CN104680530A - ICP algorithm - Google Patents
ICP algorithm Download PDFInfo
- Publication number
- CN104680530A CN104680530A CN201510091075.XA CN201510091075A CN104680530A CN 104680530 A CN104680530 A CN 104680530A CN 201510091075 A CN201510091075 A CN 201510091075A CN 104680530 A CN104680530 A CN 104680530A
- Authority
- CN
- China
- Prior art keywords
- point
- point set
- algorithm
- icp algorithm
- icp
- 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
Abstract
The invention discloses an ICP algorithm and provides a novel method for collecting aligning points. The method adopts the nonlinear optimization algorithm for directly minimizing the aligning error and is similar to the ICP algorithm in speed, the ICP algorithm is specially used for the special application of the aligning work. As the program can directly minimize one energy function, the energy function is easily expanded to the merged robustness estimation through the Hunber kernel, and the generated interval of convergence is plenty of times wider than the interval of convergence generated by the existing method. The minimized data structure based on tangent plane distance conversion is provided, and the algorithm generated by the conversion is faster than the method provided before and higher in robustness.
Description
Technical field
The invention belongs to computerized algorithm field, more particularly, the present invention relates to a kind of ICP algorithm.
Background technology
Iteration with regard to the development of proximal method through the more than ten years with regard to proximal method ICP, constantly obtains and improves and supplement.The people such as Chen and Medioni and Bergevin proposes the Precision Registration of point-to-plane search with regard near point.Rusinkiewicz and Levoy proposes the rapid registering method of point-to-p rojection search with regard near point.Soon-Yong and Murali proposes the method for registering of Contractive-projection-point search with regard near point.In addition, Andrew and Sing is extracted the Registration of Measuring Data method based on color three dimension scan-data point texture information, in ICP algorithm, mainly consider that the texture color information of 3-D scanning point carries out searching for regard near point.The people such as Natasha analyze the cloud data quality of registration problem in ICP algorithm, and three dimensions R3 exists the point set that two groups contain n coordinate points, is respectively: PL and PR.In three dimensions point set PL each point through three dimensions conversion after with point set PR mid point one_to_one corresponding.
Summary of the invention
Problem to be solved by this invention is to provide a kind of based on the minimized data structure of tangent plane range conversion, and the ICP algorithm of the fast strong robustness of speed.
To achieve these goals, the technical scheme that the present invention takes is:
A kind of figure ICP algorithm, comprises the steps:
(1) according to the point coordinate in point set Plk, curved surface S searches for corresponding closest approach point set Prk;
(2) the centre of gravity place coordinate of two point sets is calculated;
(3) calculate positive definite matrix N by new point set, and calculate the maximal eigenvector of N;
(4) due to maximal eigenvector be equivalent to residual sum of squares (RSS) minimum time rotation hypercomplex number, hypercomplex number is converted to rotation matrix R;
(5) after rotation matrix R is determined, be only the center of gravity difference of two point sets by translation vector t;
(6) according to formula P
r={ P
r1, P
r2, P
r3, P
rm; P
ri∈ R}, calculates postrotational point set P ' lk by point set Plk.By Plk and P ' lk calculates square distance and value is fk+1;
(7) at that time, ICP registration Algorithm just stops iteration, otherwise repeats 1 to 6 step, until stop iteration after satisfying condition.
Preferably, the method for described step (1) mean camber S searching for closest approach is Point to Point.
Preferably, increase in described step (2) and carry out point set centralization and generate new point set.
Preferably, the eigenvalue of maximum calculating positive definite matrix N is increased in described step (3).
Preferably, determine in described step 5) that the large method of two point set center of gravity difference is can be determined by the focus point in two coordinate systems and rotation matrix.
Preferably, in described (6) using double square distance and difference absolute value judge numerical value as iteration.
The invention provides a kind of ICP algorithm, propose a kind of new method of alignment point set.The method adopts alignment error directly to be minimized by nonlinear optimization algorithm, and speed can be equal to mutually with ICP algorithm, and ICP algorithm is the algorithm of the specific use being specifically designed to alignment work.Because program directly minimizes an energy function, so easily via Hunber core, it is extended in the Robust estimation of merging, the interval of convergence of generation is wider than the interval of convergence of Existing methods a lot of times.Finally introduce a kind of based on the minimized data structure of tangent plane range conversion, the algorithm that this conversion produces is fast and strong robustness than the method speed described in the past.
Embodiment
An ICP algorithm for images match, comprises the steps:
(1) according to the point coordinate in point set Plk, curved surface S searches for corresponding closest approach point set Prk, the method for curved surface S searching for closest approach is Point to Point;
(2) calculate the centre of gravity place coordinate of two point sets, carry out point set centralization and generate new point set;
(3) calculate positive definite matrix N by new point set, and calculate maximal eigenvector and the eigenvalue of maximum of N;
(4) due to maximal eigenvector be equivalent to residual sum of squares (RSS) minimum time rotation hypercomplex number, hypercomplex number is converted to rotation matrix R;
(5) after rotation matrix R is determined, be only the center of gravity difference of two point sets by translation vector t, can be determined by the focus point in two coordinate systems and rotation matrix;
(6) according to formula P
r={ P
r1, P
r2, P
r3, P
rm; P
ri∈ R}, calculates postrotational point set P ' lk by point set Plk.By Plk and P ' lk calculates square distance and value for fk+1, using double square distance and difference absolute value judge numerical value as iteration;
(7) at that time, ICP registration Algorithm just stops iteration, otherwise repeats 1 to 6 step, until stop iteration after satisfying condition;
The invention provides a kind of ICP algorithm, propose a kind of new method of alignment point set.The method adopts alignment error directly to be minimized by nonlinear optimization algorithm, and speed can be equal to mutually with ICP algorithm, and ICP algorithm is the algorithm of the specific use being specifically designed to alignment work.Because program directly minimizes an energy function, so easily via Hunber core, it is extended in the Robust estimation of merging, the interval of convergence of generation is wider than the interval of convergence of Existing methods a lot of times.Finally introduce a kind of based on the minimized data structure of tangent plane range conversion, the algorithm that this conversion produces is fast and strong robustness than the method speed described in the past.
The foregoing is only embodiments of the invention; not thereby the scope of the claims of the present invention is limited; every utilize description of the present invention to do equivalent structure or equivalent flow process conversion; or be directly or indirectly used in other relevant technical fields, be all in like manner included in scope of patent protection of the present invention.
Claims (6)
1. an ICP algorithm, is characterized in that, comprises the steps:
(1) according to the point coordinate in point set Plk, curved surface S searches for corresponding closest approach point set Prk;
(2) the centre of gravity place coordinate of two point sets is calculated;
(3) calculate positive definite matrix N by new point set, and calculate the maximal eigenvector of N;
(4) due to maximal eigenvector be equivalent to residual sum of squares (RSS) minimum time rotation hypercomplex number, hypercomplex number is converted to rotation matrix R;
(5) after rotation matrix R is determined, be only the center of gravity difference of two point sets by translation vector t;
(6) according to formula P
r={ P
r1, P
r2, P
r3, P
rm; P
ri∈ R}, calculates postrotational point set P ' lk by point set Plk;
By Plk and P ' lk calculates square distance and value is fk+1;
(7) at that time, ICP registration Algorithm just stops iteration, otherwise repeats 1 to 6 step, until stop iteration after satisfying condition.
2. ICP algorithm as claimed in claim 1, is characterized in that: the method for described step (1) mean camber S searching for closest approach is Point to Point.
3. ICP algorithm as claimed in claim 1, is characterized in that: the middle increase of described step (2) is carried out point set centralization and generated new point set.
4. ICP algorithm as claimed in claim 1, is characterized in that: increase the eigenvalue of maximum calculating positive definite matrix N in described step (3).
5. ICP algorithm as claimed in claim 1, is characterized in that: determine in described step (5) that the large method of two point set center of gravity difference is can be determined by the focus point in two coordinate systems and rotation matrix.
6. ICP algorithm as claimed in claim 5, its feature exists: in described step (6) using double square distance and difference absolute value judge numerical value as iteration.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510091075.XA CN104680530A (en) | 2015-03-01 | 2015-03-01 | ICP algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510091075.XA CN104680530A (en) | 2015-03-01 | 2015-03-01 | ICP algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104680530A true CN104680530A (en) | 2015-06-03 |
Family
ID=53315523
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510091075.XA Pending CN104680530A (en) | 2015-03-01 | 2015-03-01 | ICP algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104680530A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107644433A (en) * | 2017-08-30 | 2018-01-30 | 北京控制工程研究所 | Improved closest approach iteration point cloud registration method |
| CN111882614A (en) * | 2020-07-30 | 2020-11-03 | 南京溧航仿生产业研究院有限公司 | KNN-ICP algorithm-based free-form surface positioning method |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103236064A (en) * | 2013-05-06 | 2013-08-07 | 东南大学 | Point cloud automatic registration method based on normal vector |
| CN103955939A (en) * | 2014-05-16 | 2014-07-30 | 重庆理工大学 | Boundary feature point registering method for point cloud splicing in three-dimensional scanning system |
-
2015
- 2015-03-01 CN CN201510091075.XA patent/CN104680530A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103236064A (en) * | 2013-05-06 | 2013-08-07 | 东南大学 | Point cloud automatic registration method based on normal vector |
| CN103955939A (en) * | 2014-05-16 | 2014-07-30 | 重庆理工大学 | Boundary feature point registering method for point cloud splicing in three-dimensional scanning system |
Non-Patent Citations (1)
| Title |
|---|
| 郑德华: ""ICP算法及其在建筑物扫描点云数据配准中的应用"", 《测绘科学》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107644433A (en) * | 2017-08-30 | 2018-01-30 | 北京控制工程研究所 | Improved closest approach iteration point cloud registration method |
| CN107644433B (en) * | 2017-08-30 | 2020-07-14 | 北京控制工程研究所 | Improved closest point iteration point cloud registration method |
| CN111882614A (en) * | 2020-07-30 | 2020-11-03 | 南京溧航仿生产业研究院有限公司 | KNN-ICP algorithm-based free-form surface positioning method |
| WO2022021479A1 (en) * | 2020-07-30 | 2022-02-03 | 南京溧航仿生产业研究院有限公司 | Freeform surface positioning method based on knn-icp algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101692257B (en) | Method for registering complex curved surface | |
| CN103955939A (en) | Boundary feature point registering method for point cloud splicing in three-dimensional scanning system | |
| CN107044852B (en) | Total station survey method under out-of-flatness state | |
| CN103886593B (en) | A kind of based on three-dimensional point cloud curved surface circular hole detection method | |
| CN106584090A (en) | Workpiece assembling method based on structured light three-dimensional measuring system | |
| CN110363707A (en) | A kind of more view three-dimensional point cloud joining methods based on bridle virtual feature | |
| CN104392476A (en) | Method of extracting three-dimensional axis of tunnel based on minimum bounding box algorithm | |
| CN103810271A (en) | Three-dimensional point cloud object shape feature matching method based on path following | |
| CN102831101A (en) | Point cloud data splicing method based on automatic identification of plurality of mark points | |
| CN110335296A (en) | A kind of point cloud registration method based on hand and eye calibrating | |
| CN103278126A (en) | Sphericity error assessment method for part based on minimum area | |
| CN108959182A (en) | The small feature loss gravitational field modeling method returned based on Gaussian process | |
| CN103294920B (en) | A kind of part Cylindricity error evaluation based on Minimum Area | |
| Zheng et al. | Reassembling 3D thin fragments of unknown geometry in cultural heritage | |
| CN104680530A (en) | ICP algorithm | |
| Cheung et al. | Measurement and characterization of ultra-precision freeform surfaces using an intrinsic surface feature-based method | |
| CN101847262A (en) | Fast three-dimensional point cloud searching and matching method | |
| CN103292773A (en) | Symmetry error evaluation method based on minimum zone | |
| Chen et al. | Stereo vision sensor calibration based on random spatial points given by CMM | |
| CN102968400A (en) | Multi-view three-dimensional data registration method based on spatial line recognition and matching | |
| CN105157572A (en) | Center offset error compensation method used for aspheric annular subaperture stitching | |
| CN102162728A (en) | Method for evaluating minimum area of line profile error of cross section of skirt part of variable-ellipse piston | |
| CN106767405B (en) | The method and device of the quick corresponding point matching of phase mapping assist three-dimensional imaging system | |
| CN201576308U (en) | Registration system for complex curved surface | |
| CN110415281A (en) | A kind of point set rigid registration method based on Loam curvature weighting |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150603 |
|
| WD01 | Invention patent application deemed withdrawn after publication |