CN108376408A - A kind of three dimensional point cloud based on curvature feature quickly weights method for registering - Google Patents
A kind of three dimensional point cloud based on curvature feature quickly weights method for registering Download PDFInfo
- Publication number
- CN108376408A CN108376408A CN201810091369.6A CN201810091369A CN108376408A CN 108376408 A CN108376408 A CN 108376408A CN 201810091369 A CN201810091369 A CN 201810091369A CN 108376408 A CN108376408 A CN 108376408A
- Authority
- CN
- China
- Prior art keywords
- point
- matched
- matching
- point cloud
- points
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
- G06T7/33—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06F18/2413—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on distances to training or reference patterns
- G06F18/24147—Distances to closest patterns, e.g. nearest neighbour classification
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The present invention discloses the three dimensional point cloud based on curvature feature and quickly weights method for registering, including：Matching point cloud data P and model point cloud data M is treated respectively and carries out down-sampling according to curvature feature, obtains to be matched cloud sample data P ' and model point cloud sample data M '；The curvature feature of sampled point is calculated, and multiple points pair to be matched are obtained using the match point of index acceleration method each point to be matched in search P ' in M ' according to the curvature feature being calculated；To obtained multiple points pair to be matched, the screening based on Euclidean distance is carried out to remove the point pair to be matched that Euclidean distance is more than a predetermined threshold value；To the point to be matched after screening to weighted least-squares method is slightly matched again using iteration, preliminary rigid body translation matrix is obtained；P and M is slightly matched using preliminary rigid body translation matrix；Using preliminary rigid body translation matrix as initial rigid body translation matrix, using by distance weighted Trimmed ICP algorithms, smart matching is carried out to thick matching result.
Description
Technical Field
The invention relates to the field of registration of three-dimensional point cloud data, in particular to a three-dimensional point cloud data rapid weighting registration method based on curvature features.
Background
In recent years, three-dimensional point cloud data is widely applied to the fields of three-dimensional modeling, target identification, object surface detection and the like, and a multi-angle rapid and accurate registration technology of the three-dimensional point cloud data is one of the hotspots of the current research.
The aim of three-dimensional point cloud data registration is to find a three-dimensional rigid body transformation and quickly and accurately match and fuse discrete three-dimensional point cloud data of different angles of an object. Theoretically, if three non-coplanar corresponding point pairs determined on two groups of point clouds can be found, the registration of the point cloud data can be accurately and quickly completed, but the three determined corresponding point pairs are difficult to find in practice due to the limited resolution and precision of the data acquisition equipment, so that the point cloud registration problem is actually the data optimization problem.
The traditional ICP (Iterative Closest Points) algorithm is widely applied, however, the method is to calculate the minimum Euclidean distance point by point, the operation amount is huge, the time consumption is long, the operation speed and the convergence of the algorithm depend on the given initial transformation estimation to a great extent, and the inaccurate initial transformation estimation can directly cause the registration result to fall into a local optimal value; in addition, registration failure can also result if only a small portion of the point cloud data overlaps; on the other hand, with the rapid development of the technology, the amount of the obtained three-dimensional point cloud point data is huge, and how to achieve rapid and high-precision registration becomes a difficult point for research.
The above background disclosure is only for the purpose of assisting understanding of the inventive concept and technical solutions of the present invention, and does not necessarily belong to the prior art of the present patent application, and should not be used for evaluating the novelty and inventive step of the present application in the case that there is no clear evidence that the above content is disclosed before the filing date of the present patent application.
Disclosure of Invention
The invention mainly aims to overcome the defects of the traditional ICP algorithm, improve the practicability of the algorithm, provide a curvature feature-based three-dimensional point cloud data rapid weighting registration method, utilize the rotation invariance of the curvature feature and a KD-Tree rapid nearest neighbor search method based on the curvature change rate of the point cloud data, after removing edge points and error matching points, utilize an IRLS-ICP algorithm of iterative reweighting to perform rough matching on the points to be matched, then utilize a TrimmedICP algorithm weighted according to distance to perform fine matching, obtain rigid body transformation with higher accuracy, and perform rapid and accurate registration.
The technical scheme provided by the invention for achieving the purpose is as follows:
a three-dimensional point cloud data rapid weighting registration method based on curvature features is used for registering two groups of three-dimensional point cloud data, and comprises the following steps:
s1, respectively carrying out down-sampling on the point cloud data to be matched and the model point cloud data according to curvature characteristics to obtain point cloud sample data to be matched and model point cloud sample data;
s2, calculating curvature characteristics of the point cloud sample data to be matched and sampling points in the model point cloud sample data, and searching matching points of the point cloud sample data to be matched in the model point cloud sample data by using an index acceleration method according to the calculated curvature characteristics to obtain a plurality of point pairs to be matched;
s3, screening the plurality of point pairs to be matched obtained in the step S2 based on Euclidean distance to remove the point pairs to be matched with the Euclidean distance larger than a preset threshold value;
s4, carrying out rough matching on the point pairs to be matched after being screened in the step S3 by using an iteration reweighting least square method to obtain a preliminary rigid body transformation matrix;
s5, performing rough matching on the point cloud data to be matched and the model point cloud data by using the preliminary rigid body transformation matrix to obtain a rough matching result;
s6, taking the preliminary rigid body transformation matrix obtained in the step S4 as an initial rigid body transformation matrix, and carrying out fine matching on the coarse matching result obtained in the step S5 by using a distance-weighted Trimmed ICP algorithm.
