CN116091727A - Complex Qu Miandian cloud registration method based on multi-scale feature description, electronic equipment and storage medium - Google Patents

Complex Qu Miandian cloud registration method based on multi-scale feature description, electronic equipment and storage medium Download PDF

Info

Publication number
CN116091727A
CN116091727A CN202211349965.2A CN202211349965A CN116091727A CN 116091727 A CN116091727 A CN 116091727A CN 202211349965 A CN202211349965 A CN 202211349965A CN 116091727 A CN116091727 A CN 116091727A
Authority
CN
China
Prior art keywords
point
point cloud
feature description
scale
registration
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
Application number
CN202211349965.2A
Other languages
Chinese (zh)
Inventor
张明德
陈星宇
谢乐
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Chongqing University of Technology
Original Assignee
Chongqing University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Chongqing University of Technology filed Critical Chongqing University of Technology
Priority to CN202211349965.2A priority Critical patent/CN116091727A/en
Publication of CN116091727A publication Critical patent/CN116091727A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • G06T17/20—Finite element generation, e.g. wire-frame surface description, tesselation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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
    • G06T7/337—Determination of transform parameters for the alignment of images, i.e. image registration using feature-based methods involving reference images or patches
    • 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
    • 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
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Mathematical Optimization (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Mathematics (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Computing Systems (AREA)
  • Computer Graphics (AREA)
  • Geometry (AREA)
  • Algebra (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Databases & Information Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Image Analysis (AREA)

Abstract

The invention provides a complex Qu Miandian cloud registration method based on multi-scale feature description, electronic equipment and a storage medium, comprising the following steps: step 1, preprocessing point cloud data, which comprises the steps of obtaining point cloud data and simplifying point cloud voxels; step 2, initial registration, which comprises extracting key points, describing key point characteristics, screening corresponding relations and solving a transformation matrix, and rotating and translating object point cloud data to a same coordinate system of theoretical point cloud data by utilizing a rigid transformation matrix to finish the initial registration of the point cloud data; and 3, taking the initial alignment point cloud data as an initial value, calculating a rigid transformation matrix by utilizing an improved ICP algorithm introduced by normal vector weights until the iteration times or objective functions meet the requirements, and finishing point cloud registration. The method can effectively solve the problem that the general registration algorithm is not ideal for the complex Qu Miandian cloud registration effect, and simultaneously shortens the operation time and improves the registration precision.

