CN106056614A - Building segmentation and contour line extraction method of ground laser point cloud data - Google Patents
Building segmentation and contour line extraction method of ground laser point cloud data Download PDFInfo
- Publication number
- CN106056614A CN106056614A CN201610393443.0A CN201610393443A CN106056614A CN 106056614 A CN106056614 A CN 106056614A CN 201610393443 A CN201610393443 A CN 201610393443A CN 106056614 A CN106056614 A CN 106056614A
- Authority
- CN
- China
- Prior art keywords
- point
- point cloud
- building
- neighborhood
- segmentation
- 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.)
- Pending
Links
- 230000011218 segmentation Effects 0.000 title claims abstract description 52
- 238000000605 extraction Methods 0.000 title claims abstract description 16
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 33
- 238000000513 principal component analysis Methods 0.000 claims abstract description 21
- 238000004364 calculation method Methods 0.000 claims abstract description 6
- 238000000034 method Methods 0.000 claims description 58
- 239000011159 matrix material Substances 0.000 claims description 20
- VMXUWOKSQNHOCA-UKTHLTGXSA-N ranitidine Chemical compound [O-][N+](=O)\C=C(/NC)NCCSCC1=CC=C(CN(C)C)O1 VMXUWOKSQNHOCA-UKTHLTGXSA-N 0.000 claims description 8
- 238000009826 distribution Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 6
- 238000002372 labelling Methods 0.000 claims description 6
- 238000012217 deletion Methods 0.000 claims description 5
- 230000037430 deletion Effects 0.000 claims description 5
- 238000003491 array Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000003709 image segmentation Methods 0.000 claims description 3
- 230000000717 retained effect Effects 0.000 claims description 3
- 238000005070 sampling Methods 0.000 claims description 3
- 230000005484 gravity Effects 0.000 claims 1
- 239000006185 dispersion Substances 0.000 description 3
- 230000007812 deficiency Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 238000000638 solvent extraction Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/06—Topological mapping of higher dimensional structures onto lower dimensional surfaces
- G06T3/067—Reshaping or unfolding 3D tree structures onto 2D planes
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T3/00—Geometric image transformations in the plane of the image
- G06T3/06—Topological mapping of higher dimensional structures onto lower dimensional surfaces
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
- G06T2207/10012—Stereo images
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (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)
- Multimedia (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a building segmentation and contour line extraction method of ground laser point cloud data. The building segmentation and contour line extraction method comprises the steps of vertically projecting a point cloud, generating a two-dimensional grayscale image, segmenting a building point cloud by adopting an Otsu algorithm, carrying out principal component analysis, conducting best neighborhood calculation, extracting contour lines and the like. The building segmentation and contour line extraction method realizes the segmentation of the building point cloud from the original point clout quickly and accurately, and completes the extraction of the building contour line point cloud in a full-automatic manner.
Description
Technical field
The invention belongs to ground laser point cloud data and process applied technical field, relate to a kind of to initial three-dimensional cloud data
Carry out the Accurate Segmentation of building and automatic contour line extracting method.
Background technology
Territorial laser scanning technology is an emerging three-dimensional measurement scanning technique, its high-precision some cloud of magnanimity obtained
Data, it is provided that the finest three-dimensional information, preferably reflect details and the hierarchical information of building, high-precision to building
For number of degrees word model, it it is a kind of brand-new method.While ground laser provides high-quality cloud data for us, data volume
We are processed and it create certain difficulty by growth the most at double.The reason causing this problem mainly have two: one be
While scanning target structures object point cloud, inevitably introduce noise and non-targeted atural object cloud data；Two is that great majority are built
Building thing facade to regard the plane differed in size by many as and form, each plane removes the some cloud of a small amount of outline to follow-up modeling
Helpful, major part internal plane point cloud can be deleted.Present invention seek to address that above-mentioned two problems, and by invention point
Two parts are extracted for building point cloud segmentation and contour line.
At present, the method for building point cloud segmentation has the side that region growth method, method based on edge, feature based cluster
Method, method based on bidimensional image, method based on echo strength etc..In above-mentioned partitioning algorithm, some precision are high but calculate multiple
Miscellaneous, have the shortest but there is segmentation or adaptive problem by mistake.At present, also there is no a kind of perfect algorithm, at the meter of segmentation
Calculation can take into account accuracy and speed well.The present invention combines method based on bidimensional image and region growth method, and makes
Certain improvement, completes a kind of partitioning algorithm taking into account accuracy and speed.
At present, the method that building object point cloud contour line extracts mainly has method based on triangular grid, based on grid partition
Method, the method etc. of characteristic information based on point.Mostly these methods are the detailed information according to a cloud, utilize feature extraction or
Person's constraint completes the extraction of contour line point cloud, and it is complex that they have common problem to calculate exactly, time complexity
Height, and differ for different cloud data experiment effect quality.Primary study of the present invention some cloud character description method and
The method that best scale determines, makes full use of neighborhood information and the space geometry relation of a cloud, and introduces principal component analysis with fragrant
The calculating of agriculture entropy, not only increases precision and the adaptivity of contour line extraction algorithm, and for different ground laser spots
Cloud data, it is not necessary to repeatedly adjust threshold value, it is achieved that automatically extracting of contour line.
Summary of the invention
In order to solve above-mentioned technical problem, the invention provides building segmentation and the wheel of a kind of ground laser point cloud data
Profile extracting method, the invention mainly relates to two parts content, i.e. building point cloud segmentation module and building object point cloud contour line
Extraction module.It is an object of the invention to be capable of from initial three-dimensional point cloud to be partitioned into quickly and accurately building object point cloud,
And fully automatically complete the extraction of contour of building line point cloud.
The technical solution adopted in the present invention is: the building segmentation of a kind of ground laser point cloud data is extracted with contour line
Method, it is characterised in that comprise the following steps:
Step 1 a: cloud is carried out vertical projection；
Input initial three-dimensional point cloud, carries out vertical projection to initial three-dimensional point cloud, highlights the some cloud distribution spy of different atural object
Point；
Step 2: generate 2-D gray image；
X, Y coordinate scope according to a cloud set up horizontal grid, add up each grid inner projection point number, and generate with this
2-D gray image, sets up contacting between three-dimensional point cloud and bidimensional image；
Step 3: utilize Otsu algorithm segmentation building object point cloud；
Initial three-dimensional point cloud is carried out building segmentation；According to building gray scale feature in 2-D gray image, make
With Otsu algorithm, image is split, be three-dimensional point cloud by video conversion again after image segmentation, building segmentation can be completed；
Step 4: principal component analysis；
Building object point cloud after segmentation is carried out principal component analysis, calculates the dimensional characteristics value of each point；
Step 5: optimal neighborhood calculates；
Under different scale, carry out principal component analysis respectively, calculate dimensional characteristics value under different scale, further according to entropy function,
Automatically calculate the optimal neighborhood of each point, draw the main dimension of each point, thus a cloud is divided into wire point cloud, planar point
Cloud, dispersion point cloud three class；
Step 6: contour line extracts；
According to geometry site, planar point cloud is carried out contour line extraction process；First determine its place for each point
Plane, then judge that this point is in the edge of plane or internal plane, the line segment formed with its neighborhood point according to the point in plane
Angle and the about 360 ° points filtered in plane, thus retain the contour line point cloud of each plane.
As preferably, described in step 1, initial three-dimensional point cloud is carried out vertical projection, be along Z by initial three-dimensional point cloud
Axial projection, on X, Y plane, only retains the X of each point, Y coordinate；Its projection formula is:
As preferably, the statistics each grid inner projection point number described in step 2, it is designated as subpoint density DoPP, adjusts
The span of the DoPP of whole all grid, by its linear stretch to 0-255, simulates the ash of each grid by the value after stretching
Degree, generates 2-D gray image.
As preferably, image is split by the Otsu algorithm described in step 3, is to calculate gray threshold T, with threshold value
Image is divided into target and background two class by T, and the inter-class variance g making them is maximum, and g is calculated by formula (2)；
G=ω0ω1(μ0-μ1)2(2)；
Wherein ω0、ω1It is two classes each accounting example, μ0、μ1It it is the average gray of two classes.
As preferably, described Otsu algorithm include following 2 improvement:
(1) sub solving method；Carrying out Otsu algorithm when, it is not that all of cloud data is the most once counted
Calculate, and be based on building facade distribution situation in whole scene, scene domain is divided, in each subfield scape
Carry out an Otsu algorithm respectively；Division principle is, as much as possible satisfied, in each subfield scape, contains building simultaneously
Facade data and other atural object data, especially ensure that the part that in building object point cloud, DoPP value is relatively small meets above-mentioned wanting
Asking, do so can be obviously improved the accuracy of point cloud segmentation；
Image L is respectively divided into m and n equal portions in x and y direction, obtains the scope of each subfield scape:
Lij=xi*yi(i=1,2 ..., m；J=1,2 ..., n) (3)；
At each LijInside carry out an Otsu algorithm, finally take the conjunction of the building facade data that each subfield scape is partitioned into
Collection, obtains final segmentation result；
(2) combine building facade point cloud and can form this feature of continuous print line segment after vertical projection, for leakage segmentation
Some cloud part, utilize region growing algorithm to compensate；Its concrete calculation process includes following sub-step:
Step 3.1: selected seed point；If the pixel grey scale that the projection of initial three-dimensional point cloud generates gray level image is g [i] [j],
After Otsu sub solving method algorithm process, pixel grey scale is G [i] [j], sets labelling array B [i] [j], records each pixel
Attribute, the corresponding same pixel of the identical subscript of three arrays；Travel through each pixel, if there being pixel p to meet G [px][py]>
0 and its four neighborhood or eight neighborhood have any one pixel q to meet G [qx][qy]=0 and g [qx][qy] > 0, then B [px][py]=
1, otherwise B [px][py]=0, wherein pixel p, the relation of q meet formula (4)；Choose all pixels of 1 of being labeled as seed
Point；
Step 3.2: setting regions condition of growth, i.e. threshold value T；If seed points pixel is p, its neighborhood point pixel is q, will plant
Son point G [px][py] and its neighborhood point g [qx][qy] make comparisons, if its business meets G [px][py]/g[qx][qy] < T, then by g [qx]
[qy] value imparting G [qx][qy], and by B [qx][qy] it is labeled as 1, if its business meets G [px][py]/g[qx][qy] >=T, then by B
[qx][qy] it is labeled as 0, when all q that p is corresponding all complete aforesaid operations, by B [px][py] it is labeled as 0；
Step 3.3: region increases；The pixel of B=1 is regarded as new seed points, and returns execution step 3.2；
Step 3.4: termination condition；All elements in array B is 0, i.e. labelling a little be 0, region increases
Terminate.
As preferably, the principal component analysis described in step 4 is by each point and the three-dimensional coordinate structure of neighborhood point thereof
Build covariance matrix, calculate three eigenvalues of matrix, analyze each point with this and belong to the probability a of three dimensional characteristics1D、
a2D、a3D；It implements process and includes following sub-step:
Step 4.1: according to some cloud density and a required precision, set the span [r of radius of neighbourhood rmin, rmax], and
Value is spaced；
Step 4.2: under the different radius of neighbourhood, carry out principal component analysis respectively；
First with each scanning element XiAnd neighborhood point νrThree-dimensional coordinate { Xi=(xi, yi, zi)|i∈vrStructure association
Variance matrix:
Matrix C is the matrix of a 3*3, whereinFor point set νrWeight
Heart coordinate, the concrete form of matrix M is:
Then three eigenvalue λ of Matrix C are calculated1、λ2、λ3, and according to λ1≥λ2≥λ3Rule arrange；
Step 4.3: calculate three dimensional characteristics of each point；
OrderThree dimensional characteristics of each point, the most one-dimensional wire is calculated according to formula (7)
Feature a1D, two dimension planar feature a2DWith 3 d discrete point feature a3D；Wherein a1D、a2D、a3DAnd be 1, in other words, a1D、a2D、
a3DRepresent scanning element respectively and belong to the probability of three dimensional characteristics；
As preferably, the optimal neighborhood described in step 5 calculates, and is the optimal neighborhood calculating each point according to entropy function,
See formula (8):
Ef(vr)=-a1Dln(a1D)-a2Dln(a2D)-a3Dln(a3D) (8)；
Wherein a1D、a2D、a3DIt it is each point obtained in the principal component analysis probability that belongs to three dimensional characteristics.
Entropy according to the neighborhood point set under the different r value of formula (8) calculating；When entropy takes minima, represent at this neighborhood
The main dimensional characteristics of this point lower is the most prominent, now the radius r of correspondence*It is the optimal radius of neighbourhood.
As preferably, described in step 6 according to geometry site, planar point cloud is carried out contour line extraction process；
It implements process and includes following sub-step:
Step 6.1: according to optimal neighborhood r*Determine the optimal neighborhood point set of X point
Step 6.2: determine point set place plane.According to RANCAC stochastic sampling unification algorism, fromIn randomly draw two
Individual differenceAnd repeatedly (number of repetition withSome number identical), calculate X,
Xi、Xj3 place plane equations are also added upIn remaining point to plane Euclidean distance and, choose and make distance and minimum flat
Face SxpqTwo neighborhood point Xp、Xq, by Xp、XqFromMiddle deletion；
Step 6.3: delete flat outer point.TraversalCalculate each XiWith Xp、Xq3 planes S determinedipqEquation,
Set angle threshold θ1, when plane SipqWith plane SxpqAngle less than θ1Time, assert XiIn plane SxpqOn, otherwise by XiFrom
Middle deletion.
Step 6.4: set XXpFor playing initial line, traversalCalculate each ∠ XPXXi, set angle threshold θ2, choose less than θ2
In make ∠ XPXXiMinimum neighborhood point X1'；
Step 6.5: judge X1' and XpDirection, if X1' at XpClockwise direction, just according to clockwise going to search
Rope；Otherwise, according to the next neighborhood point of search counterclockwise；
Step 6.6: the line segment that neighborhood point step 6.4 obtained and X are constituted has been set to initial line, repeats step 6.4 and seeks
Meet the next neighborhood point X of condition2',X3',…,Xn', until search is less than the neighborhood point meeting condition；
Step 6.7: calculateAngle and, and if approximate 360 °, be considered as putting down by an X
Point in face, is filtered, and is otherwise considered as marginal point, is retained.
Owing to traditional building point cloud segmentation method and contour line extracting method are in precision, speed, the suitability, automatization
Etc. aspect have deficiency in various degree, and cannot accomplish to take into account simultaneously, therefore the present invention is directed to problem above and changed
Enter, it is an advantage of the current invention that:
1, the building point cloud segmentation method that the present invention proposes takes into account speed and precision.
(1) present invention is in building point cloud segmentation, and three-dimensional point cloud is converted into bidimensional image, uses the side of overall segmentation
Method, it is to avoid the calculating put for each, thus improve calculating speed.
(2) for Otsu deficiency of precision in building point cloud segmentation, it is proposed that what sub solving method and region increased changes
Enter method, both improve the segmentation precision of building object point cloud, the most do not increased too much calculating, it is ensured that fast excellent of splitting speed
Point.
2, the building object point cloud contour line extracting method that the present invention proposes is a kind of full automatic method, it is not necessary to manually do
In advance, and the suitability is strong.The present invention can calculate the optimal radius of neighbourhood of each point automatically by principal component analysis and entropy function,
And a cloud is classified according to dimensional characteristics, the geometrical relationship intrinsic finally according to planar point cloud internal point automatically extracts profile
Line point cloud, is specially converted to angle threshold by distance threshold in algorithm designs, and closes with the relative position of neighborhood point with each point
System substitutes absolute positional relation, so for different cloud datas without the threshold value in amendment experiment, reaches the mesh automatically extracted
's.
Accompanying drawing explanation
The flow chart of Fig. 1: the embodiment of the present invention.
Detailed description of the invention
Understand and implement the present invention for the ease of those of ordinary skill in the art, below in conjunction with the accompanying drawings and embodiment is to this
Bright it is described in further detail, it will be appreciated that enforcement example described herein is merely to illustrate and explains the present invention, not
For limiting the present invention.
Ask for an interview Fig. 1, building segmentation and the contour line extracting method of a kind of ground laser point cloud data that the present invention provides,
Comprise the following steps:
Step 1 a: cloud is carried out vertical projection；
Input initial three-dimensional point cloud, carries out vertical projection to initial three-dimensional point cloud, highlights the some cloud distribution spy of different atural object
Point；
Initial three-dimensional point cloud is carried out vertical projection, is that initial three-dimensional point cloud is projected to X, Y plane along Z axis, only protects
Stay the X of each point, Y coordinate；Its projection formula is:
Step 2: generate 2-D gray image；
X, Y coordinate scope according to a cloud set up horizontal grid, add up each grid inner projection point number, are designated as subpoint
Density DoPP, adjusts the span of the DoPP of all grid, by its linear stretch to 0-255, every with the value simulation after stretching
The gray scale of individual grid, and generate 2-D gray image with this, set up contacting between three-dimensional point cloud and bidimensional image；
Step 3 a: cloud is carried out initial building segmentation；
According to partial building gray scale feature in 2-D gray image, use Otsu algorithm that image is split,
The cardinal principle of Otsu algorithm is to calculate gray threshold T, with threshold value T, image is divided into target and background two class, makes theirs
Inter-class variance g is maximum, and g is calculated by formula (2).Wherein ω0、ω1It is two classes each accounting example, μ0、μ1It is the average of two classes
Gray scale.
G=ω0ω1(μ0-μ1)2(2)；
It is three-dimensional point cloud by video conversion again after image segmentation, building initial segmentation can be completed.Generally, just
Begin the available building main part of segmentation, but there is partial error cut zone.
Step 4 a: cloud is carried out accurate building segmentation；
In conjunction with in initial segmentation result, the characteristic distributions in erroneous segmentation region, use the Otsu algorithm of a kind of improvement to carry out
Accurate Segmentation.Innovatory algorithm comprises two parts, sub solving method and region and increases compensation leakage cut zone.Through this step, can
With accurately, complete Ground Split and go out to build object point cloud, reject the non-targeted atural object cloud data such as large number of ground, trees.
(1) sub solving method；Carrying out Otsu algorithm when, it is not that all of cloud data is the most once counted
Calculate, and be based on building facade distribution situation in whole scene, scene domain is divided, in each subfield scape
Carry out an Otsu algorithm respectively；Division principle is, as much as possible satisfied, in each subfield scape, contains building simultaneously
Facade data and other atural object data, especially ensure that the part that in building object point cloud, DoPP value is relatively small meets above-mentioned wanting
Asking, do so can be obviously improved the accuracy of point cloud segmentation；
Image L is respectively divided into m and n equal portions in x and y direction, obtains the scope of each subfield scape:
Lij=xi*yi(i=1,2 ..., m；J=1,2 ..., n) (3)；
At each LijInside carry out an Otsu algorithm, finally take the conjunction of the building facade data that each subfield scape is partitioned into
Collection, obtains final segmentation result；
(2) combine building facade point cloud and can form this feature of continuous print line segment after vertical projection, for leakage segmentation
Some cloud part, utilize region growing algorithm to compensate；Its concrete calculation process includes following sub-step:
Step 4.1: selected seed point；If the pixel grey scale that the projection of initial three-dimensional point cloud generates gray level image is g [i] [j],
After Otsu sub solving method algorithm process, pixel grey scale is G [i] [j], sets labelling array B [i] [j], records each pixel
Attribute, the corresponding same pixel of the identical subscript of three arrays；Travel through each pixel, if there being pixel p to meet G [px][py]>
0 and its four neighborhood or eight neighborhood have any one pixel q to meet G [qx][qy]=0 and g [qx][qy] > 0, then B [px][py]=
1, otherwise B [px][py]=0, wherein pixel p, the relation of q meet formula (4)；Choose all pixels of 1 of being labeled as seed
Point；
Step 4.2: setting regions condition of growth, i.e. threshold value T；If seed points pixel is p, its neighborhood point pixel is q, will plant
Son point G [px][py] and its neighborhood point g [qx][qy] make comparisons, if its business meets G [px][py]/g[qx][qy] < T, then by g [qx]
[qy] value imparting G [qx][qy], and by B [qx][qy] it is labeled as 1, if its business meets G [px][py]/g[qx][qy] >=T, then by B
[qx][qy] it is labeled as 0, when all q that p is corresponding all complete aforesaid operations, by B [px][py] it is labeled as 0；
Step 4.3: region increases；The pixel of B=1 is regarded as new seed points, and returns execution step 3.2；
Step 4.4: termination condition；All elements in array B is 0, i.e. labelling a little be 0, region increases
Terminate.
