CN108986048A - Based on the quick compound filter processing method of line laser structured light three-dimensional point cloud - Google Patents
Based on the quick compound filter processing method of line laser structured light three-dimensional point cloud Download PDFInfo
- Publication number
- CN108986048A CN108986048A CN201810787767.1A CN201810787767A CN108986048A CN 108986048 A CN108986048 A CN 108986048A CN 201810787767 A CN201810787767 A CN 201810787767A CN 108986048 A CN108986048 A CN 108986048A
- Authority
- CN
- China
- Prior art keywords
- point cloud
- cloud
- point
- filtering
- noise
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 150000001875 compounds Chemical class 0.000 title claims abstract description 15
- 238000003672 processing method Methods 0.000 title claims abstract description 9
- 238000001914 filtration Methods 0.000 claims abstract description 65
- 238000000034 method Methods 0.000 claims abstract description 36
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 18
- 238000009826 distribution Methods 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000012217 deletion Methods 0.000 claims description 2
- 230000037430 deletion Effects 0.000 claims description 2
- 230000008030 elimination Effects 0.000 claims description 2
- 238000003379 elimination reaction Methods 0.000 claims description 2
- 238000010408 sweeping Methods 0.000 claims 1
- 230000009466 transformation Effects 0.000 claims 1
- 238000005259 measurement Methods 0.000 abstract description 7
- 230000002146 bilateral effect Effects 0.000 description 7
- 230000000694 effects Effects 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000691 measurement method Methods 0.000 description 2
- NAWXUBYGYWOOIX-SFHVURJKSA-N (2s)-2-[[4-[2-(2,4-diaminoquinazolin-6-yl)ethyl]benzoyl]amino]-4-methylidenepentanedioic acid Chemical compound C1=CC2=NC(N)=NC(N)=C2C=C1CCC1=CC=C(C(=O)N[C@@H](CC(=C)C(O)=O)C(O)=O)C=C1 NAWXUBYGYWOOIX-SFHVURJKSA-N 0.000 description 1
- 239000006002 Pepper Substances 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000000919 ceramic Substances 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000004883 computer application Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 230000002401 inhibitory effect Effects 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- 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
- G06T2207/10028—Range image; Depth image; 3D point clouds
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
- G06T2207/20032—Median filtering
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Length Measuring Devices By Optical Means (AREA)
- Image Processing (AREA)
Abstract
The present invention is based on the quick compound filter processing methods of line laser structured light three-dimensional point cloud to belong to vision measurement field, is related to a kind of based on the quick compound filter processing method of line laser structured light formula three-dimensional point cloud.The process employs voxel filtering algorithm, median filtering and moving window LSR filter algorithms to carry out compound filter, the Null Spot cloud and isolated point cloud outside main body point cloud are removed by voxel filtering algorithm first, then the high-frequency noise in main body point cloud and edge is mingled in using median filtering removal, by-line fitting filtering is finally carried out to scan-type point cloud using moving window least squares filtering, the small scale noise being mingled in main body point cloud is filtered out, effective three-dimensional point cloud after finally obtaining removal noise.This method effectively improves the shortcomings that single filtering mode can not remove a variety of noises simultaneously, effectively extends the scope of application of filtering, is a kind of filtering method with wide application prospect.
Description
Technical field
The invention belongs to vision measurement fields, are related to a kind of based on the quick compound filter processing of line laser structured light three-dimensional point cloud
Method.