Compared with the prior art, the technical scheme provided by the invention has the following beneficial effects:
1) because the curvature features represented by the feature values have rotational invariance, massive original point cloud data are sampled by adopting a method of sampling according to the curvature features, and the integrity of the geometric features of the point cloud data can be ensured as much as possible on the basis of simplifying the data and removing redundancy;
2) in the searching strategy of the matching points, one of the geometric features of the point cloud data, namely curvature features, is utilized, and the searching of the matching points is accelerated by adopting an index acceleration method;
3) sorting according to the Euclidean distance of the nearest neighbor matching points, removing error points in a certain proportion, and further weakening the influence of error data on matching precision;
4) the iterative reweighting matching algorithm IRLS-ICP is applied to carry out coarse matching, so that the robustness of the coarse matching can be improved;
5) the method of weighting according to the distance is added into the TrimmedICP algorithm for fine matching, so that not only can a more accurate number of matching point pairs be obtained, but also the effect of correct matching point pairs can be further enhanced, the influence of wrong matching point pairs is weakened, and the precision and the robustness of the result are greatly improved.
Drawings
Fig. 1 is a flowchart of a curvature feature-based three-dimensional point cloud data fast weighted registration method according to an embodiment of the present invention;
FIGS. 2-1 and 2-2 are schematic diagrams of two examples of point cloud data to be matched and model point cloud data;
3-1 and 3-2 are schematic diagrams of registration results obtained by respectively registering the point cloud data of FIG. 2-1 and FIG. 2-2 by using a conventional ICP algorithm;
4-1 and 4-2 are schematic diagrams of registration results obtained by registering the point cloud data of FIGS. 2-1 and 2-2 by using the method of the present invention;
fig. 5 is a graphical comparison of the difference in the change in MSE for the conventional ICP algorithm and the method of the invention.
Detailed Description
The invention is further described with reference to the following figures and detailed description of embodiments.
The specific embodiment of the invention provides a curvature feature-based three-dimensional point cloud data rapid weighting registration method (hereinafter referred to as a "registration method"), which can be used for registering any two groups of mass point cloud data. Referring to fig. 1, the registration method includes the following steps S1 to S6:
and step S1, respectively carrying out down-sampling on the point cloud data P to be matched and the model point cloud data M according to curvature characteristics to obtain point cloud sample data P 'to be matched and the model point cloud sample data M'. Wherein the down-sampling can be performed with a curvature-characterized sampling function in matlab. In mass point cloud data, for any point x, there are generally two types of geometric features describing the point, namely, a feature value and a corresponding feature vector. Curvature features are an important basis for feature recognition, and curvature features of points are reversedThe concave-convex degree of the point on the surface of the point cloud is reflected, and the characteristic value of the point cloud can effectively search for matching points of two groups of scattered point clouds. For example, referring to fig. 2, the point cloud data P to be matched is derived from an actual model 100 of the object, P ═ P1,p2…,pm}; model point cloud data M is derived from an ideal model 200 of an object, M ═ y1,y2…,yn}。
Step S2, curvature features of sampling points in the point cloud sample data P ' to be matched and the model point cloud sample data M ' are calculated, and matching points of all points to be matched in the point cloud sample data M ' to be matched are searched by an index acceleration method according to the curvature features obtained through calculation, so that a plurality of point pairs to be matched are obtained. The curvature characteristics of the points can be obtained by analyzing covariance matrixes of k nearest neighbor points of the points, the value of k influences a matching result, and k represents the number of the points in a search radius of the points, so that in order to enhance robustness, the nearest neighbor points are searched by gradually enlarging the search radius, and then the difference value of the curvature characteristics calculated under different search radii is taken as matching information to search for matching points for two groups of scattered point cloud data.