Description

Complex Qu Miandian cloud registration method based on multi-scale feature description, electronic equipment and storage medium
Technical Field
The invention relates to three-dimensional point cloud data registration, in particular to a point cloud registration method based on multi-scale feature description.
Background
In the three-dimensional scanning imaging technique, data information acquired by a three-dimensional scanner device is generally represented by a large number of discrete points, and a set of these discrete points including three-dimensional coordinates, RGB color values, object reflection surface intensities, and the like is called a point cloud. The three-dimensional point cloud registration is to splice partial three-dimensional point clouds of the real objects acquired from different view angles to obtain a three-dimensional point cloud of the complete real objects under a unified coordinate system. Considering the three-dimensional point cloud of each part as one rigid body, the three-dimensional point cloud registration problem can be attributed to the coordinate transformation problem of the three-dimensional rigid body. The point cloud registration is used as a key step of point cloud data processing, and the registration result directly influences the accuracy of subsequent data processing. Most of the current registration algorithms are converged on the premise that the initial values of registration parameters are known and good. And when the initial value is improper, the algorithm cannot obtain a correct result, and the point cloud rough registration is a key point of the point cloud registration. In recent years, three-dimensional laser scanning devices have been attracting attention, and hardware and software technologies thereof have been continuously developed, and scanning accuracy has been making breakthrough progress. The research and application fields of the technology are mainly focused on reverse process, digital city, unmanned operation, deformation detection, topographic mapping and the like, and a point cloud registration algorithm is a key step, so that whether an original target object is correctly expressed and presented is directly influenced by the registration result, and the establishment precision of a three-dimensional model is further influenced.
Disclosure of Invention
Aiming at the problem that the general registration algorithm is not ideal for the complex Qu Miandian cloud registration effect, the invention provides a complex Qu Miandian cloud registration method based on multi-scale feature description, electronic equipment and a storage medium, which can realize high-precision point cloud data registration and effectively solve the problem of point cloud registration failure caused by initial position and noise points.
The technical scheme of the invention is as follows:
according to one aspect of the present application, there is provided a complex Qu Miandian cloud registration method based on multi-scale feature description, comprising the steps of:
step 1, preprocessing point cloud data, which comprises the following steps:
step 1.1, acquiring point cloud data: acquiring real objects of parts, namely original point cloud data O and theoretical model three-dimensional point cloud data T;
step 1.2, simplifying point cloud voxels: and simplifying the acquired point cloud data by using an improved voxel downsampling algorithm.
Specifically, setting a voxel side length L, and dividing the acquired point cloud data into M grids with the side length L to obtain the voxels of the point cloud data. Calculating a voxel centroid, taking a point closest to the voxel centroid in the voxel as a downsampling point, and realizing point cloud simplification to obtain simplified point clouds O 'and T';
step 2, initial registration, comprising:
step 2.1, extracting key points: extraction of key point sets in O 'and T' using ISS algorithm
Figure BDA0003919315430000021
And->
Figure BDA0003919315430000022
Step 2.2, key point feature description: and carrying out multi-scale feature description on the key point set, and calculating a key point feature description covariance matrix.
Specifically, changing the search radius t carries out neighborhood radius search on the key points, so as to obtain a neighborhood point set of the key points under different scales.
Computing a feature description d= { C over multiple scales r1 …C rs }；
The features of a single scale are described as:
Figure BDA0003919315430000023
p is the key point and r is the search scale radius.
The feature descriptors are:
Figure BDA0003919315430000031
in the method, in the process of the invention,
Figure BDA0003919315430000032
is normal to the key pointQuantity (S)>
Figure BDA0003919315430000033
Vector formed by key point and key point adjacent point, < ->
Figure BDA0003919315430000034
Is the normal vector of the neighboring points of the key point, +.>
Figure BDA0003919315430000035
Forming a vector for the point cloud centroid and the key points; />
Figure BDA0003919315430000036
Vector formed by point cloud centroid and neighboring points of key point +.>
Figure BDA0003919315430000037
Is a principal component of the point cloud.
Calculating a feature description covariance matrix D of a key point set c ＝(M r1 …M rs )；
Figure BDA0003919315430000038
In->
Figure BDA0003919315430000039
The mean is described for the feature.
Step 2.3, screening the corresponding relation: and calculating the multi-scale feature description similarity, searching the initial corresponding relation in a two-way mode, and screening the corresponding relation by using the multi-scale transformation difference.
Specifically, the similarity Sim of the covariance matrix is described by calculating key point characteristics of an original point cloud and a theoretical point cloud, and two points with minimum similarity are selected as initial corresponding relations by using a bidirectional selection mode, wherein a similarity calculation formula is as follows:
Figure BDA00039193154300000310
s is the number of scales.
And eliminating the error corresponding relation by using the conversion difference vector of the corresponding point multi-scale neighborhood coordinate system.
Step 2.4, solving a transformation matrix: and calculating a rigid body transformation matrix according to the corresponding relation, and performing matrix transformation on the down-sampling point cloud to realize initial registration.
Specifically, a rigid transformation matrix of the effective point pairs is solved by singular value decomposition, matrix transformation is carried out on the point cloud O', the point cloud O″ is obtained, and initial registration of the point cloud is realized.
Step 3, accurate registration: an improved ICP algorithm introduced by normal vector weights is used for realizing accurate point cloud registration.
In this step, a correspondence weight assignment is introduced for the ICP registration algorithm, specifically
For the point of the point cloud O '', searching a corresponding point pair by adopting a nearest neighbor search algorithm, adding weight for the corresponding point pair by utilizing a normal vector, wherein a weight calculation formula is as follows:
Figure BDA0003919315430000041
Figure BDA0003919315430000042
and->
Figure BDA0003919315430000043
Is the normal vector of the corresponding point pair.
According to another aspect of the present application, there is provided an electronic device, comprising: a processor; and a memory storing a program, wherein the program comprises instructions that when executed by the processor cause the processor to perform the complex Qu Miandian cloud registration method described above based on the multi-scale feature description.
According to another aspect of the present application, a non-transitory computer-readable storage medium storing computer instructions for causing the computer to perform a complex Qu Miandian cloud registration method based on a multi-scale feature description as described above is presented.
Compared with the prior art, the invention has the following beneficial effects:
1. the feature description combines the main component and the local geometric feature, adopts the feature descriptor fused by the integral feature and the local feature, directly calculates the feature descriptor, and searches the corresponding relation through the normalized feature matrix similarity of the multi-scale feature descriptor.
2. According to the invention, the point cloud downsampling is realized by using the voxel sampling algorithm with the improved gravity center adjacent point, the original structural characteristics of the origin cloud can be saved in the downsampling process, and the operation efficiency is improved.
3. According to the method, the point clouds in the multi-scale range are obtained by using different search radiuses, and the point clouds are initially matched through the similarity of the feature matrix described by the multi-scale features, so that mismatching of the complex Qu Miandian cloud caused by the high similarity of the local geometric features can be effectively avoided.
4. According to the invention, a multi-scale misregistration rejection mode is adopted, local coordinate systems with different scale radiuses are established for initial corresponding points, transformation difference vectors are obtained through rigid body transformation matrixes of the two local coordinate systems, and the transformation difference vectors are screened through standard deviation threshold values to reject error corresponding relations, so that the robustness to noise and the initial registration accuracy can be improved.
5. According to the invention, weight distribution of the basis and normal vector is introduced in the fine registration, so that the situation of local optimum is less involved, and the precision of the fine registration is improved.
Drawings
Fig. 1 is a flow chart of a point cloud registration method based on a multi-scale feature description according to an exemplary embodiment of the present application.
Detailed Description
Embodiments of the present application will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present application are shown in the drawings, it is to be understood that the present application may be embodied in various forms and should not be construed as limited to the embodiments set forth herein, but rather are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and examples of the present application are for illustrative purposes only and are not intended to limit the scope of the present application.
It should be understood that the various steps recited in the method embodiments of the present application may be performed in a different order and/or performed in parallel. Furthermore, method embodiments may include additional steps and/or omit performing the illustrated steps. The scope of the present application is not limited in this respect.
Complex curved surface parts such as aero-engine blades and the like have similar geometrical characteristics, and the registration effect of a common point cloud registration algorithm is not ideal.
Fig. 1 illustrates a point cloud registration method based on multi-scale feature description according to an exemplary embodiment of the present application, firstly, acquiring real point cloud data of a part, and acquiring theoretical point cloud data; then, simplifying the acquired point cloud by utilizing an improved voxel downsampling algorithm; extracting key points of the obtained point cloud by using an ISS algorithm to obtain a point set with obvious space geometric characteristics; then carrying out multi-scale feature description on the obtained key point set by utilizing the space point pair feature pairs to obtain a feature description matrix; then calculating the similarity of the feature matrix, carrying out feature matching, searching for a point pair relation, screening the point pair relation through multi-scale average conversion difference, and calculating a rigid body transformation matrix by utilizing the point pair relation obtained by matching; then, rotating and translating the object point cloud data to the same coordinate system of the theoretical point cloud data by utilizing a rigid body transformation matrix, and completing the initial registration of the point cloud data; and finally, taking the initial alignment point cloud data as an initial value, calculating a rigid transformation matrix by utilizing an improved ICP algorithm introduced by normal vector weights until the iteration times or objective function meet the requirements, and finishing point cloud registration.
According to an exemplary embodiment of the present application, the following details of each step are described:
and step 1, preprocessing point cloud data.
Step 1.1, acquiring point cloud data:
and acquiring three-dimensional point cloud data O of the object to be detected by using monocular structured light. And using NX10 three-dimensional modeling software to derive a theoretical three-dimensional point cloud T from the theoretical three-dimensional model of the real object to be detected.
And 1.2, performing voxel reduction on the obtained point cloud.
And setting the side length L of the voxels, and dividing the acquired point cloud data into M grids with the side length L to obtain the voxels of the point cloud data. And calculating a voxel centroid, taking the point closest to the voxel centroid in the voxel as a downsampling point, and realizing point cloud reduction to obtain reduced point clouds O 'and T'.
And 2, step 2. Initial registration:
step 2.1, extracting the key point set in O 'and T' by using an ISS algorithm
Figure BDA0003919315430000061
And->
Figure BDA0003919315430000062
Step 2.2, carrying out feature description on the key point set:
and changing the search radius r to search the neighborhood radius of the key points, so as to obtain a neighborhood point set of the key points under different scales.
Calculating feature descriptions D= { C under different scales r1 …C rs }。
The features of a single scale are described as:
Figure BDA0003919315430000071
p is the key point and r is the search radius.
The feature descriptors are:
Figure BDA0003919315430000072
in the method, in the process of the invention,
Figure BDA0003919315430000073
is the normal vector of the key point->
Figure BDA0003919315430000074
Vector formed by key point and key point adjacent point, < ->
Figure BDA0003919315430000075
Is the normal vector of the neighboring points of the key point, +.>
Figure BDA0003919315430000076
Forming a vector for the point cloud centroid and the key points; />
Figure BDA0003919315430000077
Vector formed by point cloud centroid and neighboring points of key point +.>
Figure BDA0003919315430000078
Is a principal component of the point cloud.
Calculating a feature description covariance matrix D of a key point set c ＝(M r1 …M rs )；
Figure BDA0003919315430000079
In->
Figure BDA00039193154300000710
The mean is described for the feature.
Step 2.3, calculating the feature description covariance matrix similarity Sim of the original point cloud and the theoretical point cloud, and selecting two points with minimum similarity as an initial corresponding relation by using a bidirectional selection mode, wherein a similarity calculation formula is as follows:
Figure BDA0003919315430000081
establishing a coordinate system by using the neighborhood points of a plurality of scales of each initial corresponding key point and using the neighborhood centroid as an origin
Figure BDA0003919315430000082
And->
Figure BDA0003919315430000083
Obtaining a rotation matrix R of the corresponding neighborhood k And translation vector T k Calculating the quaternion of the rotation matrix to obtain a vector RT formed by the quaternion and the translation quantity k ；
Vector RT using standard deviation threshold k Screening is carried out to obtain effective point pairs, and the error corresponding relation is removed.
And 2.4, solving a rigid transformation matrix of the effective point pairs by adopting singular value decomposition, and performing matrix transformation on the point cloud O ' to obtain a point cloud O ' ', thereby realizing initial registration of the point cloud.
Step 3, fine registering is carried out on the initial registering result:
normal vector weights are introduced for point-to-point relationships in the ICP registration algorithm.
For the point of the point cloud O '', searching a corresponding point pair by adopting a nearest neighbor search algorithm, adding weight for the corresponding point pair by utilizing a normal vector, wherein a weight calculation formula is as follows:
Figure BDA0003919315430000084
Figure BDA0003919315430000085
and->
Figure BDA0003919315430000086
Normal vector of the corresponding point pair;
and performing iterative operation on the point clouds O ' ' and T ' by utilizing an improved ICP algorithm introduced by normal vector weights to obtain a final transformation matrix.
And performing matrix transformation on the original point cloud data O by using a final transformation matrix to realize point cloud fine registration.
The exemplary embodiment of the application also provides an electronic device, including: at least one processor; and a memory communicatively coupled to the at least one processor. The memory stores a computer program executable by the at least one processor for causing the electronic device to perform a method according to an embodiment of the present application when executed by the at least one processor.
The present exemplary embodiments also provide a non-transitory computer readable storage medium storing a computer program, wherein the computer program, when executed by a processor of a computer, is for causing the computer to perform a method according to an embodiment of the present application.
In this application, an electronic device is intended to represent various forms of digital electronic computer devices, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the application described and/or claimed herein.
Program code for carrying out methods of the present application may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this application, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
As used herein, the terms "machine-readable medium" and "computer-readable medium" refer to any computer program product, apparatus, and/or device (e.g., magnetic discs, optical disks, memory, programmable Logic Devices (PLDs)) used to provide machine instructions and/or data to a programmable processor, including a machine-readable medium that receives machine instructions as a machine-readable signal. The term "machine-readable signal" refers to any signal used to provide machine instructions and/or data to a programmable processor.

Claims (9)

1. The complex Qu Miandian cloud registration method based on the multi-scale feature description is characterized by comprising the following steps of:
step 1, preprocessing point cloud data, which comprises the following steps:
step 1.1, acquiring point cloud data: acquiring real object point cloud data O of the part and three-dimensional point cloud data T of the theoretical model;
step 1.2, simplifying point cloud voxels: simplifying the acquired point cloud data by using an improved voxel downsampling algorithm;
step 2, initial registration, comprising:
step 2.1, extracting key points: extraction of key point set of down-sampling point cloud data using ISS algorithm
Figure QLYQS_1
And->
Figure QLYQS_2
Step 2.2, key point feature description: performing multi-scale feature description on the key point set, and calculating a key point feature description covariance matrix;
step 2.3, screening the corresponding relation: calculating the multi-scale feature description similarity, searching for initial corresponding relations in a two-way mode, and screening the corresponding relations by using multi-scale transformation differences;
step 2.4, solving a transformation matrix: calculating a rigid body transformation matrix according to the corresponding relation, and performing matrix transformation on the down-sampling point cloud to realize initial registration;
step 3, accurate registration: an improved ICP algorithm introduced by normal vector weights is used for realizing accurate point cloud registration.
2. The complex Qu Miandian cloud registration method based on multi-scale feature description of claim 1, wherein: the step 1.2 of point cloud voxel reduction specifically comprises the following steps: setting a voxel side length L, and dividing the acquired point cloud data into M grids with the side length L to obtain the voxels of the point cloud data; and calculating a voxel centroid, taking the point closest to the voxel centroid in the voxel as a downsampling point, and realizing point cloud reduction to obtain reduced point clouds O 'and T'.
3. The complex Qu Miandian cloud registration method based on multi-scale feature descriptions of claim 1 or 2, wherein: in the step 2.2, the multi-scale feature description is specifically performed on the key point set, namely, the search radius r is changed to perform neighborhood radius search on the key point, so that a neighborhood point set of the key point under different scales is obtained; computing a feature description d= { C over multiple scales r1 …C rs }；
The features of a single scale are described as:
Figure QLYQS_3
p is the key point and r is the search scale radius.
4. A complex Qu Miandian cloud registration method based on multi-scale feature description as claimed in claim 3, wherein: the feature descriptors in the step 2.2 are as follows:
Figure QLYQS_4
in the method, in the process of the invention,
Figure QLYQS_5
is the normal vector of the key point->
Figure QLYQS_6
Vector formed by key point and key point adjacent point, < ->
Figure QLYQS_7
Is the normal vector of the neighboring points of the key point, +.>
Figure QLYQS_8
Forming a vector for the point cloud centroid and the key points; />
Figure QLYQS_9
Vector formed by point cloud centroid and neighboring points of key point +.>
Figure QLYQS_10
And r is the searching radius, and p is the key point.
5. A complex Qu Miandian cloud registration method based on multi-scale feature description as claimed in claim 3, wherein: in the step 2.3, the calculation formula of the multi-scale feature description similarity is as follows:
Figure QLYQS_11
in the middle of
Figure QLYQS_12
And->
Figure QLYQS_13
Covariance matrix is described for the object point cloud and the theoretical point Yun Di s scale radius neighborhood characteristics, and s is the scale number.
6. A complex Qu Miandian cloud registration method based on multi-scale feature description as claimed in claim 3, wherein: and 2.3, filtering the corresponding relation by using the multi-scale transformation difference, namely, eliminating the error corresponding relation by using a transformation difference matrix of a corresponding point multi-scale field coordinate system.
7. A complex Qu Miandian cloud registration method based on multi-scale feature description as claimed in claim 3, wherein: in the step 3, normal vector constraint is introduced to add weight to corresponding points in the ICP algorithm, and a weight calculation formula is as follows:
Figure QLYQS_14
in the method, in the process of the invention,
Figure QLYQS_15
and->
Figure QLYQS_16
Is the normal vector of the corresponding point pair.
8. An electronic device, comprising: a processor; and a memory storing a program, wherein the program comprises instructions that when executed by the processor cause the processor to perform the complex Qu Miandian cloud registration method based on the multi-scale feature description of any one of claims 1-7.
9. A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are for causing the computer to perform the complex Qu Miandian cloud registration method based on multi-scale feature descriptions according to any one of claims 1-7.
CN202211349965.2A 2022-10-31 2022-10-31 Complex Qu Miandian cloud registration method based on multi-scale feature description, electronic equipment and storage medium Pending CN116091727A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211349965.2A CN116091727A (en) 2022-10-31 2022-10-31 Complex Qu Miandian cloud registration method based on multi-scale feature description, electronic equipment and storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211349965.2A CN116091727A (en) 2022-10-31 2022-10-31 Complex Qu Miandian cloud registration method based on multi-scale feature description, electronic equipment and storage medium

Publications (1)

Publication Number Publication Date
CN116091727A true CN116091727A (en) 2023-05-09

Family

ID=86201302

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211349965.2A Pending CN116091727A (en) 2022-10-31 2022-10-31 Complex Qu Miandian cloud registration method based on multi-scale feature description, electronic equipment and storage medium

Country Status (1)

Country Link
CN (1) CN116091727A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117541631A (en) * 2023-11-07 2024-02-09 四川大学 Blade profile data registration method based on multi-scale features and bidirectional distances
CN118196104A (en) * 2024-05-17 2024-06-14 深圳大学 Intelligent steel member quality detection method and system

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117541631A (en) * 2023-11-07 2024-02-09 四川大学 Blade profile data registration method based on multi-scale features and bidirectional distances
CN118196104A (en) * 2024-05-17 2024-06-14 深圳大学 Intelligent steel member quality detection method and system

Similar Documents

Publication Publication Date Title
JP6216508B2 (en) Method for recognition and pose determination of 3D objects in 3D scenes
JP5705147B2 (en) Representing 3D objects or objects using descriptors
CN111311650B (en) Point cloud data registration method, device and storage medium
CN116091727A (en) Complex Qu Miandian cloud registration method based on multi-scale feature description, electronic equipment and storage medium
WO2017199141A1 (en) Point cloud matching method
CN108022262A (en) A kind of point cloud registration method based on neighborhood of a point center of gravity vector characteristics
US7747106B2 (en) Method and system for filtering, registering, and matching 2.5D normal maps
CN114677418B (en) Registration method based on point cloud feature point extraction
CN110069989B (en) Face image processing method and device and computer readable storage medium
CN111340862B (en) Point cloud registration method and device based on multi-feature fusion and storage medium
CN113628263A (en) Point cloud registration method based on local curvature and neighbor characteristics thereof
CN112651944A (en) 3C component high-precision six-dimensional pose estimation method and system based on CAD model
CN115861397A (en) Point cloud registration method based on improved FPFH-ICP
CN111915517A (en) Global positioning method for RGB-D camera in indoor illumination adverse environment
CN113160129A (en) Combined type simplified point cloud data rapid registration method
CN117132630A (en) Point cloud registration method based on second-order spatial compatibility measurement
CN111739071A (en) Rapid iterative registration method, medium, terminal and device based on initial value
CN111798453A (en) Point cloud registration method and system for unmanned auxiliary positioning
CN112418250B (en) Optimized matching method for complex 3D point cloud
CN113159103B (en) Image matching method, device, electronic equipment and storage medium
CN113706381A (en) Three-dimensional point cloud data splicing method and device
CN115147433A (en) Point cloud registration method
CN115239776B (en) Point cloud registration method, device, equipment and medium
CN113256693A (en) Multi-view registration method based on K-means and normal distribution transformation
Wang et al. Guiding local feature matching with surface curvature

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