Background technique
With the continuous development of science and technology, components measurement gradually becomes important link indispensable in production, packet
Various types of measurement problems such as profile, gap, pattern and defect are included all to propose the accuracy of measurement method and rapidity
Very high requirement out.Existing measurement method middle line laser measurement mode with it quickly, high-precision and the measurement such as non-contact
Advantage is widely applied in the industrial production.Line laser scanner is the key that the accurate measured object three-dimensional surface shape that obtains is surveyed
Measuring appratus, with its precision is high, adaptability is good the advantages that, are widely used in fields of measurement.Line laser scanner scans measured object table
Face obtains the three-dimensional point cloud on measured object surface, due to measure field there are it is reflective, complex situations, the measured object surface three such as block
Dimension point cloud is easy to appear the noise and Null Spot cloud of large and small scale.Null Spot cloud is generally off-site from measured object main body point cloud, mostly
The point cloud of measured object support construction.It is easy to be easy to deposit inside it there are large scale high-frequency noise point at the edge of main body point cloud
In small scale low-frequency noise.For the filtering and noise reduction method of cloud, it is divided into two major classes according to the type of cloud: orderly point cloud filtering
It is filtered with dispersion point cloud.The scan-type point cloud genera is mainly wrapped in orderly point cloud, the filter processing method for orderly putting noise in cloud
Include gaussian filtering, median filtering and mean filter.These three point cloud filtering methods are developed from image processing method,
For different types of noise, three kinds of filtering methods have different effects.Wherein, mean filter is using all the points in template
Height value average value replace template center's point height value, all types of high-frequency noises are all had with certain decrease
Effect, effect are general;Median filtering is height value of the intermediate value of the height value of all the points in calculation template as template center's point,
For salt-pepper noise have good removal effect and can preferable retention point cloud edge feature, in high-frequency noise removal and
It is got well than mean filter in terms of retaining edge feature.Gaussian filtering is endowed to the height value of different location point in certain vertex neighborhood
Different weights has the noise of Normal Distribution good then using weighted average as the height value of the point
Inhibitory effect.It is comprehensive that three kinds of orderly classical cloud filtering methods, single filtering mode still can not effectively remove far above
Null Spot cloud, large and small scale noise from main body point cloud.
The Cao Shuan of Hohai University has been delivered equal to the S2 phase in " Southeast China University's journal " natural science edition in 2013 and " has been based on
The bilateral filtering denoising method of a feature selecting " text removes the first and second noise like points using minimum bounding box method, and leads to
It crosses neighborhood point and grid point calculates separately the different bilateral filtering factors, avoid the problem of bilateral filtering algorithm crosses fairing, but
The process of feature selecting is relatively complicated.Yuan Hua of Guilin Electronic Science and Technology Univ. et al. is in 2015 in " computer application " the 8th phase
" the bilateral filtering point cloud Denoising Algorithm based on noise classification " text has been delivered, has been combined using statistical filtering and radius filtering
Mode removes the first, second noise like point, and is improved the bilateral filtering factor, it is contemplated that normal vector angle is to weight
It influences, but has lost the depth information of a cloud.Wu Lushen of University Of Nanchang et al. is in 2016 in " optical precision engineering " the 6th phase
" the three-dimensional point data de-noising based on characteristic information classification " text has been delivered, a kind of three-dimensional based on characteristic information classification is proposed
Point cloud data denoising method is classified to characteristic information using normal vector and Curvature Estimation method, and is distinguished different characteristic areas
It carries out mean filter and bilateral filtering, this method effectively prevents the bilateral filtering flat site less to characteristic information and generated
Fairing phenomenon, but large scale noise can not be removed.Therefore, it is necessary to invent a kind of compound filter method, can be suitable for
A variety of noises remove and preferable can must retain the original characteristic information of measurand.
Summary of the invention
The present invention in order to overcome the shortcomings of the prior art, has invented a kind of based on the quick composite filter of line laser structured light three-dimensional point cloud
Wave processing method, small ruler in the outer isolated noise point of main body point cloud and main body point cloud can not be removed simultaneously by improving single filtering method
The defect for spending noise spot, realizes large scale noise by a cloud compound filter algorithm and effectively goes one by one to small scale noise
It removes, this method effectively increases the adaptability, robustness and validity of filtering algorithm, is a kind of with widely applied filtering calculation
Method.
The technical solution adopted by the present invention is that a kind of be based on the quick compound filter processing method of line laser structured light three-dimensional point cloud,
It is characterized in that the process employs the progress of voxel filtering algorithm, median filtering and moving window LSR filter algorithm is compound
Filtering is removed Null Spot cloud and isolated point cloud outside main body point cloud by voxel filtering algorithm first, is then filtered using intermediate value
Wave removal is mingled in main body point cloud and the high-frequency noise at edge, finally using moving window least squares filtering to scan-type
Point cloud carries out by-line fitting filtering, filters out the small scale noise being mingled in main body point cloud, having after finally obtaining removal noise
Three-dimensional point cloud is imitated, specific step is as follows for method:
Step 1: building point cloud acquisition system, scan-type point cloud is obtained
Line laser scanner is installed to PI electric control platform 1 by fixture 2, while by the I/O output end of PI electric control platform 1
Mouth is connect with the external trigger end of line laser scanner 3, and standard ball 5 is placed on standard ball seat 4, and is put together with support platform 6
It sets in the scan depth of line laser scanner 3;When line laser scanner 3 is in the limit on the left position PI, with line laser structured light
The movement Y-axis of the X of instrument 3, Z axis and PI electric control platform 1 establish global frame of reference O-XYZ, and setting PI electric control platform 1 is suitable
Movement velocity and triggering frequency carry out fixed-point spacing from scanning, by 7 collection point cloud of computer, by sitting to measured object standard ball 5
Subscript conversion obtains the noise-containing initial three-dimensional point cloud matrix P of measured object;
Step 2: obtaining three-dimensional available point cloud using voxel filter removal Null Spot cloud
Interception removal Null Spot cloud is carried out to three-dimensional point cloud using voxel filter, detailed process is tested by the first step
The original point of object converges the D coordinates value of P and single-point:
P={ g (xi,yi,zi),xi,yi,zi∈R,i≤n} (1)
g(xi,yi,zi)=f (xi,yi,zi)+ev(xi,yi,zi)+es(xi,yi,zi) (2)
Wherein, P is that initial three-dimensional point converges, g (xi,yi,zi) it is i-th coordinate value during original point is converged, n is that point Yun is total
Quantity, f (xi,yi,zi) it is muting cloud, ev(xi,yi,zi) it is the noise that point converges, es(xi,yi,zi) it is during point converges
Null Spot cloud, wherein ev(xi,yi,zi) cannot completely remove and can only reduce, and es(xi,yi,zi) filtered by voxel filter
It removes;
According to measured object three-dimensional body point cloud distribution situation and noise profile situation, chooses suitable point cloud and intercept boundary
xmin,xmax,ymin,ymax,zmin,zmax, the point cloud deleted outside boundary obtains new point and converges Q and its boundary length:
Q={ g (xi,yi,zi),xmin≤xi≤xmax,ymin≤yi≤ymax,zmin≤zi≤zmax,i≤n} (3)
K in formula (4)X,KY,KZFor the boundary length for newly putting the X, Y for converging Q, Z-direction;
Step 3: removing high-frequency noise using median filtering
The elimination for carrying out high-frequency noise to cloud using median filtering, first presses scanning direction, i.e. Y-direction converges progress to point
Grouping
The scanning element cloud that line laser scanner 3 obtains is divided into 0.02m, the point cloud interval T of Y-direction between the point cloud of X-direction
It is specifically set by the surface complex situations of sweep object;Assuming that every c × d is a scanning window, according to the point cloud of c × d size
Battle array is one group, is divided intoGroup;
Then, the Z value of this group of center point is replaced with the Z value median of each group point
The median of each group point Z value is calculated, and the Z value is replaced to the new Z coordinate value of this group of center point, with such
It pushes away, until having traversed all groups of point cloud, achievees the purpose that remove each group high-frequency noise using this method, be gone by this noise
It removes, obtains new point and converge S;
Step 4: using LSR filter algorithm smothing filtering
It is carried out using LSR filter algorithm smothing filtering since scanning wire type point cloud has apparent topological relation
By-line matched curve achievees the purpose that the small scale noise of removal, and detailed process is as follows:
1) measurand three-dimensional surface feature difference causes the number of matched curve different, needs according to specific surface characteristics
Choose suitable polynomial fitting number k;
Wherein;
Wherein: a0,a1,a2,a3,.....,ak-1,akFor k+1 coefficient of polynomial fitting, f (x) is the z value of corresponding points
Match value, x are the abscissa of corresponding points;
2) deviation the sum of of the calculating each point to curve distance
3) a is asked to formula (6) the rightiPartial derivative
Above formula is converted are as follows:
A(x)a(x)-pT(x) (9) Z=0
Wherein,
4) coefficient vector and fitting function are sought
A (x)=A-1(x)B(x)Z (10)
Fitting function is as follows:
F (x)=pT(x)A-1(x)p(x)Z (11)
5) removal of small scale noise
If z0It is a threshold value in the direction cloud z, if | f (x)-z (x) | > z0, then by the point deletion;
By second step to the 4th step, gradually eliminate Null Spot cloud far from main body point cloud, inside main body point cloud and
Small scale noise inside the high-frequency noise at edge, main body point cloud.