In a preferred embodiment, each point P to be matched in P 'is searched in M' by using a KD-Tree acceleration methodiMatching point y ofiThe process specifically comprises the following steps: for each point p to be matchediCalculating curvature characteristics obtained by calculation under different search radiuses, calculating curvature characteristic difference values between adjacent search radiuses, and forming a matrix by the calculated curvature characteristic difference values to serve as a matching calculation matrix of the points to be matched; and searching the nearest point of the point to be matched in the model point cloud data by utilizing a KD-Tree acceleration method based on the matching calculation matrix so as to obtain the matching point of the point to be matched. For example, for any point p to be matched, assuming there are D different search radii, then for any search radius rdIts covariance matrix CdComprises the following steps:
wherein D is 1, …, D, Kd＝{xi|||xi-p||≤rdIs the search radius rdNumber of interior points, xiFor the searched radius rdAnd (3) a nearest neighbor point set of the inner point P, wherein P is any point in the point cloud sample data P' to be matched. 3 x 3 covariance matrix C obtained by singular value decompositiondSo as to obtain three eigenvalues and eigenvectors thereof, and selecting three eigenvalues lambda from large to smalld1、λd2、λd3Describing the curvature characteristic of the point, normalizing the vector consisting of the three characteristic values to obtain the p-in-search radius rdLower curvature feature sdAnd is recorded as:
since the curvature features represented by the feature values have rotation invariance, s is gradually increased along with the gradually increased search radiusdSlowly varying, therefore, the difference Δ s of curvature characteristics under adjacent search radii is chosendMatching information as a matching point of the search is defined as Δ sd＝sd+1-sdAll Δ s at that pointdForming a matrix N as a matching calculation matrix when searching for matching points, wherein the expression is as follows:
based on the matrix N, points P to be matched in P 'can be quickly found in M' (or in M) by using a KD-Tree acceleration methodiMatching point y ofiThereby obtaining a plurality of point pairs to be matched<pi,yi>。
Step S3, performing euclidean distance-based screening on the plurality of point pairs to be matched obtained in step S2 to remove the point pairs to be matched whose euclidean distance is greater than a preset threshold. Since the point pair to be matched obtained in step S2 has an error inevitably, that is, two points in some point pairs may not actually be matched, in this step, a screening based on euclidean distance is adopted to remove the point pair that may be matched incorrectly, which is a key for performing the rough matching of the point cloud subsequently.
In a specific embodiment, the step S3 of performing the euclidean distance-based screening process substantially includes: firstly, calculating the Euclidean distance between two points in each point pair to be matched; secondly, sorting all the point pairs to be matched according to the Euclidean distance, and then removing 10% of the point pairs to be matched with the worst point (the greater the Euclidean distance is, the worse is). Alternatively, a threshold value may be preset, and the point pairs that are more than the threshold value in euclidean distance may be removed.
In a more preferred embodiment, the secondary screening for removing the sharp point can be further performed on the point pairs to be matched after the euclidean distance-based screening. The secondary screening adopts a triangulation point cloud data method to find sharp points, any one of the two groups of point cloud data can be triangulated to find the sharp points outside a triangular plane, and then the sharp points and matching points of the sharp points in the other group of point cloud data are removed together. Such as: triangulation is carried out on the point cloud data P to be matched, then sharp points located outside a triangular plane are found out from the points to be matched of the screened matching point pairs based on Euclidean distance, and the matching point pairs containing the sharp points are deleted; or triangulating the model point cloud data M, finding out sharp points out of a triangular plane from model points of the screened matching point pairs based on the Euclidean distance, and deleting the matching point pairs containing the sharp points.
After the wrong matching points and the point pairs to be matched where the boundary points are located are removed through the two screening, the disturbance of the matching results of the wrong points is greatly weakened, and a foundation is laid for the next coarse registration.
And step S4, performing coarse matching on the point pairs to be matched, which are screened in the step S3, by using an iterative reweighting least squares method (IRLS-ICP), and obtaining a preliminary rigid body transformation matrix. In the course of coarse registration, an iterative weighted least squares method (IRLS-ICP) is applied, and the registration of two groups of scattered point clouds is essentially to find a rigid transformation matrix containing a rotation matrix R and a translation vector t, and the rigid transformation matrix can transform two groups of mass scattered point cloud data in different coordinate systems into the same coordinate system and realize accurate registration and registration. The algorithm (IRLS-ICP) essentially comprises two steps: finding the nearest neighbor point and finding a rigid transformation matrix which can match the two groups of point clouds specifically as follows:
first, a rigid body transformation (R) is initialized(0)，t(0))，R(0)＝I,t(0)0. Setting symbol C as nearest neighbor operator, then finding out point p to be matchediNearest neighbor y in model point cloud data M (or M')iThe operation formula of (c) can be expressed as:
where k-1 represents the number of iterations.
Thus, iteratively updated rigid body transformations (R)*，t*) Can be obtained by the following formula:
[R*,t*]＝argming(R,t) (5)
wherein,
wherein w (r) is Tukey's bi-weight function, r is Euclidean distance between two points in the point pair,kTu7.0589, Tukey's bi-weight weighting function is very robust to erroneous interference points and can eliminate unnecessary erroneous points.
And after the iterative update of the rigid body transformation is completed by using all the point pairs to be matched, the rough matching of the point pairs to be matched is completed, and a preliminary rigid body transformation matrix is obtained.
And S5, performing rough matching on the point cloud data P to be matched and the model point cloud data M by using the preliminary rigid body transformation matrix obtained in the step S4 to obtain a rough matching result.
And S6, taking the preliminary rigid body transformation matrix obtained in the step S4 as an initial rigid body transformation matrix, and performing fine matching on the coarse matching result obtained in the step S5 by using a distance-weighted Trimmed ICP algorithm. Thus, the accurate registration of the two groups of point cloud data P and M is completed.
In order to overcome the limitations of the conventional ICP algorithm, the improved method is generally studied from three aspects of improving the robustness, convergence (convergence speed) and accuracy of the algorithm. The core of the invention is to reduce redundant data as much as possible on the basis of ensuring the integrity of data, properly simplify the number of data points to improve the operation speed, and simultaneously increase the number of correct matching points as much as possible, particularly points with obvious characteristics, so as to improve the robustness and accuracy of the algorithm. In order to reduce the error substituted by the wrong matching point pair, the effect of the correct matching point pair can be strengthened by weighting the weighting coefficient of the point pair to be matched, and the influence of the wrong matching point pair can be reduced. Therefore, step S6 proposes to add the weighted euclidean distance to the Trimmed ICP algorithm for fine matching. The specific method comprises the following steps:
firstly, a weighting coefficient based on the euclidean distance is calculated for each matching point pair finally obtained by the Trimmed ICP algorithm, for example, a matching point pair (p, q), wherein the weighting coefficient represented by the euclidean distance is as follows:
then, using a weighting factor wPAnd then carrying out fine matching on the point cloud data P' and the point cloud data M.
The Trimmed ICP algorithm is an improved version of the traditional ICP algorithm for improving robustness, the algorithm sorts the square errors of all matching point pairs, only optimizes a certain number of smaller values, the number is obtained according to the initial overlapping rate of two groups of point cloud data, and the Trimmed ICP has better convergence rate and robustness. The method adds the distance weighting method into the Trimmed ICP algorithm, so that not only is a more accurate number of matching point pairs obtained, but also the effect of correct matching point pairs can be further enhanced, the influence of wrong matching point pairs is weakened, the calculation amount of mass data is simplified, and the accuracy of results is improved.
The curvature characteristic-based fast weighted registration method provided by the invention is subjected to simulation verification to evaluate the effectiveness of the algorithm. In particular, the method of the present invention is compared to the commonly used ICP algorithm. Comparative evaluations were performed using the Bunny dataset and Dragon dataset in the Stanford university graphical Laboratory dataset. In the experiment, the real-time performance, the accuracy and the robustness of the traditional ICP algorithm and the algorithm provided by the invention are evaluated and compared by using the time consumed by the algorithm, the Mean Square Error (MSE) and the change difference value of the MSE and the time curve.
As shown in fig. 2-1 and 2-2, the Bunny data set and the Dragon data set used each include a point cloud model 100 to be matched (all three-dimensional point clouds on the model constitute point cloud data P to be matched) and a theoretical point cloud model 200 (all three-dimensional point clouds on the model constitute model point cloud data M). For the same two data sets, the Bunny data set and the Dragon data set, the results obtained after the registration by the registration method of the invention are shown in fig. 4-1 and 4-2, while the results obtained after the registration by the traditional ICP algorithm are shown in fig. 3-1 and 3-2, it can be seen that the registration result of the invention is more accurate, and the registration result in fig. 3-1 and 3-2 is poorer, such as the configuration shown in fig. 3-1As a result, there are several regions such as 10, 20, 30, etc. that do not overlap. Experimental results show that the conventional ICP algorithm has a certain limitation on initial conditions, so that an accurate registration result is difficult to obtain, and as the complexity of point cloud data (such as a Dragon data set) is increased, the registration result is worse than that of a Bunny data set with lower complexity. In reverse view of the present invention, the proposed algorithm can accomplish accurate registration of both data sets. On the other hand, through performance comparison, the traditional ICP algorithm finds the nearest neighbor point by a point-by-point Euclidean distance method, the calculated amount is huge, the time consumption is long, the matching time is about 113.2s, the mean square error is 1.594 multiplied by 10-5And m is selected. The algorithm provided by the invention has good registration effect in time and precision, the matching time is about 31.1s, and the mean square error is 1.97 multiplied by 10-6m, the matching precision is improved by nearly ten times.
Comparing the change difference of the MSE of the invention with the MSE of the traditional ICP algorithm, the curve shown in FIG. 5 can be obtained, the abscissa represents time, and the ordinate represents the change difference of the MSE of the algorithm, and the graph shows that: the traditional ICP algorithm converges very slowly and is also accompanied by large fluctuations; the algorithm provided by the invention has better effect in convergence and robustness.
With the rapid development of the technology, the amount of the acquired three-dimensional point cloud data is more and more huge, and how to process massive three-dimensional data and achieve rapid and high-precision registration becomes a difficult point for research. Aiming at the limitation of the traditional ICP algorithm in the scattered three-dimensional point cloud registration, the algorithm firstly carries out down-sampling according to curvature characteristics, quickly searches matching points by utilizing the curvature characteristics, removes error point pairs, carries out rough matching on the scattered point cloud by utilizing the IRLS-ICP algorithm, and finally carries out fine matching by utilizing the TrimmedICP algorithm weighted according to distance. The experimental results prove that the method of the invention achieves good convergence and robustness, and greatly reduces the registration time.
The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several equivalent substitutions or obvious modifications can be made without departing from the spirit of the invention, and all the properties or uses are considered to be within the scope of the invention.
Claims (8)
1. A three-dimensional point cloud data rapid weighting registration method based on curvature features is used for registering two groups of three-dimensional point cloud data, and is characterized by comprising the following steps:
s1, respectively carrying out down-sampling on the point cloud data to be matched and the model point cloud data according to curvature characteristics to obtain point cloud sample data to be matched and model point cloud sample data;
s2, calculating curvature characteristics of the point cloud sample data to be matched and sampling points in the model point cloud sample data, and searching matching points of the point cloud sample data to be matched in the model point cloud sample data by using an index acceleration method according to the calculated curvature characteristics to obtain a plurality of point pairs to be matched;
s3, screening the plurality of point pairs to be matched obtained in the step S2 based on Euclidean distance to remove the point pairs to be matched with the Euclidean distance larger than a preset threshold value;
s4, carrying out rough matching on the point pairs to be matched after being screened in the step S3 by using an iteration reweighting least square method to obtain a preliminary rigid body transformation matrix;
s5, performing rough matching on the point cloud data to be matched and the model point cloud data by using the preliminary rigid body transformation matrix to obtain a rough matching result;
s6, taking the preliminary rigid body transformation matrix obtained in the step S4 as an initial rigid body transformation matrix, and carrying out fine matching on the coarse matching result obtained in the step S5 by using a distance-weighted Trimmed ICP algorithm.
2. The registration method of claim 1, wherein the down-sampling in step S1 is performed using a curvature-feature-by-curvature sampling function in matlab.
3. The registration method as claimed in claim 1, wherein the curvature features of the sampling points in step S2 are obtained by analyzing covariance matrices of k nearest neighbors of the sampling points; where k is the number of points within the search radius of the sample point.
4. The registration method according to claim 3, wherein the matching point of the point to be matched is searched by gradually enlarging a search radius in the model point cloud sample data by using a KD-Tree acceleration method in step S2.
5. The registration method according to claim 4, wherein the searching for the matching point of each point to be matched using a KD-Tree acceleration method in step S2 specifically includes: for each point to be matched, curvature features obtained by calculation under different search radiuses are taken, curvature feature difference values between adjacent search radiuses are calculated, and the curvature feature difference values obtained by calculation form a matrix which is used as a matching calculation matrix of the point to be matched; and searching the nearest point of the point to be matched in the model point cloud sample data by utilizing a KD-Tree acceleration method based on the matching calculation matrix so as to obtain the matching point of the point to be matched.
6. The registration method of claim 1, wherein step S3 further includes performing a secondary filtering for removing sharp points on the filtered pairs of matching points based on euclidean distance.
7. The registration method of claim 6, wherein the secondary screening comprises:
triangulation is carried out on the point cloud data to be matched, sharp points located outside a triangular plane are found out from the points to be matched of the screened matching point pairs based on Euclidean distance, and the matching point pairs containing the sharp points are deleted; or
And triangulating the model point cloud data, finding out sharp points out of a triangular plane from model points of the screened matching point pairs based on the Euclidean distance, and deleting the matching point pairs containing the sharp points.
8. The registration method according to claim 1, wherein the fine matching of the coarse matching result obtained in step S5 by using a distance-weighted TrimmedICP algorithm in step S6 specifically comprises: and calculating a weighting coefficient based on Euclidean distance for each matching point pair obtained by the Trimmed ICP algorithm, and then performing the fine matching based on the weighting coefficient.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810091369.6A CN108376408B (en) | 2018-01-30 | 2018-01-30 | Three-dimensional point cloud data rapid weighting registration method based on curvature features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810091369.6A CN108376408B (en) | 2018-01-30 | 2018-01-30 | Three-dimensional point cloud data rapid weighting registration method based on curvature features |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108376408A true CN108376408A (en) | 2018-08-07 |
| CN108376408B CN108376408B (en) | 2020-04-03 |
Family
ID=63017044
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810091369.6A Active CN108376408B (en) | 2018-01-30 | 2018-01-30 | Three-dimensional point cloud data rapid weighting registration method based on curvature features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108376408B (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110287873A (en) * | 2019-06-25 | 2019-09-27 | 清华大学深圳研究生院 | Noncooperative target pose measuring method, system and terminal device based on deep neural network |
| CN110335297A (en) * | 2019-06-21 | 2019-10-15 | 华中科技大学 | A kind of point cloud registration method based on feature extraction |
| CN110415281A (en) * | 2019-07-30 | 2019-11-05 | 西安交通大学深圳研究院 | A kind of point set rigid registration method based on Loam curvature weighting |
| CN110473239A (en) * | 2019-08-08 | 2019-11-19 | 刘秀萍 | A kind of high-precision point cloud registration method of 3 D laser scanning |
| CN111462202A (en) * | 2020-04-08 | 2020-07-28 | 中国科学技术大学 | Non-rigid registration method and system |
| CN111612847A (en) * | 2020-04-30 | 2020-09-01 | 重庆见芒信息技术咨询服务有限公司 | Point cloud data matching method and system for robot grabbing operation |
| CN111815686A (en) * | 2019-04-12 | 2020-10-23 | 四川大学 | Coarse-to-fine point cloud registration method based on geometric features |
| CN112015138A (en) * | 2020-09-04 | 2020-12-01 | 中国工程物理研究院机械制造工艺研究所 | Blade contour error evaluation method based on K nearest neighbor iterative nearest grid algorithm |
| CN112435299A (en) * | 2021-01-28 | 2021-03-02 | 深圳大学 | Airborne point cloud assisted satellite image stereo matching point cloud precision orientation method |
| CN113470084A (en) * | 2021-05-18 | 2021-10-01 | 西安电子科技大学 | Point set registration method based on outline rough matching |
| CN114004958A (en) * | 2021-11-02 | 2022-02-01 | 广州虎牙科技有限公司 | Texture image processing method, device and electronic device |
| CN114332359A (en) * | 2021-12-10 | 2022-04-12 | 欧冶云商股份有限公司 | System, method and device for three-dimensional reconstruction of warehouse steel point cloud based on laser radar |
| CN114937122A (en) * | 2022-06-16 | 2022-08-23 | 黄冈强源电力设计有限公司 | Rapid three-dimensional model reconstruction method for cement fiberboard house |
| CN115239932A (en) * | 2022-08-02 | 2022-10-25 | 广东奥普特科技股份有限公司 | A three-dimensional data registration method and system based on geometric sampling of curvature features |
| CN115409806A (en) * | 2022-08-31 | 2022-11-29 | 深圳前海瑞集科技有限公司 | Weld joint identification method and device, computer readable medium and electronic equipment |
| CN115496667A (en) * | 2021-06-17 | 2022-12-20 | 北京四维图新科技股份有限公司 | Point cloud data smoothing method and device, electronic equipment and storage medium |
| CN117132631A (en) * | 2023-09-07 | 2023-11-28 | 重庆理工大学 | Heteroscale point cloud registration method based on pairwise constrained matching of line vectors |
| CN118397209A (en) * | 2024-06-27 | 2024-07-26 | 河北华勘地质勘查有限公司 | Visualization method and system for mine goaf measurement data |
| CN119722767A (en) * | 2025-02-28 | 2025-03-28 | 零磁装备(德清)有限公司 | An automatic co-registration method for brain MRI images and MEG sensors |
| CN120852495A (en) * | 2025-09-23 | 2025-10-28 | 天津博诺智创机器人技术有限公司 | A point cloud registration method based on curvature density feature extraction |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050089213A1 (en) * | 2003-10-23 | 2005-04-28 | Geng Z. J. | Method and apparatus for three-dimensional modeling via an image mosaic system |
| CN101645170A (en) * | 2009-09-03 | 2010-02-10 | 北京信息科技大学 | Precise registration method of multilook point cloud |
| CN103456038A (en) * | 2013-08-19 | 2013-12-18 | 华中科技大学 | Method for rebuilding three-dimensional scene of downhole environment |
| CN104112289A (en) * | 2014-01-29 | 2014-10-22 | 辽宁师范大学 | Three-dimensional object point cloud registration method based on parallel cascaded EM-ICP |
| CN104143210A (en) * | 2014-07-31 | 2014-11-12 | 哈尔滨工程大学 | A multi-scale normal feature point cloud registration method |
| CN104778443A (en) * | 2015-03-06 | 2015-07-15 | 哈尔滨理工大学 | Natural road sign recognition based on real-time local characteristic description |
| CN105654422A (en) * | 2015-12-23 | 2016-06-08 | 北京观著信息技术有限公司 | Point cloud registration method and system |
| CN106097324A (en) * | 2016-06-07 | 2016-11-09 | 中国农业大学 | A kind of non-rigid 3D shape corresponding point determine method |
| US20170301104A1 (en) * | 2015-12-16 | 2017-10-19 | Objectvideo, Inc. | Profile matching of buildings and urban structures |
| CN107492120A (en) * | 2017-07-18 | 2017-12-19 | 北京航空航天大学 | Point cloud registration method |
-
2018
- 2018-01-30 CN CN201810091369.6A patent/CN108376408B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050089213A1 (en) * | 2003-10-23 | 2005-04-28 | Geng Z. J. | Method and apparatus for three-dimensional modeling via an image mosaic system |
| CN101645170A (en) * | 2009-09-03 | 2010-02-10 | 北京信息科技大学 | Precise registration method of multilook point cloud |
| CN103456038A (en) * | 2013-08-19 | 2013-12-18 | 华中科技大学 | Method for rebuilding three-dimensional scene of downhole environment |
| CN104112289A (en) * | 2014-01-29 | 2014-10-22 | 辽宁师范大学 | Three-dimensional object point cloud registration method based on parallel cascaded EM-ICP |
| CN104143210A (en) * | 2014-07-31 | 2014-11-12 | 哈尔滨工程大学 | A multi-scale normal feature point cloud registration method |
| CN104778443A (en) * | 2015-03-06 | 2015-07-15 | 哈尔滨理工大学 | Natural road sign recognition based on real-time local characteristic description |
| US20170301104A1 (en) * | 2015-12-16 | 2017-10-19 | Objectvideo, Inc. | Profile matching of buildings and urban structures |
| CN105654422A (en) * | 2015-12-23 | 2016-06-08 | 北京观著信息技术有限公司 | Point cloud registration method and system |
| CN106097324A (en) * | 2016-06-07 | 2016-11-09 | 中国农业大学 | A kind of non-rigid 3D shape corresponding point determine method |
| CN107492120A (en) * | 2017-07-18 | 2017-12-19 | 北京航空航天大学 | Point cloud registration method |
Cited By (31)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111815686A (en) * | 2019-04-12 | 2020-10-23 | 四川大学 | Coarse-to-fine point cloud registration method based on geometric features |
| CN111815686B (en) * | 2019-04-12 | 2023-08-18 | 四川大学 | Coarse to fine point cloud registration method based on geometric features |
| CN110335297A (en) * | 2019-06-21 | 2019-10-15 | 华中科技大学 | A kind of point cloud registration method based on feature extraction |
| CN110335297B (en) * | 2019-06-21 | 2021-10-08 | 华中科技大学 | A point cloud registration method based on feature extraction |
| CN110287873B (en) * | 2019-06-25 | 2021-06-29 | 清华大学深圳研究生院 | Non-cooperative target pose measurement method and system based on deep neural network and terminal equipment |
| CN110287873A (en) * | 2019-06-25 | 2019-09-27 | 清华大学深圳研究生院 | Noncooperative target pose measuring method, system and terminal device based on deep neural network |
| CN110415281B (en) * | 2019-07-30 | 2022-04-22 | 西安交通大学深圳研究院 | Loam curvature weighting-based point set rigid registration method |
| CN110415281A (en) * | 2019-07-30 | 2019-11-05 | 西安交通大学深圳研究院 | A kind of point set rigid registration method based on Loam curvature weighting |
| CN110473239A (en) * | 2019-08-08 | 2019-11-19 | 刘秀萍 | A kind of high-precision point cloud registration method of 3 D laser scanning |
| CN111462202A (en) * | 2020-04-08 | 2020-07-28 | 中国科学技术大学 | Non-rigid registration method and system |
| CN111462202B (en) * | 2020-04-08 | 2022-09-02 | 中国科学技术大学 | Non-rigid registration method and system |
| CN111612847A (en) * | 2020-04-30 | 2020-09-01 | 重庆见芒信息技术咨询服务有限公司 | Point cloud data matching method and system for robot grabbing operation |
| CN111612847B (en) * | 2020-04-30 | 2023-10-20 | 湖北煌朝智能自动化装备有限公司 | Point cloud data matching method and system for robot grabbing operation |
| CN112015138A (en) * | 2020-09-04 | 2020-12-01 | 中国工程物理研究院机械制造工艺研究所 | Blade contour error evaluation method based on K nearest neighbor iterative nearest grid algorithm |
| CN112015138B (en) * | 2020-09-04 | 2021-08-10 | 中国工程物理研究院机械制造工艺研究所 | Blade contour error evaluation method based on K nearest neighbor iterative nearest grid algorithm |
| CN112435299A (en) * | 2021-01-28 | 2021-03-02 | 深圳大学 | Airborne point cloud assisted satellite image stereo matching point cloud precision orientation method |
| CN113470084B (en) * | 2021-05-18 | 2024-01-30 | 西安电子科技大学 | Point set registration method based on outline rough matching |
| CN113470084A (en) * | 2021-05-18 | 2021-10-01 | 西安电子科技大学 | Point set registration method based on outline rough matching |
| CN115496667A (en) * | 2021-06-17 | 2022-12-20 | 北京四维图新科技股份有限公司 | Point cloud data smoothing method and device, electronic equipment and storage medium |
| CN114004958A (en) * | 2021-11-02 | 2022-02-01 | 广州虎牙科技有限公司 | Texture image processing method, device and electronic device |
| CN114332359A (en) * | 2021-12-10 | 2022-04-12 | 欧冶云商股份有限公司 | System, method and device for three-dimensional reconstruction of warehouse steel point cloud based on laser radar |
| CN114937122A (en) * | 2022-06-16 | 2022-08-23 | 黄冈强源电力设计有限公司 | Rapid three-dimensional model reconstruction method for cement fiberboard house |
| CN114937122B (en) * | 2022-06-16 | 2024-10-18 | 黄冈强源电力设计有限公司 | A method for rapid 3D model reconstruction of cement fiber board house |
| CN115239932A (en) * | 2022-08-02 | 2022-10-25 | 广东奥普特科技股份有限公司 | A three-dimensional data registration method and system based on geometric sampling of curvature features |
| CN115409806A (en) * | 2022-08-31 | 2022-11-29 | 深圳前海瑞集科技有限公司 | Weld joint identification method and device, computer readable medium and electronic equipment |
| CN117132631A (en) * | 2023-09-07 | 2023-11-28 | 重庆理工大学 | Heteroscale point cloud registration method based on pairwise constrained matching of line vectors |
| CN118397209A (en) * | 2024-06-27 | 2024-07-26 | 河北华勘地质勘查有限公司 | Visualization method and system for mine goaf measurement data |
| CN118397209B (en) * | 2024-06-27 | 2024-08-20 | 河北华勘地质勘查有限公司 | Visualization method and system for mine goaf measurement data |
| CN119722767A (en) * | 2025-02-28 | 2025-03-28 | 零磁装备(德清)有限公司 | An automatic co-registration method for brain MRI images and MEG sensors |
| CN120852495A (en) * | 2025-09-23 | 2025-10-28 | 天津博诺智创机器人技术有限公司 | A point cloud registration method based on curvature density feature extraction |
| CN120852495B (en) * | 2025-09-23 | 2025-11-21 | 天津博诺智创机器人技术有限公司 | Point cloud registration method based on curvature density feature extraction |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108376408B (en) | 2020-04-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108376408B (en) | Three-dimensional point cloud data rapid weighting registration method based on curvature features | |
| CN109887015B (en) | A Point Cloud Automatic Registration Method Based on Local Surface Feature Histogram | |
| CN111080684B (en) | A point cloud registration method based on point neighborhood scale difference description | |
| Yao et al. | Point cloud registration algorithm based on curvature feature similarity | |
| CN114677418B (en) | A registration method based on point cloud feature point extraction | |
| CN112017220B (en) | A Point Cloud Accurate Registration Method Based on Robust Constrained Least Squares Algorithm | |
| CN111553409B (en) | A Point Cloud Recognition Method Based on Voxel Shape Descriptors | |
| CN114972459B (en) | Point cloud registration method based on low-dimensional point cloud local feature descriptor | |
| CN114219022B (en) | Multi-sensor multi-target tracking method combining cluster analysis and particle swarm optimization algorithm | |
| CN117745780A (en) | Outdoor large scene 3D point cloud registration method based on isolated cluster removal | |
| CN108594816B (en) | A method and system for positioning and composition by improving ORB-SLAM algorithm | |
| CN105551015A (en) | Scattered-point cloud image registering method | |
| CN117132630B (en) | A point cloud registration method based on second-order spatial compatibility metric | |
| CN109523582B (en) | A point cloud coarse registration method considering normal vectors and multi-scale sparse features | |
| CN119006543A (en) | Point cloud registration method based on neighborhood normal vector and curvature | |
| CN117455967A (en) | A large-scale point cloud registration method based on deep semantic graph matching | |
| CN117274339A (en) | A point cloud registration method based on improved ISS-3DSC features combined with ICP | |
| CN112884057B (en) | Point cloud data-based three-dimensional curved surface quality classification method and system and storage medium | |
| CN108537805A (en) | A kind of target identification method of feature based geometry income | |
| CN117372480A (en) | A point cloud registration method using adaptive downsampling and curvature enhancement point pair features | |
| CN115294398A (en) | SAR image target recognition method based on multi-attitude angle joint learning | |
| CN116883463A (en) | Three-dimensional registration reconstruction method based on multi-domain multi-dimensional feature map | |
| CN114332172B (en) | Laser point cloud registration method based on covariance matrix improvement | |
| CN116416287A (en) | Point cloud registration method, electronic equipment and storage medium based on feature description of key points and neighborhood point pairs | |
| CN118999582B (en) | Path planning method, device, equipment and medium based on particle filter optimization |
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 |