Step 5: principal component analysis；
The first step that contour line extracts is according to point and the position relationship of its neighborhood point, a cloud is divided into wire point cloud,
Planar point cloud, dispersion point cloud three class.Complicated unordered cloud data can be analyzed calculating by principal component analysis, and then obtains
Three dimensions X, relation between Y, Z, determine the type of each point.The main calculation process of principal component analysis is by each
The three-dimensional coordinate of point and neighborhood point thereof builds covariance matrix, calculates three eigenvalues of matrix, analyzes each point with this
Belong to the probability a of three dimensional characteristics1D、a2D、a3D；It implements process and includes following sub-step:
Step 5.1: according to some cloud density and a required precision, set the span [r of radius of neighbourhood rmin, rmax], and
Value is spaced；
Step 5.2: under the different radius of neighbourhood, carry out principal component analysis respectively；
First with each scanning element XiAnd neighborhood point νrThree-dimensional coordinate { Xi=(xi, yi, zi)|i∈vrStructure association
Variance matrix:
Matrix C is the matrix of a 3*3, whereinFor point set νrWeight
Heart coordinate, the concrete form of matrix M is:
Then three eigenvalue λ of Matrix C are calculated1、λ2、λ3, and according to λ1≥λ2≥λ3Rule arrange；
Step 5.3: calculate three dimensional characteristics of each point；
OrderThree dimensional characteristics of each point, the most one-dimensional wire is calculated according to formula (7)
Feature a1D, two dimension planar feature a2DWith 3 d discrete point feature a3D；Wherein a1D、a2D、a3DAnd be 1, in other words, a1D、a2D、
a3DRepresent scanning element respectively and belong to the probability of three dimensional characteristics；
Step 6: optimal neighborhood calculates；
Under different scale, carry out principal component analysis respectively, calculate dimensional characteristics value under different scale, further according to entropy function,
Automatically calculate the optimal neighborhood of each point, draw the main dimension of each point, thus a cloud is divided into wire point cloud, planar point
Cloud, dispersion point cloud three class；
When principal component analysis calculates the dimensional characteristics of each point, the dimensional characteristics of point can be along with the change of neighborhood point set
Change, for the cloud data of zones of different, the most true due to the difference of the degree of depth of each point, some cloud density and neighborhood point quantity
Qualitative, we are difficult to obtain the most correct feature point set under a certain fixing yardstick.To this end, calculate each point according to entropy function
Optimal neighborhood, see formula (8)；
Ef(vr)=-a1Dln(a1D)-a2Dln(a2D)-a3Dln(a3D) (8)；
Entropy according to the neighborhood point set under the different r value of formula (8) calculating；When entropy takes minima, represent at this neighborhood
The main dimensional characteristics of this point lower is the most prominent, now the radius r of correspondence*It is the optimal radius of neighbourhood.
Step 7: contour line extracts；
According to geometry site, planar point cloud is carried out contour line extraction process；First determine its place for each point
Plane, then judge that this point is in the edge of plane or internal plane, the line segment formed with its neighborhood point according to the point in plane
Angle and the about 360 ° points filtered in plane, thus retain the contour line point cloud of each plane.
According to geometry site, planar point cloud is carried out contour line extraction process；It implements process and includes following
Sub-step:
Step 7.1: according to optimal neighborhood r*Determine the optimal neighborhood point set of X point
Step 7.2: determine point set place plane.According to RANCAC stochastic sampling unification algorism, fromIn randomly draw two
Individual differenceAnd repeatedly (number of repetition withSome number identical), calculate X, Xi、
Xj3 place plane equations are also added upIn remaining point to plane Euclidean distance and, choose and make distance and minimum plane
SxpqTwo neighborhood point Xp、Xq, by Xp、XqFromMiddle deletion；
Step 7.3: delete flat outer point.TraversalCalculate each XiWith Xp、Xq3 planes S determinedipqEquation,
Set angle threshold θ1, when plane SipqWith plane SxpqAngle less than θ1Time, assert XiIn plane SxpqOn, otherwise by XiFrom
Middle deletion.
Step 7.4: set XXpFor playing initial line, traversalCalculate each ∠ XPXXi, set angle threshold θ2, choose less than θ2
In make ∠ XPXXiMinimum neighborhood point X1'；
Step 7.5: judge X1' and XpDirection, if X1' at XpClockwise direction, just according to clockwise going to search
Rope；Otherwise, according to the next neighborhood point of search counterclockwise；
Step 7.6: the line segment that neighborhood point step 7.4 obtained and X are constituted has been set to initial line, repeats step 7.4 and seeks
Meet the next neighborhood point X of condition2',X3',…,Xn', until search is less than the neighborhood point meeting condition；
Step 7.7: calculateAngle and, and if approximate 360 °, be considered as putting down by an X
Point in face, is filtered, and is otherwise considered as marginal point, is retained.
It should be appreciated that the part that this specification does not elaborates belongs to prior art.
It should be appreciated that the above-mentioned description for preferred embodiment is more detailed, can not therefore be considered this
The restriction of invention patent protection scope, those of ordinary skill in the art, under the enlightenment of the present invention, is weighing without departing from the present invention
Profit requires under the ambit protected, it is also possible to make replacement or deformation, within each falling within protection scope of the present invention, this
The bright scope that is claimed should be as the criterion with claims.
Claims (8)
1. the building of a ground laser point cloud data is split and contour line extracting method, it is characterised in that include following step
Rapid:
Step 1 a: cloud is carried out vertical projection；
Input initial three-dimensional point cloud, carries out vertical projection to initial three-dimensional point cloud, highlights the some cloud characteristic distributions of different atural object；
Step 2: generate 2-D gray image；
X, Y coordinate scope according to a cloud set up horizontal grid, add up each grid inner projection point number, and generate two dimension with this
Gray level image, sets up contacting between three-dimensional point cloud and bidimensional image；
Step 3: utilize Otsu algorithm segmentation building object point cloud；
Initial three-dimensional point cloud is carried out building segmentation；According to building gray scale feature in 2-D gray image, use
Image is split by Otsu algorithm, is three-dimensional point cloud by video conversion again after image segmentation, can complete building segmentation；
Step 4: principal component analysis；
Building object point cloud after segmentation is carried out principal component analysis, calculates the dimensional characteristics value of each point；
Step 5: optimal neighborhood calculates；
Under different scale, carry out principal component analysis respectively, calculate dimensional characteristics value under different scale, further according to entropy function, automatically
Calculate the optimal neighborhood of each point, draw the main dimension of each point, thus a cloud is divided into wire point cloud, planar point cloud, dissipates
Disorderly some cloud three class；
Step 6: contour line extracts；
According to geometry site, planar point cloud is carried out contour line extraction process；The plane at its place is first determined for each point,
Judge that this point is in the edge of plane or internal plane again, the line segment angle that formed according to the point in plane and its neighborhood point and
It is about 360 ° of points filtered in plane, thus retains the contour line point cloud of each plane.
Building segmentation and contour line extracting method, its feature of ground the most according to claim 1 laser point cloud data
It is: described in step 1, initial three-dimensional point cloud is carried out vertical projection, is that initial three-dimensional point cloud is put down to X, Y along Z axis projection
On face, only retain the X of each point, Y coordinate；Its projection formula is:
Building segmentation and contour line extracting method, its feature of ground the most according to claim 1 laser point cloud data
It is: the statistics each grid inner projection point number described in step 2, is designated as subpoint density DoPP, adjusts all grid
The span of DoPP, by its linear stretch to 0-255, simulates the gray scale of each grid, generates two dimension ash by the value after stretching
Degree image.
Building segmentation and contour line extracting method, its feature of ground the most according to claim 1 laser point cloud data
It is: image is split by the Otsu algorithm described in step 3, calculates gray threshold T, with threshold value T, image is divided into mesh
Mark and background two class, the inter-class variance g making them is maximum, and g is calculated by formula (2)；
G=ω0ω1(μ0-μ1)2(2)；
Wherein ω0、ω1It is two classes each accounting example, μ0、μ1It it is the average gray of two classes.
Building segmentation and contour line extracting method, its feature of ground the most according to claim 4 laser point cloud data
Be, described Otsu algorithm include following 2 improvement:
(1) sub solving method；Carrying out Otsu algorithm when, it is not that all of cloud data is the most once calculated, and
It is based on building facade distribution situation in whole scene, scene domain is divided, in each subfield scape respectively
Carry out an Otsu algorithm；Division principle is, as much as possible satisfied, in each subfield scape, contains building facade simultaneously
Data and other atural object data, especially ensure that the part that in building object point cloud, DoPP value is relatively small meets above-mentioned requirements, this
Sample does the accuracy that can be obviously improved point cloud segmentation；
Image L is respectively divided into m and n equal portions in x and y direction, obtains the scope of each subfield scape:
Lij=xi*yi(i=1,2 ..., m；J=1,2 ..., n) (3)；
At each LijInside carry out an Otsu algorithm, finally take the intersection of the building facade data that each subfield scape is partitioned into,
Obtain final segmentation result；
(2) combine building facade point cloud and can form this feature of continuous print line segment after vertical projection, for the point of leakage segmentation
Cloud part, utilizes region growing algorithm to compensate；Its concrete calculation process includes following sub-step:
Step 3.1: selected seed point；If the pixel grey scale that the projection of initial three-dimensional point cloud generates gray level image is g [i] [j], pass through
After Otsu sub solving method algorithm process, pixel grey scale is G [i] [j], sets labelling array B [i] [j], records the genus of each pixel
Property, the corresponding same pixel of identical subscript of three arrays；Travel through each pixel, if there being pixel p to meet G [px][py] > 0 and
Its four neighborhood or eight neighborhood have any one pixel q to meet G [qx][qy]=0 and g [qx][qy] > 0, then B [px][py]=1,
Otherwise B [px][py]=0, wherein pixel p, the relation of q meet formula (4)；Choose all pixels of 1 of being labeled as seed points；
Step 3.2: setting regions condition of growth, i.e. threshold value T；If seed points pixel is p, its neighborhood point pixel is q, by seed points
G[px][py] and its neighborhood point g [qx][qy] make comparisons, if its business meets G [px][py]/g[qx][qy] < T, then by g [qx][qy]
Value gives G [qx][qy], and by B [qx][qy] it is labeled as 1, if its business meets G [px][py]/g[qx][qy] >=T, then by B [qx]
[qy] it is labeled as 0, when all q that p is corresponding all complete aforesaid operations, by B [px][py] it is labeled as 0；
Step 3.3: region increases；The pixel of B=1 is regarded as new seed points, and returns execution step 3.2；
Step 3.4: termination condition；All elements in array B is 0, i.e. labelling a little be 0, region increases and terminates.
Building segmentation and contour line extracting method, its feature of ground the most according to claim 1 laser point cloud data
It is: the principal component analysis described in step 4 is by the three-dimensional coordinate of each point and neighborhood point thereof is built covariance matrix,
Calculate three eigenvalues of matrix, analyze each point with this and belong to the probability a of three dimensional characteristics1D、a2D、a3D；It is concrete
The process of realization includes following sub-step:
Step 4.1: according to some cloud density and a required precision, set the span [r of radius of neighbourhood rmin, rmax], and value
Interval；
Step 4.2: under the different radius of neighbourhood, carry out principal component analysis respectively；
First with each scanning element XiAnd neighborhood point νrThree-dimensional coordinate { Xi=(xi, yi, zi)|i∈vrStructure covariance square
Battle array:
Matrix C is the matrix of a 3*3, wherein For point set νrCenter of gravity sit
Mark, the concrete form of matrix M is:
Then three eigenvalue λ of Matrix C are calculated1、λ2、λ3, and according to λ1≥λ2≥λ3Rule arrange；
Step 4.3: calculate three dimensional characteristics of each point；
OrderThree dimensional characteristics of each point, the most one-dimensional line feature is calculated according to formula (7)
a1D, two dimension planar feature a2DWith 3 d discrete point feature a3D；Wherein a1D、a2D、a3DAnd be 1, in other words, a1D、a2D、a3DPoint
Do not represent scanning element and belong to the probability of three dimensional characteristics；
Building segmentation and contour line extracting method, its feature of ground the most according to claim 6 laser point cloud data
It is: optimal neighborhood described in step 5 calculates, is the optimal neighborhood calculating each point according to entropy function, see formula (8):
Ef(vr)=-a1Dln(a1D)-a2Dln(a2D)-a3Dln(a3D) (8)；
Wherein a1D、a2D、a3DIt it is each point obtained in the principal component analysis probability that belongs to three dimensional characteristics；
Entropy according to the neighborhood point set under the different r value of formula (8) calculating；When entropy takes minima, representing should under this neighborhood
The main dimensional characteristics of point is the most prominent, now corresponding radius r*It is the optimal radius of neighbourhood.
Building segmentation and contour line extracting method, its feature of ground the most according to claim 7 laser point cloud data
Be: described in step 6 according to geometry site, planar point cloud is carried out contour line extraction process；It implemented
Journey includes following sub-step:
Step 6.1: according to optimal neighborhood r*Determine the optimal neighborhood point set of X point
Step 6.2: determine point set place plane；According to RANCAC stochastic sampling unification algorism, fromIn randomly draw two not
Same pointAnd repeatedly, number of repetition withSome number identical, calculate X, Xi、XjThree
Point place plane equation is also added upIn remaining point to plane Euclidean distance and, choose and make distance and minimum plane Sxpq's
Two neighborhood point Xp、Xq, by Xp、XqFromMiddle deletion；
Step 6.3: delete flat outer point；TraversalCalculate each XiWith Xp、Xq3 planes S determinedipqEquation, sets
Angle threshold θ1, when plane SipqWith plane SxpqAngle less than θ1Time, assert XiIn plane SxpqOn, otherwise by XiFromIn delete
Remove；
Step 6.4: set XXpFor playing initial line, traversalCalculate each ∠ XPXXi, set angle threshold θ2, choose less than θ2In make
∠XPXXiMinimum neighborhood point X1'；
Step 6.5: judge X1' and XpDirection, if X1' at XpClockwise direction, just according to clockwise direction removal search；No
Then, according to the next neighborhood point of search counterclockwise；
Step 6.6: the line segment that neighborhood point step 6.4 obtained and X are constituted has been set to initial line, repeats step 6.4 and seeks to meet
The next neighborhood point X of condition2',X3',…,Xn', until search is less than the neighborhood point meeting condition；
Step 6.7: calculateAngle and, and if approximate 360 °, an X is considered as in plane
Point, filtered, be otherwise considered as marginal point, retained.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610393443.0A CN106056614A (en) | 2016-06-03 | 2016-06-03 | Building segmentation and contour line extraction method of ground laser point cloud data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610393443.0A CN106056614A (en) | 2016-06-03 | 2016-06-03 | Building segmentation and contour line extraction method of ground laser point cloud data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN106056614A true CN106056614A (en) | 2016-10-26 |
Family
ID=57170323
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610393443.0A Pending CN106056614A (en) | 2016-06-03 | 2016-06-03 | Building segmentation and contour line extraction method of ground laser point cloud data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106056614A (en) |
Cited By (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407925A (en) * | 2016-09-09 | 2017-02-15 | 厦门大学 | Automatic extracting method of laser scanning point cloud tree based on local interval maximal value |
| CN106846494A (en) * | 2017-01-16 | 2017-06-13 | 青岛海大新星软件咨询有限公司 | Oblique photograph three-dimensional building thing model automatic single-body algorithm |
| CN107862738A (en) * | 2017-11-28 | 2018-03-30 | 武汉大学 | One kind carries out doors structure three-dimensional rebuilding method based on mobile laser measurement point cloud |
| CN108241871A (en) * | 2017-12-27 | 2018-07-03 | 华北水利水电大学 | Laser point cloud and visual fusion data classification method based on multiple features |
| CN108734769A (en) * | 2017-04-17 | 2018-11-02 | 宏达国际电子股份有限公司 | Threedimensional model analysis method, electronic device and non-transient computer readable media |
| CN108984599A (en) * | 2018-06-01 | 2018-12-11 | 青岛秀山移动测量有限公司 | A kind of vehicle-mounted laser point cloud road surface extracting method referred to using driving trace |
| CN108986024A (en) * | 2017-06-03 | 2018-12-11 | 西南大学 | A kind of regularly arranged processing method of laser point cloud based on grid |
| CN109118500A (en) * | 2018-07-16 | 2019-01-01 | 重庆大学产业技术研究院 | A kind of dividing method of the Point Cloud Data from Three Dimension Laser Scanning based on image |
| CN109325953A (en) * | 2018-02-05 | 2019-02-12 | 黑龙江科技大学 | A kind of determination method of extensive dense point cloud normal |
| CN110047036A (en) * | 2019-04-22 | 2019-07-23 | 重庆交通大学 | Territorial laser scanning data building facade extracting method based on polar coordinates grid |
| CN110223267A (en) * | 2018-03-14 | 2019-09-10 | 浙江大学山东工业技术研究院 | The recognition methods of refractory brick deep defects based on height histogram divion |
| CN110377776A (en) * | 2018-07-23 | 2019-10-25 | 北京京东尚科信息技术有限公司 | The method and apparatus for generating point cloud data |
| CN110969624A (en) * | 2019-11-07 | 2020-04-07 | 哈尔滨工程大学 | Laser radar three-dimensional point cloud segmentation method |
| CN111179297A (en) * | 2018-11-13 | 2020-05-19 | 北京航空航天大学 | Method, device and system for generating multiple outlines of point cloud |
| CN111754385A (en) * | 2019-03-26 | 2020-10-09 | 深圳中科飞测科技有限公司 | Data point model processing method and system, detection method and system and readable medium |
| CN111783721A (en) * | 2020-07-13 | 2020-10-16 | 湖北亿咖通科技有限公司 | Lane line extraction method of laser point cloud and electronic equipment |
| CN111783722A (en) * | 2020-07-13 | 2020-10-16 | 湖北亿咖通科技有限公司 | Lane line extraction method of laser point cloud and electronic equipment |
| CN111833286A (en) * | 2019-03-26 | 2020-10-27 | 深圳中科飞测科技有限公司 | Point cloud processing method and system, detection method and system and readable medium |
| WO2020234678A1 (en) * | 2019-05-21 | 2020-11-26 | International Business Machines Corporation | Progressive 3d point cloud segmentation into object and background from tracking sessions |
| CN112132969A (en) * | 2020-09-01 | 2020-12-25 | 济南市房产测绘研究院(济南市房屋安全检测鉴定中心) | Vehicle-mounted laser point cloud building target classification method |
| CN112232248A (en) * | 2020-10-22 | 2021-01-15 | 中国人民解放军战略支援部队信息工程大学 | Method and device for extracting plane features of multi-line LiDAR point cloud data |
| CN112488010A (en) * | 2020-12-05 | 2021-03-12 | 武汉中海庭数据技术有限公司 | High-precision target extraction method and system based on unmanned aerial vehicle point cloud data |
| CN112529952A (en) * | 2020-12-15 | 2021-03-19 | 武汉万集信息技术有限公司 | Object volume measuring method and device and electronic equipment |
| CN112595258A (en) * | 2020-11-23 | 2021-04-02 | 扆亮海 | Ground object contour extraction method based on ground laser point cloud |
| CN112633657A (en) * | 2020-12-16 | 2021-04-09 | 中冶建筑研究总院有限公司 | Construction quality management method, device, equipment and storage medium |
| CN113409332A (en) * | 2021-06-11 | 2021-09-17 | 电子科技大学 | Building plane segmentation method based on three-dimensional point cloud |
| CN113591597A (en) * | 2021-07-07 | 2021-11-02 | 东莞市鑫泰仪器仪表有限公司 | Intelligent public security information system based on thermal imaging |
| CN113838072A (en) * | 2021-11-01 | 2021-12-24 | 江苏集萃智能光电系统研究所有限公司 | High-dynamic star atlas image segmentation method |
| CN114820986A (en) * | 2022-05-13 | 2022-07-29 | 广西微车检智能科技有限公司 | Trailer outline parameter measuring method based on laser radar |
| WO2022198745A1 (en) * | 2021-03-24 | 2022-09-29 | Shanghai Huizi Cosmetics Co., Ltd. | Method and system for extracting nail contours |
| US11532120B2 (en) | 2017-10-06 | 2022-12-20 | Interdigital Vc Holdings, Inc. | Method and device for hole filling of a point cloud |
| CN115880325A (en) * | 2022-12-07 | 2023-03-31 | 重庆市地理信息和遥感应用中心 | Building outline automatic extraction method based on point cloud dimension and spatial distance clustering |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020637A (en) * | 2012-11-22 | 2013-04-03 | 北京航空航天大学 | Point cloud data segmentation method on top surface of building based on K-plane algorithm |
| CN104484668A (en) * | 2015-01-19 | 2015-04-01 | 武汉大学 | Unmanned aerial vehicle multi-overlapped-remote-sensing-image method for extracting building contour line |
-
2016
- 2016-06-03 CN CN201610393443.0A patent/CN106056614A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020637A (en) * | 2012-11-22 | 2013-04-03 | 北京航空航天大学 | Point cloud data segmentation method on top surface of building based on K-plane algorithm |
| CN104484668A (en) * | 2015-01-19 | 2015-04-01 | 武汉大学 | Unmanned aerial vehicle multi-overlapped-remote-sensing-image method for extracting building contour line |
Non-Patent Citations (2)
| Title |
|---|
| 秦家鑫等: "一种建筑物点云轮廓线的自动提取方法", 《遥感信息》 * |
| 秦家鑫等: "基于Otsu的建筑物点云分割改进算法", 《地理空间信息》 * |
Cited By (48)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106407925B (en) * | 2016-09-09 | 2019-09-27 | 厦门大学 | Laser scanning point cloud trees extraction method based on local section maximum |
| CN106407925A (en) * | 2016-09-09 | 2017-02-15 | 厦门大学 | Automatic extracting method of laser scanning point cloud tree based on local interval maximal value |
| CN106846494A (en) * | 2017-01-16 | 2017-06-13 | 青岛海大新星软件咨询有限公司 | Oblique photograph three-dimensional building thing model automatic single-body algorithm |
| CN108734769A (en) * | 2017-04-17 | 2018-11-02 | 宏达国际电子股份有限公司 | Threedimensional model analysis method, electronic device and non-transient computer readable media |
| CN108986024A (en) * | 2017-06-03 | 2018-12-11 | 西南大学 | A kind of regularly arranged processing method of laser point cloud based on grid |
| CN108986024B (en) * | 2017-06-03 | 2024-01-23 | 西南大学 | Grid-based laser point cloud rule arrangement processing method |
| US11532120B2 (en) | 2017-10-06 | 2022-12-20 | Interdigital Vc Holdings, Inc. | Method and device for hole filling of a point cloud |
| CN107862738A (en) * | 2017-11-28 | 2018-03-30 | 武汉大学 | One kind carries out doors structure three-dimensional rebuilding method based on mobile laser measurement point cloud |
| CN107862738B (en) * | 2017-11-28 | 2019-10-11 | 武汉大学 | One kind carrying out doors structure three-dimensional rebuilding method based on mobile laser measurement point cloud |
| CN108241871A (en) * | 2017-12-27 | 2018-07-03 | 华北水利水电大学 | Laser point cloud and visual fusion data classification method based on multiple features |
| CN109325953B (en) * | 2018-02-05 | 2021-09-21 | 黑龙江科技大学 | Method for determining large-scale dense point cloud normal |
| CN109325953A (en) * | 2018-02-05 | 2019-02-12 | 黑龙江科技大学 | A kind of determination method of extensive dense point cloud normal |
| CN110223267A (en) * | 2018-03-14 | 2019-09-10 | 浙江大学山东工业技术研究院 | The recognition methods of refractory brick deep defects based on height histogram divion |
| CN108984599A (en) * | 2018-06-01 | 2018-12-11 | 青岛秀山移动测量有限公司 | A kind of vehicle-mounted laser point cloud road surface extracting method referred to using driving trace |
| CN108984599B (en) * | 2018-06-01 | 2021-08-20 | 青岛秀山移动测量有限公司 | Vehicle-mounted laser point cloud road surface extraction method using travel track reference |
| CN109118500B (en) * | 2018-07-16 | 2022-05-10 | 重庆大学产业技术研究院 | Image-based three-dimensional laser scanning point cloud data segmentation method |
| CN109118500A (en) * | 2018-07-16 | 2019-01-01 | 重庆大学产业技术研究院 | A kind of dividing method of the Point Cloud Data from Three Dimension Laser Scanning based on image |
| CN110377776A (en) * | 2018-07-23 | 2019-10-25 | 北京京东尚科信息技术有限公司 | The method and apparatus for generating point cloud data |
| CN111179297A (en) * | 2018-11-13 | 2020-05-19 | 北京航空航天大学 | Method, device and system for generating multiple outlines of point cloud |
| CN111179297B (en) * | 2018-11-13 | 2023-09-19 | 北京航空航天大学 | Multi-contour generation method, device and system of point cloud |
| CN111754385A (en) * | 2019-03-26 | 2020-10-09 | 深圳中科飞测科技有限公司 | Data point model processing method and system, detection method and system and readable medium |
| CN111833286A (en) * | 2019-03-26 | 2020-10-27 | 深圳中科飞测科技有限公司 | Point cloud processing method and system, detection method and system and readable medium |
| CN111754385B (en) * | 2019-03-26 | 2024-06-11 | 深圳中科飞测科技股份有限公司 | Data point model processing method and system, detection method and system and readable medium |
| CN110047036A (en) * | 2019-04-22 | 2019-07-23 | 重庆交通大学 | Territorial laser scanning data building facade extracting method based on polar coordinates grid |
| GB2598512B (en) * | 2019-05-21 | 2022-10-05 | Ibm | Progressive 3D point cloud segmentation into object and background from tracking sessions |
| WO2020234678A1 (en) * | 2019-05-21 | 2020-11-26 | International Business Machines Corporation | Progressive 3d point cloud segmentation into object and background from tracking sessions |
| GB2598512A (en) * | 2019-05-21 | 2022-03-02 | Ibm | Progressive 3D point cloud segmentation into object and background from tracking sessions |
| CN110969624A (en) * | 2019-11-07 | 2020-04-07 | 哈尔滨工程大学 | Laser radar three-dimensional point cloud segmentation method |
| CN111783722A (en) * | 2020-07-13 | 2020-10-16 | 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 |
| CN111783721A (en) * | 2020-07-13 | 2020-10-16 | 湖北亿咖通科技有限公司 | Lane line extraction method of laser point cloud and electronic equipment |
| CN112132969B (en) * | 2020-09-01 | 2023-10-10 | 济南市房产测绘研究院(济南市房屋安全检测鉴定中心) | Vehicle-mounted laser point cloud building target classification method |
| CN112132969A (en) * | 2020-09-01 | 2020-12-25 | 济南市房产测绘研究院(济南市房屋安全检测鉴定中心) | Vehicle-mounted laser point cloud building target classification method |
| CN112232248A (en) * | 2020-10-22 | 2021-01-15 | 中国人民解放军战略支援部队信息工程大学 | Method and device for extracting plane features of multi-line LiDAR point cloud data |
| CN112595258A (en) * | 2020-11-23 | 2021-04-02 | 扆亮海 | Ground object contour extraction method based on ground laser point cloud |
| CN112595258B (en) * | 2020-11-23 | 2022-04-22 | 湖南航天智远科技有限公司 | Ground object contour extraction method based on ground laser point cloud |
| CN112488010A (en) * | 2020-12-05 | 2021-03-12 | 武汉中海庭数据技术有限公司 | High-precision target extraction method and system based on unmanned aerial vehicle point cloud data |
| CN112529952A (en) * | 2020-12-15 | 2021-03-19 | 武汉万集信息技术有限公司 | Object volume measuring method and device and electronic equipment |
| CN112529952B (en) * | 2020-12-15 | 2023-11-14 | 武汉万集光电技术有限公司 | Object volume measurement method and device and electronic equipment |
| CN112633657B (en) * | 2020-12-16 | 2024-05-14 | 中冶建筑研究总院有限公司 | Construction quality management method, device, equipment and storage medium |
| CN112633657A (en) * | 2020-12-16 | 2021-04-09 | 中冶建筑研究总院有限公司 | Construction quality management method, device, equipment and storage medium |
| WO2022198745A1 (en) * | 2021-03-24 | 2022-09-29 | Shanghai Huizi Cosmetics Co., Ltd. | Method and system for extracting nail contours |
| CN113409332A (en) * | 2021-06-11 | 2021-09-17 | 电子科技大学 | Building plane segmentation method based on three-dimensional point cloud |
| CN113591597A (en) * | 2021-07-07 | 2021-11-02 | 东莞市鑫泰仪器仪表有限公司 | Intelligent public security information system based on thermal imaging |
| CN113838072A (en) * | 2021-11-01 | 2021-12-24 | 江苏集萃智能光电系统研究所有限公司 | High-dynamic star atlas image segmentation method |
| CN114820986A (en) * | 2022-05-13 | 2022-07-29 | 广西微车检智能科技有限公司 | Trailer outline parameter measuring method based on laser radar |
| CN114820986B (en) * | 2022-05-13 | 2024-04-09 | 广西微车检智能科技有限公司 | Laser radar-based trailer outline parameter measurement method |
| CN115880325A (en) * | 2022-12-07 | 2023-03-31 | 重庆市地理信息和遥感应用中心 | Building outline automatic extraction method based on point cloud dimension and spatial distance clustering |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106056614A (en) | Building segmentation and contour line extraction method of ground laser point cloud data | |
| Jin et al. | Deep learning: individual maize segmentation from terrestrial lidar data using faster R-CNN and regional growth algorithms | |
| CN104049245B (en) | Urban building change detection method based on LiDAR point cloud spatial difference analysis | |
| CN108109139B (en) | Airborne LIDAR three-dimensional building detection method based on gray voxel model | |
| CN106372277B (en) | Method for optimizing variation function model in forest land index space-time estimation | |
| CN109711410A (en) | Three-dimensional object rapid segmentation and identification method, device and system | |
| CN105469098A (en) | Precise LINDAR data ground object classification method based on adaptive characteristic weight synthesis | |
| CN107316048A (en) | Point cloud classifications method and device | |
| CN106199557A (en) | A kind of airborne laser radar data vegetation extracting method | |
| Marinelli et al. | A novel approach to 3-D change detection in multitemporal LiDAR data acquired in forest areas | |
| CN104091321A (en) | Multi-level-point-set characteristic extraction method applicable to ground laser radar point cloud classification | |
| CN103679655A (en) | LiDAR point cloud filter method based on gradient and area growth | |
| CN103745441A (en) | Method of filtering airborne LiDAR (Light Detection and Ranging) point cloud | |
| Stone et al. | Determining an optimal model for processing lidar data at the plot level: results for a Pinus radiata plantation in New South Wales, Australia | |
| CN110532963B (en) | Vehicle-mounted laser radar point cloud driven road marking accurate extraction method | |
| CN105809194A (en) | Method for translating SAR image into optical image | |
| CN112257605A (en) | Three-dimensional target detection method, system and device based on self-labeling training sample | |
| CN110363771B (en) | Isolation guardrail shape point extraction method and device based on three-dimensional point cloud data | |
| CN103927557B (en) | LIDAR data ground object classification method based on layered fuzzy evidence synthesis | |
| CN114037836A (en) | Method for applying artificial intelligence recognition technology to three-dimensional power transmission and transformation engineering measurement and calculation | |
| CN107292039B (en) | UUV bank patrolling profile construction method based on wavelet clustering | |
| Ayrey et al. | Ecologically-based metrics for assessing structure in developing area-based, enhanced forest inventories from LiDAR | |
| Pawłuszek et al. | Landslides identification using airborne laser scanning data derived topographic terrain attributes and support vector machine classification | |
| CN117765006A (en) | Multi-level dense crown segmentation method based on unmanned aerial vehicle image and laser point cloud | |
| Cherlinka | Using geostatistics, DEM and remote sensing to clarify soil cover maps of Ukraine |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20161026 |
|
| RJ01 | Rejection of invention patent application after publication |