The characteristics of scan-type is presented for line laser scanner collection point cloud the beneficial effects of the invention are as follows this method, invention
A kind of quick compound filter method uses voxel filtering, median filtering and minimum for removing Null Spot cloud and noise
Square law compound filter can gradually remove Null Spot cloud, main body point cloud edge and the high frequency of inside far from main body point cloud
Small scale noise inside noise and main body point cloud improves single filtering mode and only rises to a certain or a certain noise like
The shortcomings that effect, effectively extends the scope of application of filtering method, is a kind of with the filtering side being widely applied with high robust
Method.
Detailed description of the invention
Fig. 1 is quick compound filter method flow diagram
Fig. 2 is line laser scanner acquisition three-dimensional point cloud system diagram, wherein the automatically controlled mobile platform of 1-PI, the right angle 2- fixture,
3- line laser scanner, 4- standard ball seat, 5- standard ball, 6- support platform, 7- computer.
Specific embodiment
With reference to the accompanying drawing with the technical solution specific implementation that the present invention will be described in detail.
The line laser scanner 3 that the present embodiment is selected is the scanner of the LJ-7060 model of Keyemce company, the mark of selection
Quasi- ball 5 is the matt ceramic ball of normal diameter 15mm, and PI electric control platform 1 is the electric control platform of PI Corp. M521.DD1 model.
Step 1: building point cloud acquisition system, scan-type three-dimensional point cloud is obtained
Line laser scanner is installed to PI electric control platform 1 by fixture 2, while by the I/O output end of PI electric control platform 1
Mouth is connect with the external trigger end of line laser scanner 3, and standard ball 5 is placed on standard ball seat 4, and is put together with support platform 6
It sets in the scan depth of line laser scanner 3;When line laser scanner 3 is in the limit on the left position PI, with line laser structured light
The movement Y-axis of the X of instrument 3, Z axis and PI electric control platform 1 establish global frame of reference O-XYZ, and setting PI electric control platform 1 is suitable
Movement velocity and triggering frequency carry out fixed-point spacing from scanning, by 7 collection point cloud of computer, by sitting to measured object standard ball 5
Subscript conversion obtains the noise-containing initial three-dimensional point cloud matrix P of measured object.
Step 2: obtaining three-dimensional available point cloud using voxel filter removal Null Spot cloud
The point cloud obtained by the first step, there are high and low frequency noise, gross error point and (non-tested pairs of Null Spot cloud
Image point cloud), three-dimensional point cloud is intercepted using voxel filter, converges note with the original point that formula (1) obtains measurand
For P, the D coordinates value of single-point is sought with formula (2).
According to measured object three-dimensional body point cloud distribution situation and noise profile situation, chooses suitable point cloud and intercept boundary
xmin,xmax,ymin,ymax,zmin,zmax, the point cloud outside boundary is deleted, new point is obtained using formula (3) and formula (4) and converges Q.
Step 3: removing high-frequency noise using median filtering
High-frequency noise in point cloud is removed using median filtering, point is converged by scanning direction (Y-direction) and is grouped,
The scanning element cloud that LJ-7060 line laser scanner 3 obtains is divided into 0.02m between the point cloud of X-direction, according to the surface of measurand
Three-dimensional appearance sets and is divided into T=0.05 between the point cloud of Y-direction;Window size is set as 3 × 3, according to the point cloud battle array of 3 × 3 sizes
It is one group, is divided intoGroup.
Replace the Z value of this group of center point with the Z value median of each group point, calculates each group point Z value (depth information coordinate
Value) median, and the Z value is replaced to the new Z coordinate value of this group of center point, and so on, until having traversed all groups
Point cloud, achieve the purpose that removing each group high-frequency noise by this noise remove obtains new point and converge S using this method.
Step 4: using LSR filter algorithm smothing filtering
The low-frequency noise in cloud is removed using LSR filter algorithm, since scanning wire type point cloud is with bright
Aobvious topological relation, therefore by-line fitting scanning can be carried out, achieve the purpose that remove the small scale noise of low frequency.Utilize formula
(5)-(6) suitable polynomial fitting number k is chosen, polynomial fitting is constructed.Then each point is calculated to you and song by formula (7)
The sum of deviation of linear distance.
A is asked to formula (7) the rightiPartial derivative obtain formula (8), (9), from formula (10) seek polynomial coefficient to
Amount, obtains fitting function by formula (11);By | f (x)-z (x) | > z0Remove the small scale noise that difference is greater than given threshold.Through
Second step is crossed to the 4th step, gradually eliminates inside Null Spot cloud and noise, main body point cloud far from main body point cloud and edge
Small scale noise inside high-frequency noise, main body point cloud.
This method is on the basis of clearly point cloud topological relation, the filtering of integrated application voxel, median filtering and least square
Filtering, intercepts a cloud using voxel filter, Null Spot cloud and noise spot far from main body point cloud is eliminated, in
Value filtering eliminates inside main body point cloud and the high frequency large scale noise at edge, eliminates main body point cloud using least squares filtering
Internal small scale noise, improves the shortcomings that single filtering mode can not remove large scale noise and small scale noise simultaneously,
Effectively extend the robustness and adaptability of filtering and noise reduction method.
Claims (1)
1. one kind is based on the quick compound filter processing method of line laser structured light three-dimensional point cloud, characterized in that the process employs bodies
Plain filtering algorithm, median filtering and moving window LSR filter algorithm carry out compound filter;It is filtered and is calculated by voxel first
Method removes the Null Spot cloud and isolated point cloud outside main body point cloud, then use median filtering removal be mingled in main body point cloud it is interior with
And the high-frequency noise at edge, by-line fitting filtering, filter are finally carried out to scan-type point cloud using moving window least squares filtering
Effective three-dimensional point cloud except the small scale noise being mingled in main body point cloud, after finally obtaining removal noise;The specific step of method
It is rapid as follows:
Step 1: building point cloud acquisition system, scan-type point cloud is obtained
Line laser scanner (3) are installed to PI electric control platform (1) by right angle fixture (2), by the I/O of PI electric control platform (1)
Output port is connect with the external trigger end of line laser scanner (3), and standard ball (5) is placed on standard ball seat (4), and with support
Platform (6) is placed on together in the scan depth of line laser scanner (3);When line laser scanner (3) is in PI electric control platform
(1) when limit on the left position, global base is established with the movement Y-axis of the X of line laser scanner (3), Z axis and PI electric control platform (1)
Conventional coordinates O-XYZ, the suitable movement velocity of setting PI electric control platform (1) and triggering frequency, carry out measured object standard ball (5)
Fixed-point spacing is from scanning, by computer (7) collection point cloud, obtains the noise-containing initial three-dimensional point of measured object by coordinate transformation
Cloud matrix P;
Step 2: obtaining three-dimensional available point cloud using voxel filter removal Null Spot cloud
Three-dimensional point cloud is intercepted using voxel filter, removes Null Spot cloud;The original of measurand is obtained by the first step
Point converges the D coordinates value of P and single-point:
P={ g (xi,yi,zi),xi,yi,zi∈R,i≤n} (1)
g(xi,yi,zi)=f (xi,yi,zi)+ev(xi,yi,zi)+es(xi,yi,zi) (2)
Wherein, P is that initial three-dimensional point converges, g (xi,yi,zi) it is i-th coordinate value during original point is converged, n is point cloud total quantity,
f(xi,yi,zi) it is muting cloud, ev(xi,yi,zi) it is the noise that point converges, es(xi,yi,zi) it is invalid during point converges
Cloud is put, wherein ev(xi,yi,zi) cannot completely remove and can only reduce, and es(xi,yi,zi) filtered out by voxel filter;
According to measured object three-dimensional body point cloud distribution situation and noise profile situation, chooses suitable point cloud and intercept boundary xmin,
xmax,ymin,ymax,zmin,zmax, the point cloud deleted outside boundary obtains new point and converges Q and its boundary length:
Q={ g (xi,yi,zi),xmin≤xi≤xmax,ymin≤yi≤ymax,zmin≤zi≤zmax,i≤n} (3)
K in formula (4)X,KY,KZFor the boundary length for newly putting the X, Y for converging Q, Z-direction;
Step 3: removing high-frequency noise using median filtering
The elimination for carrying out high-frequency noise to cloud using median filtering, first presses scanning direction, i.e., Y-direction converges minute point
Group;The scanning element cloud that line laser scanner (3) obtains is divided into 0.02m between the point cloud of X-direction, and the point cloud interval T of Y-direction is by sweeping
The surface complex situations for retouching object are specifically set;Assuming that every c × d is a scanning window, it is according to the point cloud battle array of c × d size
It one group, is divided into are as follows:Group;
Then, the Z value of this group of center point is replaced with the Z value median of each group point
The median of each group point Z value is calculated, and the Z value is replaced to the new Z coordinate value of this group of center point, and so on, directly
To the point cloud for having traversed all groups, achieve the purpose that remove each group high-frequency noise obtains by this noise remove using this method
S is converged to new point;
Step 4: using LSR filter algorithm smothing filtering
Using LSR filter algorithm smothing filtering, the low-frequency noise in cloud is removed, due to scanning wire type point cloud
With apparent topological relation, by-line matched curve is carried out, achievees the purpose that the small scale noise of removal, detailed process is as follows:
1) measurand three-dimensional surface feature difference causes the number of matched curve different, needs to be chosen according to specific surface characteristics
Suitable polynomial fitting number k;
Wherein;
P (x)=[1, x, x2,x3,x4,x5..., xk-1]T
A (x)=[a0,a1,a2,a3,a4,a5,…,ak-1]T (6)
Wherein: a0,a1,a2,a3,.....,ak-1,akFor k+1 coefficient of polynomial fitting, f (x) is that the z value of corresponding points is fitted
Value, x are the abscissa of corresponding points;
2) deviation the sum of of the calculating each point to curve distance
3) a is asked to formula (6) the rightiPartial derivative
Above formula is converted are as follows:
A(x)a(x)-pT(x) (9) Z=0
Wherein,
4) coefficient vector and fitting function are sought
A (x)=A-1(x)B(x)Z (10)
Fitting function is as follows:
F (x)=pT(x)A-1(x)p(x)Z (11)
5) removal of small scale noise
If z0It is a threshold value in the direction cloud z, if | f (x)-z (x) | > z0, then by the point deletion;
By second step to the 4th step, Null Spot cloud, main body point cloud inside and edge far from main body point cloud are gradually eliminated
High-frequency noise, the small scale noise inside main body point cloud.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810787767.1A CN108986048B (en) | 2018-07-18 | 2018-07-18 | Three-dimensional point cloud rapid composite filtering processing method based on line laser scanning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810787767.1A CN108986048B (en) | 2018-07-18 | 2018-07-18 | Three-dimensional point cloud rapid composite filtering processing method based on line laser scanning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108986048A true CN108986048A (en) | 2018-12-11 |
CN108986048B CN108986048B (en) | 2020-04-28 |
Family
ID=64549609
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810787767.1A Active CN108986048B (en) | 2018-07-18 | 2018-07-18 | Three-dimensional point cloud rapid composite filtering processing method based on line laser scanning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108986048B (en) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110349094A (en) * | 2019-06-12 | 2019-10-18 | 西安工程大学 | It is peeled off based on statistics and the 3D point cloud denoising method of adaptive bilateral mixed filtering |
CN110458772A (en) * | 2019-07-30 | 2019-11-15 | 五邑大学 | A kind of point cloud filtering method, device and storage medium based on image procossing |
CN110634185A (en) * | 2019-07-31 | 2019-12-31 | 众宏(上海)自动化股份有限公司 | Visual algorithm for quickly forming point cloud for gear repair |
CN111524084A (en) * | 2020-05-13 | 2020-08-11 | 曹彬才 | Complex scene photon counting laser radar point cloud denoising algorithm based on multimodal Gaussian fitting |
CN111912346A (en) * | 2020-06-30 | 2020-11-10 | 成都飞机工业(集团)有限责任公司 | Nest hole online detection method suitable for robot drilling and riveting system on surface of airplane |
CN112558033A (en) * | 2020-12-30 | 2021-03-26 | 成都圭目机器人有限公司 | Radar data standard processing method based on three-dimensional ground penetrating radar |
CN113218330A (en) * | 2021-04-27 | 2021-08-06 | 大连理工大学 | PCL (polycaprolactone) -library-based large-scale thin-wall component measuring and data processing method |
CN113298951A (en) * | 2021-06-07 | 2021-08-24 | 山东省科学院激光研究所 | Three-dimensional frequency surface digitization and model comparison method |
CN113763537A (en) * | 2021-09-07 | 2021-12-07 | 岱悟智能科技(上海)有限公司 | Three-dimensional point cloud plane optimization method based on plane recognition |
CN113837952A (en) * | 2020-06-24 | 2021-12-24 | 影石创新科技股份有限公司 | Three-dimensional point cloud noise reduction method and device based on normal vector, computer readable storage medium and electronic equipment |
CN113960046A (en) * | 2021-10-08 | 2022-01-21 | 深圳市霄光科技有限公司 | Surface detection method and system |
CN116165677A (en) * | 2023-04-24 | 2023-05-26 | 湖北中图勘测规划设计有限公司 | Geological investigation method and device based on laser radar |
CN116579955A (en) * | 2023-07-13 | 2023-08-11 | 厦门微图软件科技有限公司 | New energy battery cell weld reflection point denoising and point cloud complement method and system |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103886555A (en) * | 2014-03-12 | 2014-06-25 | 北京昊峰东方科技有限公司 | Processing method based on mass three-dimensional laser scanning point cloud data |
WO2017120897A1 (en) * | 2016-01-15 | 2017-07-20 | 武汉武大卓越科技有限责任公司 | Object surface deformation feature extraction method based on line scanning three-dimensional point cloud |
CN107133565A (en) * | 2017-03-31 | 2017-09-05 | 大连理工大学 | Laser incising molded line feature extracting method based on line laser |
US20170368755A1 (en) * | 2016-06-22 | 2017-12-28 | Massachusetts Institute Of Technology | Methods and Apparatus for 3D Printing of Point Cloud Data |
-
2018
- 2018-07-18 CN CN201810787767.1A patent/CN108986048B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103886555A (en) * | 2014-03-12 | 2014-06-25 | 北京昊峰东方科技有限公司 | Processing method based on mass three-dimensional laser scanning point cloud data |
WO2017120897A1 (en) * | 2016-01-15 | 2017-07-20 | 武汉武大卓越科技有限责任公司 | Object surface deformation feature extraction method based on line scanning three-dimensional point cloud |
US20170368755A1 (en) * | 2016-06-22 | 2017-12-28 | Massachusetts Institute Of Technology | Methods and Apparatus for 3D Printing of Point Cloud Data |
CN107133565A (en) * | 2017-03-31 | 2017-09-05 | 大连理工大学 | Laser incising molded line feature extracting method based on line laser |
Non-Patent Citations (3)
Title |
---|
SERGUO ORTS-ESCOLANO,ET AL.: "《Point Cloud Data Filtering and Downsampling using Growing Neural Gas》", 《PROCEEDINGS OF INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS》 * |
刘巍等: "《面向大型零部件的三维形面高精度测量方法》", 《红外与激光工程》 * |
申超: "《基于能量优化的三维点云中多几何基元识别研究》", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110349094A (en) * | 2019-06-12 | 2019-10-18 | 西安工程大学 | It is peeled off based on statistics and the 3D point cloud denoising method of adaptive bilateral mixed filtering |
CN110458772A (en) * | 2019-07-30 | 2019-11-15 | 五邑大学 | A kind of point cloud filtering method, device and storage medium based on image procossing |
CN110458772B (en) * | 2019-07-30 | 2022-11-15 | 五邑大学 | Point cloud filtering method and device based on image processing and storage medium |
CN110634185A (en) * | 2019-07-31 | 2019-12-31 | 众宏(上海)自动化股份有限公司 | Visual algorithm for quickly forming point cloud for gear repair |
CN111524084A (en) * | 2020-05-13 | 2020-08-11 | 曹彬才 | Complex scene photon counting laser radar point cloud denoising algorithm based on multimodal Gaussian fitting |
CN111524084B (en) * | 2020-05-13 | 2023-08-22 | 曹彬才 | Photon counting laser radar point cloud denoising method based on multi-peak Gaussian fitting |
CN113837952A (en) * | 2020-06-24 | 2021-12-24 | 影石创新科技股份有限公司 | Three-dimensional point cloud noise reduction method and device based on normal vector, computer readable storage medium and electronic equipment |
CN111912346A (en) * | 2020-06-30 | 2020-11-10 | 成都飞机工业(集团)有限责任公司 | Nest hole online detection method suitable for robot drilling and riveting system on surface of airplane |
CN111912346B (en) * | 2020-06-30 | 2021-12-10 | 成都飞机工业(集团)有限责任公司 | Nest hole online detection method suitable for robot drilling and riveting system on surface of airplane |
CN112558033A (en) * | 2020-12-30 | 2021-03-26 | 成都圭目机器人有限公司 | Radar data standard processing method based on three-dimensional ground penetrating radar |
CN113218330B (en) * | 2021-04-27 | 2022-04-08 | 大连理工大学 | PCL (polycaprolactone) -library-based large-scale thin-wall component measuring and data processing method |
CN113218330A (en) * | 2021-04-27 | 2021-08-06 | 大连理工大学 | PCL (polycaprolactone) -library-based large-scale thin-wall component measuring and data processing method |
CN113298951A (en) * | 2021-06-07 | 2021-08-24 | 山东省科学院激光研究所 | Three-dimensional frequency surface digitization and model comparison method |
CN113763537A (en) * | 2021-09-07 | 2021-12-07 | 岱悟智能科技(上海)有限公司 | Three-dimensional point cloud plane optimization method based on plane recognition |
CN113960046A (en) * | 2021-10-08 | 2022-01-21 | 深圳市霄光科技有限公司 | Surface detection method and system |
CN116165677A (en) * | 2023-04-24 | 2023-05-26 | 湖北中图勘测规划设计有限公司 | Geological investigation method and device based on laser radar |
CN116579955A (en) * | 2023-07-13 | 2023-08-11 | 厦门微图软件科技有限公司 | New energy battery cell weld reflection point denoising and point cloud complement method and system |
CN116579955B (en) * | 2023-07-13 | 2023-10-20 | 厦门微图软件科技有限公司 | New energy battery cell weld reflection point denoising and point cloud complement method and system |
Also Published As
Publication number | Publication date |
---|---|
CN108986048B (en) | 2020-04-28 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108986048A (en) | Based on the quick compound filter processing method of line laser structured light three-dimensional point cloud | |
CN109887015A (en) | A kind of point cloud autoegistration method based on local surface feature histogram | |
CN104981105B (en) | A kind of quickly accurate detection and method for correcting error for obtaining element central and deflection angle | |
CN105627923B (en) | A kind of planning parameters of scanning paths method that blade dense point cloud based on laser ranging is obtained | |
CN102032875B (en) | Image-processing-based cable sheath thickness measuring method | |
CN111553292B (en) | Rock mass structural plane identification and occurrence classification method based on point cloud data | |
CN104392476A (en) | Method of extracting three-dimensional axis of tunnel based on minimum bounding box algorithm | |
CN112446123B (en) | Measuring head pose planning method for blisk three-coordinate measuring machine | |
CN112033338B (en) | Blade curved surface contact type scanning measurement probe radius surface compensation method | |
CN107122333A (en) | It is a kind of to reduce the data processing method of measurement data uncertainty | |
CN110006372A (en) | A kind of three-dimensional point cloud flatness calculation method based on local optimum | |
CN112461175A (en) | Method for measuring wide chord and large torsion angle blade profile of fan blisk | |
Ren et al. | Invariant-feature-pattern-based form characterization for the measurement of ultraprecision freeform surfaces | |
CN109683552A (en) | A kind of Machining Path generation method on the complicated point cloud model of basal plane curve orientation | |
CN110379011A (en) | A kind of underwater topography dot cloud hole method for repairing and mending based on improvement B-spline Curve | |
CN104089599B (en) | Quasi morphological filtering method for extracting two-dimensional contour in contact measuring head measurement | |
CN109141266A (en) | A kind of steel construction measurement method and system | |
CN114608461B (en) | Laser scanning measurement method for non-uniform wall thickness parts | |
CN110306414A (en) | A kind of pavement structural depth detection method | |
CN110781573A (en) | Reverse design method for rotor blade of aircraft engine | |
CN102012964A (en) | Method and device for processing sampling data of laser scanning | |
CN107091603B (en) | A method of measurement integral blade disk | |
CN110648391B (en) | Point cloud processing three-dimensional reconstruction method | |
CN112365602A (en) | Pavement three-dimensional reconstruction error calculation method based on laser point cloud | |
CN106803266A (en) | A kind of hull complexity outside plate point cloud costal line extracting method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |