CN111754421B - Improved guiding filtering three-dimensional scattered point cloud rapid fairing method - Google Patents
Improved guiding filtering three-dimensional scattered point cloud rapid fairing method Download PDFInfo
- Publication number
- CN111754421B CN111754421B CN202010483771.6A CN202010483771A CN111754421B CN 111754421 B CN111754421 B CN 111754421B CN 202010483771 A CN202010483771 A CN 202010483771A CN 111754421 B CN111754421 B CN 111754421B
- Authority
- CN
- China
- Prior art keywords
- point cloud
- point
- cloud data
- formula
- data
- 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.)
- Active
Links
- 238000001914 filtration Methods 0.000 title claims abstract description 33
- 238000000034 method Methods 0.000 title claims abstract description 30
- 230000000903 blocking effect Effects 0.000 claims abstract description 5
- 238000000638 solvent extraction Methods 0.000 claims description 8
- 230000010354 integration Effects 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000002146 bilateral effect Effects 0.000 abstract description 4
- 230000001174 ascending effect Effects 0.000 abstract description 2
- 238000009499 grossing Methods 0.000 abstract 2
- 238000010586 diagram Methods 0.000 description 4
- 238000011084 recovery Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 241000283973 Oryctolagus cuniculus Species 0.000 description 1
- 230000002411 adverse Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005336 cracking Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000011555 rabbit model 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
- 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
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Complex Calculations (AREA)
- Processing Or Creating Images (AREA)
Abstract
The invention discloses an improved guide filtering three-dimensional scattered point cloud rapid smoothing method, which comprises the steps of arranging point cloud data to be processed according to an ascending order of axes, solving the maximum point cloud quantity in the data to be segmented according to the total quantity of the point clouds and the quantity of to-be-segmented, calculating the expansion processing points according to the maximum point cloud quantity, blocking the point cloud data according to the maximum point cloud quantity and the expansion processing points in the data to be segmented, generating an overlapping area of the segmented point clouds, effectively solving the splitting phenomenon generated by the point cloud data, processing the segmented point cloud data through a parallel three-dimensional guide filtering point cloud smoothing algorithm to improve the operation rate, and finally recovering and integrating the processed point cloud data to output a result. The method can effectively solve the problem of gradient inversion in the bilateral filtering fairing method and can effectively improve the operation speed of the algorithm.
Description
Technical Field
The invention relates to the field of computer vision technology and three-dimensional reconstruction reverse engineering, and particularly provides an improved guiding filtering three-dimensional scattered point cloud rapid fairing method.
Background
The three-dimensional point cloud is widely focused as data capable of expressing geometric information of the surface of an object, and the acquisition of the point cloud data is an important link of reverse engineering. However, due to defects of the acquisition equipment, system noise, interference of environmental factors and the like, noise exists in the obtained point cloud data, and the point cloud data containing the noise can have adverse effects on subsequent point cloud processing, so that denoising of the three-dimensional point cloud data has become one of hot problems in the field of reverse engineering research.
The noise contained in the three-dimensional point cloud data can be divided into drift noise points and mixed noise points. The drifting noise point is represented by a noise point which is obviously far away from the point cloud main body and floats near the point cloud main body, and the mixed noise point is represented by a noise point mixed with the point cloud main body. At present, methods used by industry practitioners in processing mixed noise points are mainly a bilateral filtering fairing method and a guided filtering fairing method. The bilateral filtering point cloud fairing method can effectively reduce image noise and can keep edge information to the maximum extent, but when a point (usually on an edge) has a plurality of similar points around, gradient inversion problem can exist due to unstable average Gaussian weighting, so that the fairing effect is poor. The guiding filtering fairing method can effectively solve the problem of gradient inversion in the bilateral filtering fairing method, but the algorithm running time is still not ideal. Both existing methods have the problem of slower running speed.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides an improved guiding filtering three-dimensional scattered point cloud rapid fairing method, so that the problem of low running speed in the prior art is solved.
The technical scheme for solving the problems is that an improved guiding filtering three-dimensional scattered point cloud rapid fairing method is provided, and comprises the following steps:
step 1: inputting three-dimensional scattered point cloud data P= { P containing noise i },i＝1,2,...,n，p i Representing each point in the point cloud P, wherein n is the total number of the point clouds;
step 2: partitioning the three-dimensional point cloud data P containing noise obtained in the step 1 to form point cloud data P j ：
(2-1) arranging point cloud data to be processed in ascending order of a Z axis;
(2-2) calculating the maximum point cloud quantity M according to the formula (1), and then calculating the expansion processing point quantity M according to the formula (2),
wherein N represents the total number of point clouds P, N represents the number of blocks to be segmented, a represents the percentage of the number M of expansion processing points to the maximum number M of point clouds, and whereinRepresentation up-takeFinishing;
(2-3) determining a blocking range according to M and M, wherein the dividing range of each block of processing point cloud is shown in a formula (3),
p in the formula j J=1, 2, N, representing the j-th block of the point cloud P;
step 3: for the block point cloud data P of the step 2 j Processing by parallel three-dimensional guided filtering point cloud fairing algorithm to obtain point cloud data Q after fairing j ；
Step 4: for the point cloud data Q processed in the step 3 j Recovery integration is carried out, and a result Q is output
According to the point cloud data Q processed by M in the step 3 j Integration into output point cloud q= { Q j J=1, 2,..n, specifically as in formula (9),
according to the parameter M in the step 2, the point cloud data Q processed in the step 3 is processed j Integration into output point cloud q= { Q j J=1, 2, N, specifically as shown in formula (9),
and (3) recovering and integrating the partitioned point clouds according to a formula (9), and finally outputting a result graph processed by the algorithm.
The step 3 comprises the following steps:
(3-1) inputting the three-dimensional point cloud data P to be processed after the partitioning j Setting the number of search neighborhood K=10 and the coefficient epsilon=0.1 of the identity matrix, and P j Is transmitted into different processing units to realize the parallel operation of the algorithm;
(3-2) searching for p Using K-D Tree Algorithm i K neighborhood of points, wherein K is the number of neighborhood points and is smaller than the total number of each block of point cloud, and a neighborhood point set is set as N (p) according to a formula (3) i )＝{p ik ∈p j P, where ik Is p i K=1, 2,..；
(3-3) from the neighborhood set, we can calculate a quantityCalled point p i The centroid of the neighborhood is shown in formula (4);
(3-4) calculating a of the point according to the formula (5) and the formula (6) i And b i ，
Thereby obtaining an optimal solution of the formula (7);
(3-5) equation (7) minimizes the reconstruction residual of equation (8), i.e., the spatial position of the point, q, can be updated according to equation (8) i ＝a i p i +b i (8)
Wherein p is i And q i Is the point before and after the guiding filtration, a i Is a scalar, b i Is a vector of 3*1;
(3-6) traversing and updating all data points in each block; judging whether the data point is the last data point, if so, finishing a filtering fairing algorithm, otherwise, returning to the step (3-2), and iterating;
(3-7) outputting the three-dimensional point cloud data Q after fairing j 。
Compared with the prior art, the method has the following beneficial effects:
compared with the traditional three-dimensional guide filtering point cloud fairing algorithm, the improved guide filtering three-dimensional scattered point cloud rapid fairing method provided by the invention improves fairing capacity and algorithm speed, blocks point cloud data to realize parallel processing, and greatly shortens operation time of the algorithm.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a three-dimensional point cloud data schematic of a rabbit model;
FIG. 3 is a schematic diagram of three-dimensional point cloud data containing noise;
FIG. 4 is a schematic diagram of three-dimensional scattered point cloud data chunking;
FIG. 5 is a flow chart of a three-dimensional point cloud oriented filtering fairing algorithm;
FIG. 6 is a graph of the results of a three-dimensional point cloud guided filter fairing algorithm;
fig. 7 is a graph of the results of a modified steering filtered three-dimensional scattered point cloud rapid fairing method.
Detailed Description
The invention is described in further detail below with reference to the attached drawings and specific embodiments:
referring to fig. 1, the invention provides an improved guiding filtering three-dimensional scattered point cloud rapid fairing method, which comprises the following steps:
step 1: taking a rabbit point cloud model of a standard point cloud library of the university of stanfu as an example, see fig. 2, a mixed noise is added thereto, denoted as P, and the number of point clouds n=40256, see fig. 3.
Step 2: the three-dimensional point cloud data P containing noise, which is obtained in the step 1, is divided into blocks with N=7
The specific operation steps comprise:
(2-1) first, the point cloud data to be processed needs to be sorted from small to large according to the Z axis.
And (2-2) calculating the maximum point cloud quantity M=5751 according to the formula (1), and calculating the expansion processing point quantity m=575 according to the formula (2), wherein a=0.1 is taken.
Wherein N represents the total number of point clouds P, N represents the number of blocks to be segmented, a represents the percentage of the number M of expansion processing points to the maximum number M of point clouds, and whereinRepresenting an upward rounding;
(2-3) the blocking range can then be determined from M and M, and the partitioning range is determined from equation (3) as shown in Table 1.
TABLE 1 specific Point cloud data partitioning Range
| The j th block | Processing point cloud ranges | Specific range |
| 1 | [1,1×M+m] | [1,6326] |
| 2 | [1×M-m,2×M+m] | [5176，12077] |
| 3 | [2×M-m,3×M+m] | [10927,17828] |
| 4 | [3×M-m,4×M+m] | [16678,23579] |
| 5 | [4×M-m,5×M+m] | [22433，29360] |
| 6 | [5×M-m,6×M+m] | [28180，35081] |
| 7 | [6×M-m,n] | [33931，40256] |
The point cloud is divided according to the method, and a block diagram is shown in fig. 4. There is an overlap between the blocks, which has the advantage that no cracking occurs when processing different blocks.
Step 3: for the block point cloud data P of the step 2 j The parallel three-dimensional guided filtering point cloud fairing algorithm processing is carried out, see fig. 5, and the specific steps include:
(3-1) inputting the three-dimensional point cloud data P to be processed after the partitioning j Setting the number of search neighborhood K=10 and the coefficient epsilon=0.1 of the identity matrix, and P j Is transmitted into different processing units to realize the parallel operation of the algorithm.
(3-2) searching for p Using K-D Tree Algorithm i K neighborhood of points, wherein K is the number of neighborhood points and is smaller than the total number of each block of point cloud, and a neighborhood point set is set as N (p) according to a formula (3) i )＝{p ik ∈p j P, where ik Is p i K=1, 2,..k.
(3-3) from the neighborhood set, we can calculate a quantityCalled point p i Centroid of neighborhood, formula (4).
(3-4) calculating a of the point according to the formula (5) and the formula (6) i And b i ，
Thereby obtaining the optimal solution of the formula (7).
(3-5) equation (7) minimizes the reconstruction residual of equation (8), i.e., the spatial location of the point can be updated according to equation (8),
q i ＝a i p i +b i (8)
wherein p is i And q i Is the point before and after the guiding filtration, a i Is a scalar, b i Is a vector of 3*1.
(3-6) traversing and updating all data points in each block. And (3) judging whether the data point is the last data point, if so, finishing a filtering fairing algorithm, otherwise, returning to the step (3-2), and iterating.
(3-7) outputting the three-dimensional point cloud data Q after fairing j 。
Step 4: for the point cloud data Q processed in the step 3 j And (5) recovering and integrating, and outputting a result Q.
Processing the processed point cloud data Q according to m=5751 in step 3 j Integration into output point cloud q= { Q j J=1, 2,..n, specifically as in formula (9),
and recovering and integrating the partitioned point clouds according to the method, wherein the specific recovery point cloud point taking range is shown in table 2.
TABLE 2 Point cloud recovery partitioning Range
| The j th block | Processing point cloud ranges | Specific range |
| 1 | [1,1×M] | [1,5751] |
| 2 | [1×M+1,2×M] | [5752，11502] |
| 3 | [2×M+1,3×M] | [11503,17253] |
| 4 | [3×M+1,4×M] | [17254,23004] |
| 5 | [4×M+1,5×M] | [23005，28755] |
| 6 | [5×M+1,6×M] | [28756，34506] |
| 7 | [6×M+1,n] | [34507，40256] |
Finally, a result graph processed by the algorithm is output, and experimental simulation results of the improved three-dimensional scattered point cloud rapid fairing method of the guided filtering are shown in fig. 7. Fig. 6 is a diagram of experimental simulation results of a non-blocking three-dimensional guided filter fairing method.
Under the conditions of a certain total number of point clouds and a certain noise adding degree, the comparison result of the three-dimensional guiding filtering fairing method and the improved guiding filtering three-dimensional scattered point cloud rapid fairing method of the invention is shown in table 3:
table 3 comparison of different algorithm results
From table 3, it can be seen that the improved three-dimensional scattered point cloud rapid fairing method of guided filtering has a significant improvement in running speed, and is consistent with the three-dimensional guided filtering method in terms of filtering precision. Thus demonstrating the feasibility of the invention.
Claims (1)
1. An improved guiding filtering three-dimensional scattered point cloud rapid fairing method is characterized in that: the method comprises the following steps:
step 1: inputting three-dimensional scattered point cloud data P= { P containing noise i },i＝1,2,...,n，p i Representing each point in the point cloud P, n is the total point cloudA number of;
step 2: partitioning the three-dimensional point cloud data P containing noise obtained in the step 1 to form point cloud data P j ：
(2-1) sorting the point cloud data to be processed from small to large according to the Z axis;
(2-2) calculating the maximum point cloud quantity M according to the formula (1), and then calculating the expansion processing point quantity M according to the formula (2),
wherein N represents the total number of point clouds P, N represents the number of blocks to be segmented, a represents the percentage of the number M of expansion processing points to the maximum number M of point clouds, and whereinRepresenting an upward rounding;
(2-3) determining a blocking range according to M and M, wherein the dividing range of each block of processing point cloud is shown as a formula (3),
p in the formula j J=1, 2, N, representing the j-th block of the point cloud P;
step 3: for the block point cloud data P of the step 2 j Processing by parallel three-dimensional guided filtering point cloud fairing algorithm to obtain point cloud data Q after fairing j ；
The specific method comprises the following steps:
(3-1) inputting the three-dimensional point cloud data P to be processed after the partitioning j Setting the number of search neighborhood K=10 and the coefficient epsilon=0.1 of the identity matrix, and P j Is transmitted into different processing units to realize the parallel operation of the algorithm;
(3-2) searching for p Using K-D Tree Algorithm i K neighborhood of points, wherein K is the number of neighborhood points and is smaller than the total number of each block of point cloud, and a neighborhood point set is set as N (p) according to a formula (3) i )＝{p ik ∈p j P, where ik Is p i K=1, 2,.;
(3-3) from the neighborhood set, we can calculate a quantityCalled point p i The centroid of the neighborhood is shown in formula (4);
(3-4) calculating a of the point according to the formula (5) and the formula (6) i And b i ，
Thereby obtaining an optimal solution of the formula (7);
(3-5) equation (7) minimizes the reconstruction residual of equation (8), i.e., the spatial position of the point, q, can be updated according to equation (8) i ＝a i p i +b i (8)
Wherein p is i And q i Is the point before and after the guiding filtration, a i Is a scalar, b i Is a vector of 3*1;
(3-6) traversing and updating all data points in each block; judging whether the data point is the last data point, if so, finishing a filtering fairing algorithm, otherwise, returning to the step (3-2), and iterating;
(3-7) outputting the three-dimensional point cloud data Q after fairing j；
Step 4: for the point cloud data Q processed in the step 3 j Recovering and integrating, and outputting a result Q;
according to the point cloud data Q processed by M in the step 3 j Integration into output point cloud q= { Q j },j＝1,2,...,N；
According to the parameter M in the step 2, the point cloud data Q processed in the step 3 is processed j Integration into output point cloud q= { Q j },j＝1,2,...,N
And (3) recovering and integrating the partitioned point clouds according to a formula (9), and finally outputting a result graph processed by the algorithm.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010483771.6A CN111754421B (en) | 2020-06-01 | 2020-06-01 | Improved guiding filtering three-dimensional scattered point cloud rapid fairing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010483771.6A CN111754421B (en) | 2020-06-01 | 2020-06-01 | Improved guiding filtering three-dimensional scattered point cloud rapid fairing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111754421A CN111754421A (en) | 2020-10-09 |
| CN111754421B true CN111754421B (en) | 2023-08-18 |
Family
ID=72674529
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010483771.6A Active CN111754421B (en) | 2020-06-01 | 2020-06-01 | Improved guiding filtering three-dimensional scattered point cloud rapid fairing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111754421B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116843563A (en) * | 2023-06-25 | 2023-10-03 | 成都飞机工业（集团）有限责任公司 | Point cloud noise reduction processing method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2007146069A2 (en) * | 2006-06-07 | 2007-12-21 | Carnegie Mellon University | A sketch-based design system, apparatus, and method for the construction and modification of three-dimensional geometry |
| CN102629367A (en) * | 2012-01-17 | 2012-08-08 | 安徽建筑工业学院 | Bilateral filtering de-noising method of point cloud data based on KDTree |
| WO2018036138A1 (en) * | 2016-08-24 | 2018-03-01 | 大连理工大学 | Method for processing actually measured three-dimensional morphology point cloud data of thin-wall shell obtained for digital photography |
| CN109146817A (en) * | 2018-08-23 | 2019-01-04 | 西安工业大学 | A kind of method for processing noise of non-iterative single object scattered point cloud data |
-
2020
- 2020-06-01 CN CN202010483771.6A patent/CN111754421B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2007146069A2 (en) * | 2006-06-07 | 2007-12-21 | Carnegie Mellon University | A sketch-based design system, apparatus, and method for the construction and modification of three-dimensional geometry |
| CN102629367A (en) * | 2012-01-17 | 2012-08-08 | 安徽建筑工业学院 | Bilateral filtering de-noising method of point cloud data based on KDTree |
| WO2018036138A1 (en) * | 2016-08-24 | 2018-03-01 | 大连理工大学 | Method for processing actually measured three-dimensional morphology point cloud data of thin-wall shell obtained for digital photography |
| CN109146817A (en) * | 2018-08-23 | 2019-01-04 | 西安工业大学 | A kind of method for processing noise of non-iterative single object scattered point cloud data |
Non-Patent Citations (1)
| Title |
|---|
| 基于方法库的点云去噪与精简算法;李仁忠;杨曼;冉媛;张缓缓;景军锋;李鹏飞;;激光与光电子学进展(01);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111754421A (en) | 2020-10-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111445488B (en) | Method for automatically identifying and dividing salt body by weak supervision learning | |
| CN112184752A (en) | Video target tracking method based on pyramid convolution | |
| CN113052955B (en) | Point cloud completion method, system and application | |
| CN113298833A (en) | Target object point cloud characteristic line and surface extraction method and system | |
| CN111340723B (en) | Terrain-adaptive airborne LiDAR point cloud regularization thin plate spline interpolation filtering method | |
| CN107393005B (en) | Three-dimensional rock and stone system modeling method | |
| CN101996328B (en) | Wood identification method | |
| CN112945196B (en) | Strip mine step line extraction and slope monitoring method based on point cloud data | |
| CN108416801B (en) | Har-SURF-RAN characteristic point matching method for stereoscopic vision three-dimensional reconstruction | |
| CN104299263A (en) | Method for modeling cloud scene based on single image | |
| CN109146817B (en) | Noise processing method for non-iterative single object scattered point cloud data | |
| CN105354829A (en) | Self-adaptive point cloud data segmenting method | |
| CN111754421B (en) | Improved guiding filtering three-dimensional scattered point cloud rapid fairing method | |
| CN105389471A (en) | Method for reducing training set of machine learning | |
| CN106530247A (en) | Multi-scale image restoring algorithm based on structure information | |
| CN117011175A (en) | Mine three-dimensional model point cloud data combined filtering method and medium | |
| CN112508851A (en) | Mud rock lithology recognition system based on CNN classification algorithm | |
| CN116091771A (en) | Method, device and equipment for partitioning point cloud of cavity of complex casing | |
| CN102800077B (en) | Bayes non-local mean image restoration method | |
| CN117152344B (en) | Tunnel surrounding rock structural surface analysis method and system based on photo reconstruction point cloud | |
| CN113688842A (en) | Local image feature extraction method based on decoupling | |
| CN109308457B (en) | Multi-particle three-dimensional tracking method under high concentration | |
| CN104103061A (en) | Medical X ray image segmentation method and system | |
| CN104766085A (en) | Multi-scale figure recognition method | |
| CN110824690B (en) | Holographic microscopic equipment for judicial marking detection based on big data |
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 |