
Whether elevation variance yields in rectangular extent is road center point according to the diversity judgement current point between variance yields；
Step 4: road connects, intermittent road is attached, smoothly；
Wherein the first step includes:
(1) point cloud data range (x is read firstmin, xmax, ymin, ymax), count quantity pt_num a little；
(2) point cloud spacing is calculated:
(3) according to cloud spacing pt_dis and calculating needs, a covering point cloud range is established
(xmin, xmax, ymin, ymax) regular grid；
(4) on-board LiDAR data is assigned in each grid according to the coordinate of point,
GridNum=XAxisNum*YAxisNum (formula 3)
XAxisNum, YAxisNum x, the direction y grid number, GridNum be grid sum；
Any point p in on-board LiDAR datai(xi,yi) where grid are as follows:
Wherein XiFor piX-direction grid number, the Y at placeiFor piY-direction grid number, GiFor piThe one-dimension array at place records
Grid number；
Non- ground filtering is carried out to on-board LiDAR data, retains comprising the ground point data including road waypoint, is with regular grid
The mathematical morphology filter of unit, grid Corrosion results are the minimum height value in w × w grid, and expansion results are w × w grid
Interior highest elevation value, wherein w represents the size of window, and carries out the operation of formula 5:
Wherein A represents target collection, B representative structure element, and A Θ B indicates corrosion,Indicate expansion, z in formulai' and zi" point
Not Wei grid cell corrosion and expansion after height value, Gridi(zmin) it is elevation minimum value in grid；
The first step further comprises:
(1) by grid distance computation number of grid, and it is grid storage allocation space, point is pressed into plane coordinates storage to phase one by one
It in the grid cell answered, while being compared with the minimum height value saved in the grid cell, the minimum after recording relatively
Height value；
(2) each grid is traversed, compares minimum height value in w × w grid of surrounding and is taken most using first formula of formula 5
Small value is the height value after current grid corrosion, and the grid highest elevation value is set as the height value after corrosion；
(3) each grid is traversed, compares w × w grid internal corrosion height value of surrounding and is taken most using second formula of formula 5
Big value is the height value after current grid expansion, and the grid minimum height value is set as the height value after expansion；
(4) each grid is traversed, the difference of the height value after seeking each point height value in grid and expanding, when its absolute value is less than
When given threshold, otherwise it is non-ground points which, which is ground point,；
The second step includes:
(1) it is directed to each point pi(xi,yi) by its place grid of the calculating of formula 4；
(2) search radius is calculated,
(3) set up an office pi(xi,yi) rectangular neighborhood in point set be Pi_near={ pj(xj, yj)|xj∈(xi- r, xi+ r), yj∈
(yi- r, yi+ r) } (wherein j be point point number in other words subscript, and with current point piDistinguish), thus with current point piFor
The point set P in surrounding grid is searched at centeri_near；
(4) P is calculatedi_nearIn each point and pi(xi,yi) the sum of depth displacement absolute value ∑ HiIf ∑ HiLess than threshold value, then currently
Point is road candidate point；
zjIt is Pi_nearIn j-th point of height value, ziIt is pi(xi,yi) height value, n Pi_nearTotal points；
(5) above step is repeated, until completing the judgement to all the points, and is P by road candidate's point set of reservationc。
2. a kind of road data extracting method based on rectangular neighborhood analysis according to claim 1, it is characterised in that: institute
Stating third step includes:
(1) to each road candidate point pi, rectangular extent point p is calculated according to formula 7i,r1, pi,r2, pi,r3, pi,r4, i.e. the four of rectangle
A vertex, pi,r1It is overlapped with current point, remaining presses arranged counterclockwise, and wherein d θ is the rotation angle interval of rectangle, by 360 degree
It is divided into n interval；
Wherein pi,r1It (x) is pi,r1X coordinate value, pi,r1It (y) is pi,r1Y-coordinate value, remaining similarly, l is rectangle length, w
For rectangle width, k is the number of rotation or is the number of iterations；
(2) according to the neighborhood judgment method of step 4, p is searched foriThe neighbor point P of pointi_near, then judge Pi_nearWhether in pi,r1,
pi,r2, pi,r3, pi,r4In the rectangular extent of composition, point set P will be saved as in rangei_r,k, and calculate its elevation variance
Elevation mean value:Wherein zjFor point set Pi_r,kIn j-th point of height value, m is point set Pi_r,kPoint number；
Elevation variance yields:
(3) according to above two step, point p is calculatediAll rectangular neighborhoods in put elevation variance yields, obtain elevation variance value setThe elevation variance yields being respectively compared in set Ψ, record elevation variance yields is minimum, second it is small and
Three small values of thirdWithWith corresponding rectangle number k1、k2、k3If meeting following two equation simultaneously
Then current point piIt is denoted as road center point；
Y is that user sets the coarse threshold value of road in formula, and μ is that user sets Road turnings threshold value；
(4) above step is repeated, until completing the judgement to all the points, and remembers that road-center point set is Pk。
3. a kind of road data extracting method based on rectangular neighborhood analysis according to claim 2, it is characterised in that: institute
Stating the 4th step includes:
According to the road candidate's point set P being calculatedcWith road-center point set Pk, road candidate point, which is concentrated to contain, largely to stop
The point of the flat region such as parking lot, sports ground, roof is rejected these points by the following method:
(1) by point set PcAnd PkSpecification grid dividing is carried out, and indicates that type a little is road candidate point or road with coding 1 and 2
Lu Zhizheng point；
(2) to point set PcIn each point, search for around it point that attribute is 2 within the scope of r/2, the point retained if having, otherwise
It is judged as that non-rice habitats point, parameter r/2 indicate the half of road width and the ratio of grid spacing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410213465.5A CN104050473B (en) | 2014-05-20 | 2014-05-20 | A kind of road data extracting method based on rectangular neighborhood analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410213465.5A CN104050473B (en) | 2014-05-20 | 2014-05-20 | A kind of road data extracting method based on rectangular neighborhood analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104050473A CN104050473A (en) | 2014-09-17 |
| CN104050473B true CN104050473B (en) | 2019-07-19 |
Family
ID=51503285
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410213465.5A Expired - Fee Related CN104050473B (en) | 2014-05-20 | 2014-05-20 | A kind of road data extracting method based on rectangular neighborhood analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104050473B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105893961A (en) * | 2016-03-30 | 2016-08-24 | 广东中冶地理信息股份有限公司 | Method for extracting road center line |
| CN108021844B (en) * | 2016-10-31 | 2020-06-02 | 阿里巴巴（中国）有限公司 | Road edge identification method and device |
| CN106709892A (en) * | 2016-11-15 | 2017-05-24 | 昂纳自动化技术（深圳）有限公司 | Rapid region expansion algorithm and device of any structure element based on stroke coding |
| CN107657636B (en) * | 2017-10-16 | 2021-07-30 | 南京市测绘勘察研究院股份有限公司 | Method for automatically extracting road topographic map elevation points based on vehicle-mounted laser radar data |
| CN111158015B (en) * | 2019-12-31 | 2020-11-24 | 飞燕航空遥感技术有限公司 | Detection method and system for point cloud data of airborne laser radar to be wrongly divided into ground points |
| CN111783721B (en) * | 2020-07-13 | 2021-07-20 | 湖北亿咖通科技有限公司 | Lane line extraction method of laser point cloud and electronic equipment |
| CN111783722B (en) * | 2020-07-13 | 2021-07-06 | 湖北亿咖通科技有限公司 | Lane line extraction method of laser point cloud and electronic equipment |
| CN111932477B (en) * | 2020-08-07 | 2023-02-07 | 武汉中海庭数据技术有限公司 | Noise removal method and device based on single line laser radar point cloud |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101901343A (en) * | 2010-07-20 | 2010-12-01 | 同济大学 | Remote sensing image road extracting method based on stereo constraint |
| CN103500338A (en) * | 2013-10-16 | 2014-01-08 | 厦门大学 | Road zebra crossing automatic extraction method based on vehicle-mounted laser scanning point cloud |
| CN103605135A (en) * | 2013-11-26 | 2014-02-26 | 中交第二公路勘察设计研究院有限公司 | Road feature extracting method based on fracture surface subdivision |
| CN103714339A (en) * | 2013-12-30 | 2014-04-09 | 武汉大学 | SAR image road damaging information extracting method based on vector data |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3780922B2 (en) * | 2001-11-30 | 2006-05-31 | 日産自動車株式会社 | Road white line recognition device |
-
2014
- 2014-05-20 CN CN201410213465.5A patent/CN104050473B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101901343A (en) * | 2010-07-20 | 2010-12-01 | 同济大学 | Remote sensing image road extracting method based on stereo constraint |
| CN103500338A (en) * | 2013-10-16 | 2014-01-08 | 厦门大学 | Road zebra crossing automatic extraction method based on vehicle-mounted laser scanning point cloud |
| CN103605135A (en) * | 2013-11-26 | 2014-02-26 | 中交第二公路勘察设计研究院有限公司 | Road feature extracting method based on fracture surface subdivision |
| CN103714339A (en) * | 2013-12-30 | 2014-04-09 | 武汉大学 | SAR image road damaging information extracting method based on vector data |
Non-Patent Citations (2)
| Title |
|---|
| "地面激光扫描数据滤波研究";王勇;《中国优秀硕士学位论文全文数据库 基础科学辑》;20130615(第06期);第1、28、29、33、34页 |
| "结合角度纹理信息和Snake方法从LiDAR点云数据中提取道路交叉口";陈卓，马洪超，李云帆;《国土资源遥感》;20131231;第25卷(第4期);第80页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104050473A (en) | 2014-09-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104050473B (en) | A kind of road data extracting method based on rectangular neighborhood analysis | |
| Guan et al. | Use of mobile LiDAR in road information inventory: A review | |
| Yu et al. | Automated extraction of urban road facilities using mobile laser scanning data | |
| CN103500338B (en) | Road zebra crossing extraction method based on Vehicle-borne Laser Scanning point cloud | |
| Monnier et al. | Trees detection from laser point clouds acquired in dense urban areas by a mobile mapping system | |
| WO2018068653A1 (en) | Point cloud data processing method and apparatus, and storage medium | |
| Yang et al. | Semi-automated extraction and delineation of 3D roads of street scene from mobile laser scanning point clouds | |
| Yu et al. | Automated detection of urban road manhole covers using mobile laser scanning data | |
| Choi et al. | An automatic approach for tree species detection and profile estimation of urban street trees using deep learning and Google street view images | |
| KR102450019B1 (en) | Water Quality Monitoring Method and System for Using Unmanned Aerial Vehicle | |
| Safaie et al. | Automated street tree inventory using mobile LiDAR point clouds based on Hough transform and active contours | |
| Yadav et al. | Identification of trees and their trunks from mobile laser scanning data of roadway scenes | |
| Franklin | Interpretation and use of geomorphometry in remote sensing: a guide and review of integrated applications | |
| Fan et al. | Identifying man-made objects along urban road corridors from mobile LiDAR data | |
| Sameen et al. | A simplified semi-automatic technique for highway extraction from high-resolution airborne LiDAR data and orthophotos | |
| Zhou et al. | UAV Laser scanning technology: A potential cost-effective tool for micro-topography detection over wooded areas for archaeological prospection | |
| Guo et al. | Extraction of dense urban buildings from photogrammetric and LiDAR point clouds | |
| Wang et al. | Road boundary, curb and surface extraction from 3D mobile LiDAR point clouds in urban environment | |
| Yadav | A multi-constraint combined method for road extraction from airborne laser scanning data | |
| CN117523522B (en) | Tunnel scene identification method, device, equipment and storage medium | |
| Guan et al. | Automated extraction of manhole covers using mobile LiDAR data | |
| Wu et al. | Building a water feature extraction model by integrating aerial image and lidar point clouds | |
| Zhao | Recognizing features in mobile laser scanning point clouds towards 3D high-definition road maps for autonomous vehicles | |
| Shokri et al. | POINTNET++ Transfer Learning for Tree Extraction from Mobile LIDAR Point Clouds | |
| Guan | Automated extraction of road information from mobile laser scanning data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190719 Termination date: 20200520 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |