CN115409886B - Part geometric feature measuring method, device and system based on point cloud - Google Patents

Part geometric feature measuring method, device and system based on point cloud Download PDF

Info

Publication number
CN115409886B
CN115409886B CN202211360827.4A CN202211360827A CN115409886B CN 115409886 B CN115409886 B CN 115409886B CN 202211360827 A CN202211360827 A CN 202211360827A CN 115409886 B CN115409886 B CN 115409886B
Authority
CN
China
Prior art keywords
point cloud
point
calculating
geometric
training set
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202211360827.4A
Other languages
Chinese (zh)
Other versions
CN115409886A (en
Inventor
汪俊
吴斯帛
李子宽
肖坤
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.)
Nanjing University of Aeronautics and Astronautics
Original Assignee
Nanjing University of Aeronautics and Astronautics
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 Nanjing University of Aeronautics and Astronautics filed Critical Nanjing University of Aeronautics and Astronautics
Priority to CN202211360827.4A priority Critical patent/CN115409886B/en
Publication of CN115409886A publication Critical patent/CN115409886A/en
Application granted granted Critical
Publication of CN115409886B publication Critical patent/CN115409886B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G06V10/443—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components by matching or filtering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/757—Matching configurations of points or features
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
    • G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30108—Industrial image inspection
    • G06T2207/30164—Workpiece; Machine component
    • 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)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Computing Systems (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Geometry (AREA)
  • Length Measuring Devices With Unspecified Measuring Means (AREA)

Abstract

The invention relates to the technical field of digital detection, solves the technical problems of large error, low efficiency and high technical requirement on detection personnel of the traditional contact type part detection method, relates to a part geometric feature measuring method, device and system based on point cloud, in particular to a part geometric feature measuring method based on point cloud, and comprises the following processes: obtaining a training set D of a mechanical part point cloud model marked with elements, wherein the training set D comprises a plurality of single-point data D i (ii) a Generating a decision tree T for element extraction according to the training set D by adopting a C4.5 algorithm; obtaining a point cloud model M of mechanical parts, wherein the point cloud M comprises a plurality of single points i . The invention realizes the intelligent measurement of the precision machining component, improves the geometric precision detection efficiency and accuracy of the precision machining component by organically integrating the three-dimensional laser scanning technology, the measurement error theory, the adjustment technology and the intelligent analysis technology, and has better practicability.

Description

Part geometric feature measuring method, device and system based on point cloud
Technical Field
The invention relates to the technical field of digital detection, in particular to a method, a device and a system for measuring geometrical characteristics of a part based on point cloud.
Background
The quality of parts in mechanical products directly influences the accuracy and stability of equipment, the traditional contact type part detection method has large error, low efficiency and high technical requirements on detection personnel, different measurement schemes need to be formulated for different measurement targets, and the measurement scheme for curved surfaces is more complex. The measuring tool with higher precision has strict requirements on the measuring environment and extremely high measuring cost. In addition, a large number of staff in the front line are required to perform complicated repeated operation in the part production process, special professional talents are required for some special posts, and the detection efficiency is urgently required to be improved.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a method, a device and a system for measuring the geometric characteristics of parts based on point cloud, which solve the technical problems of large error, low efficiency and high technical requirement on detection personnel in the traditional contact type part detection method, realize the intelligent measurement of a precisely machined component, improve the geometric precision detection efficiency and accuracy of the precisely machined component by organically combining a three-dimensional laser scanning technology, a measurement error theory, a balancing technology and an intelligent analysis technology, and have better practicability.
In order to solve the technical problems, the invention provides the following technical scheme: a part geometric feature measuring method based on point cloud comprises the following steps:
obtaining a training set D of a mechanical part point cloud model marked with elements, wherein the training set D comprises a plurality of single-point data D i ；
Generating a decision tree T for element extraction according to the training set D by adopting a C4.5 algorithm;
obtaining a point cloud model M of mechanical parts, wherein the point cloud M comprises a plurality of single points i ；
Point cloud M of each single point in point cloud model M i Performing near point query and calculating point cloud M i The local features of (a) are recorded as a set N;
inputting the set N into a decision tree T for element extraction to obtain a point cloud group P after the element extraction;
segmenting the point cloud group P by adopting a region growing algorithm to obtain a segmented point cloud group Q;
calculating, identifying and grouping the geometrical characteristics of the target surface of the point cloud group Q to obtain measurement data of the geometrical characteristics of the part;
and judging whether the design standard of the part is met or not according to the measurement data of the geometric characteristics of the part.
Further, in the process of generating the decision tree T for primitive extraction according to the training set D by using the C4.5 algorithm, the following steps are also included:
s201, aiming at each single point data D in the training set D i Using a kd-tree to perform near point query;
s202, calculating each single point data D i Gaussian curvature K;
s203, calculating single-point data D i PFH descriptor F of (a);
s204, calculating the empirical entropy of the training set D
Figure 755794DEST_PATH_IMAGE001
；
S205, recording the characteristics obtained in the steps S202-S203 as characteristics A, wherein the characteristics A have 2 different values of Gaussian curvature K and PFH descriptor F, and the characteristics A are respectively recorded as characteristics A
Figure 663707DEST_PATH_IMAGE002
That is, the training set D is divided into 2 subsets according to the value of A
Figure 471126DEST_PATH_IMAGE003
，
Figure 173503DEST_PATH_IMAGE004
Calculating the empirical conditional entropy of the features A on the training set D
Figure 66372DEST_PATH_IMAGE005
；
S206, entropy based on empirical conditions
Figure 512397DEST_PATH_IMAGE005
Calculating information gain
Figure 174323DEST_PATH_IMAGE006
；
S207, gain according to information
Figure 47601DEST_PATH_IMAGE006
Calculating an information gain ratio
Figure 663652DEST_PATH_IMAGE007
；
S208, selecting the information gain ratio
Figure 913368DEST_PATH_IMAGE007
Characteristic of medium maximum
Figure 695379DEST_PATH_IMAGE008
To characteristic(s)
Figure 536296DEST_PATH_IMAGE008
Each possible value of
Figure 341441DEST_PATH_IMAGE009
In accordance with
Figure 191585DEST_PATH_IMAGE010
Dividing the training set D into a plurality of non-empty single-point data D i Taking the class with the maximum number of examples as a mark to construct a child node, and constructing a decision tree T by the node i and the child node thereof;
s209, aiming at the node i, using single-point data D i For the training set, the
Figure 765786DEST_PATH_IMAGE011
For the feature set, the steps S204-S208 are recursively called to loop until all instances in the training set D belong to the same class C k A decision tree T is obtained.
Further, the point cloud M of each single point in the point cloud model M i Performing near point query and calculating point cloud M i In the process of recording the local features as a set N, the detailed process includes:
and (4) carrying out near point query on the point cloud Mi of each single point of the point cloud model M, calculating the point Gaussian curvature K and the PFH descriptor F based on the near points, and recording as a set N.
Further, in the process of inputting the set N into the decision tree T for primitive extraction to obtain the point cloud group P after primitive extraction, the detailed process includes:
inputting the set N into a decision tree T to obtain a point cloud M of each single point in the set N i The sample classification result is a primitive segmentation result, and the segmented point cloud group P is output.
Further, in the process of segmenting the point cloud group P by using a region growing algorithm to obtain a segmented point cloud group Q, the method further comprises the following steps:
s601, randomly selecting a point P in the point cloud group P as a seed point, and setting the curvature of the seed point P as
Figure 246446DEST_PATH_IMAGE012
；
S602, performing proximity point search on the seed point p to obtain a proximity point p i Let a near point p i Has a curvature of
Figure 335625DEST_PATH_IMAGE013
Setting a curvature change threshold value of
Figure 927143DEST_PATH_IMAGE014
；
If it is
Figure 418167DEST_PATH_IMAGE015
Then, it will be close to the point p i Polymerizing with the seed point p and adding the proximity point p i And (5) serving as new seed points until all the points are clustered to obtain a segmented point cloud group Q.
Further, in the process of calculating, identifying and grouping the geometric features of the target surface of the point cloud group Q to obtain the measurement data of the geometric features of the part, the method further comprises the following steps:
s701, using RANSAC to align the cylindrical surfaces
Figure 600887DEST_PATH_IMAGE016
Fitting to obtain the center line of the fitted cylindrical surface
Figure 380624DEST_PATH_IMAGE017
According to the cylindrical surface
Figure 572571DEST_PATH_IMAGE016
Height of (2)
Figure 855785DEST_PATH_IMAGE018
Radius, of the shaft
Figure 242029DEST_PATH_IMAGE019
Calculating a centerline
Figure 509062DEST_PATH_IMAGE017
The distance between the center line of the other cylindrical surfaces is selected as the nearest distance and the farthest distance
Figure 442383DEST_PATH_IMAGE020
And
Figure 642420DEST_PATH_IMAGE021
building up a cylindrical surface
Figure 635784DEST_PATH_IMAGE016
The feature operator of (2);
s702, using RANSAC to plane
Figure 186851DEST_PATH_IMAGE022
Fitting to obtain a fitted plane, and determining the position of the center point of the fitted plane
Figure 455022DEST_PATH_IMAGE023
Plane normal vector
Figure 978407DEST_PATH_IMAGE024
Computing
Figure 939410DEST_PATH_IMAGE025
Area of
Figure 649877DEST_PATH_IMAGE026
Building a plane
Figure 987317DEST_PATH_IMAGE025
The feature operator of (2);
s703, constructing each cylindrical surface in part standard parts
Figure 365209DEST_PATH_IMAGE027
According to the characteristic operator of each cylinder
Figure 700375DEST_PATH_IMAGE027
The feature operator of (1) completes the hash table
Figure 226034DEST_PATH_IMAGE028
Creating;
s704, constructing each plane in the part standard component
Figure 39270DEST_PATH_IMAGE029
And according to each plane
Figure 333985DEST_PATH_IMAGE029
Characteristic operator completion hash table
Figure 574473DEST_PATH_IMAGE030
Creating;
s705, characteristic operator
Figure 823314DEST_PATH_IMAGE031
And a hash table
Figure 502557DEST_PATH_IMAGE028
Matching to obtain feature pairs
Figure 589462DEST_PATH_IMAGE032
The identification of the geometric characteristic cylindrical surface of the part is realized;
s706, operating the characteristic operator
Figure 532010DEST_PATH_IMAGE033
And a hash table
Figure 766682DEST_PATH_IMAGE030
Matching to obtain feature pairs
Figure 921720DEST_PATH_IMAGE034
Realizing the identification of the geometric feature plane of the part;
s707, obtaining a geometric dimension to be checked according to a part design standard;
s708, calculating the distance between the two planes of the mechanical part to be inspected respectively
Figure 191028DEST_PATH_IMAGE035
Distance between them
Figure 773319DEST_PATH_IMAGE036
；
S709, calculating the axial distance between two cylindrical surfaces respectively for the distance between two cylindrical axes to be detected
Figure 495287DEST_PATH_IMAGE037
。
The technical scheme also provides a device suitable for the method for measuring the geometric characteristics of the part, and the device comprises the following components:
a first acquisition module, which is used for acquiring a training set D of the point cloud model of the mechanical part with marked elements, wherein the training set D comprises a plurality of single-point data D i ；
The decision tree T generation module is used for generating a decision tree T for element extraction according to the training set D by adopting a C4.5 algorithm;
a second obtaining module, configured to obtain a point cloud model M of the mechanical component, where the point cloud M includes a plurality of single points i ；
A local feature calculation module for point cloud M for each single point in the point cloud model M i Performing near point query and calculating point cloud M i The local features of (a) are recorded as a set N;
the extracting module is used for inputting the set N into the decision tree T for element extraction to obtain a point cloud group P after the element extraction;
the segmentation module is used for segmenting the point cloud group P by adopting a region growing algorithm to obtain a segmented point cloud group Q;
the measurement data obtaining module is used for calculating, identifying and grouping the geometric features of the target surface of the point cloud group Q to obtain measurement data of the geometric features of the parts;
and the judging module is used for judging whether the design standard of the part is met according to the measurement data of the geometric characteristics of the part.
The technical scheme also provides a system suitable for the method for measuring the geometric characteristics of the part, and the system comprises the following steps:
the point cloud model forming unit is used for obtaining point cloud data of the precision machining component to form a point cloud model M;
the point cloud model segmentation unit is used for carrying out element classification and segmentation on a point cloud model M of the precision machining component, realizing element classification by adopting a decision tree C4.5 method, realizing element segmentation by adopting a region growing algorithm and obtaining a segmentation result point cloud group Q;
a geometric feature identification unit for matching each of the segmented regions of the segmentation result point cloud group Q with a design structure;
and the geometric precision detection unit is used for calculating a numerical value of the geometric feature precision of the required precision machining component on the point cloud, comparing the numerical value with a design value and finally outputting a quality inspection report.
By means of the technical scheme, the invention provides a part geometric feature measuring method, device and system based on point cloud, which at least have the following beneficial effects:
1. the invention solves the technical problems of large error, low efficiency and high technical requirement on detection personnel in the traditional contact type part detection method, realizes the intelligent measurement of the precision machining component, improves the geometric precision detection efficiency and accuracy of the precision machining component by organically integrating the three-dimensional laser scanning technology, the measurement error theory, the adjustment technology and the intelligent analysis technology, and has better practicability.
2. According to the invention, the point cloud data is processed and analyzed by using the point cloud model of the precision machining component, the geometric characteristics of the part are measured, the intelligent measurement of the precision machining component is realized, the geometric precision detection is carried out by processing and analyzing the point cloud data, the detection efficiency and the accuracy are improved, the inspection efficiency and the delivery progress of the product are improved, and the method has better practicability.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
FIG. 1 is a flow chart of a part geometry measurement method of the present invention;
FIG. 2 is a schematic view of a point cloud model of a transverse shock absorber mount of the present invention;
FIG. 3 is a schematic view of a point cloud model segmentation of the transverse damper seat of the present invention;
FIG. 4 is a plan distance map of a point cloud model of the transverse damper seat of the present invention;
FIG. 5 is a block diagram of the device for measuring geometrical characteristics of a part according to the present invention;
FIG. 6 is a block diagram of a system for measuring geometrical characteristics of a part according to the present invention.
In the figure: 10. a first acquisition module; 20. a decision tree T generation module; 30. a second acquisition module; 40. a local feature calculation module; 50. an extraction module; 60. a segmentation module; 70. a measured data obtaining module; 80. a judgment module; 100. a point cloud model construction unit; 200. a point cloud model segmentation unit; 300. a geometric feature recognition unit; 400. and a geometric precision detection unit.
Detailed Description
In order to make the aforementioned objects, features and advantages of the present invention more comprehensible, the present invention is described in detail with reference to the accompanying drawings and the detailed description thereof. Therefore, the realization process of how to apply technical means to solve the technical problems and achieve the technical effects can be fully understood and implemented.
Referring to fig. 1 to fig. 6, a specific embodiment of the present embodiment is shown, in which a point cloud model of a precisely machined component is utilized to process and analyze point cloud data to measure geometric features of a part, so as to improve inspection efficiency and delivery progress of a product, provide better practicability, and improve detection efficiency and accuracy of geometric feature precision of the part.
Referring to fig. 1, a method for measuring geometric characteristics of a part based on point cloud is described in detail in this embodiment through steps S1 to S8, and includes the following steps:
s1, obtaining a training set D of a mechanical part point cloud model with marked elements, wherein the training set D comprises a plurality of single-point data
Figure 454016DEST_PATH_IMAGE038
。
Specifically, a laser scanner is used for collecting information of a mechanical part, high-precision point cloud data with a complete surface are generated, a point cloud model of the part is formed under the point cloud data, a label of an element is added to each point, namely a training set D of the point cloud model of the mechanical part with the element marked is obtained, wherein the training set D comprises a plurality of single-point data
Figure 577830DEST_PATH_IMAGE039
，
Figure 393339DEST_PATH_IMAGE039
The parameters contained in (1) are
Figure 540286DEST_PATH_IMAGE040
Wherein, in the step (A),
Figure 99444DEST_PATH_IMAGE041
as a geometric coordinate of the point,
Figure 15447DEST_PATH_IMAGE042
the primitive label to which the point belongs.
And S2, generating a decision tree T for primitive extraction according to the training set D by adopting a C4.5 algorithm.
In step S2, the specific process of generating the decision tree T for primitive extraction from the training set D by using the C4.5 algorithm includes the following steps:
s201, for each single point data D in the training set D i Using a kd-tree to perform near point query;
s202, calculating each single point data D i The process of calculating the gaussian curvature K is specifically as follows:
with a single point of data D i Taking n points uniformly in the point cloud for the center, and fitting a quadratic surface by least square method through the n points
Figure 768902DEST_PATH_IMAGE043
Calculating single-point data D according to the properties of the space curved surface after solving the coefficient i The curvature of (d). The following equation is minimized according to the least squares principle:
Figure 403145DEST_PATH_IMAGE044
Figure 765994DEST_PATH_IMAGE045
in the above-mentioned formula, the compound has the following structure,
Figure 536503DEST_PATH_IMAGE046
is a point in the field, the coefficients are differentiated by the above formula to be 0, to obtain:
Figure 959395DEST_PATH_IMAGE047
thereby solving for the conic coefficients;
if a curve r exists on the curved surface, the expression of the curve r is as follows:
Figure 80934DEST_PATH_IMAGE048
if s represents the arc length of the curve r, the differential equation of the arc length can be obtained by the complex function derivation equation:
Figure 981894DEST_PATH_IMAGE049
is provided with
Figure 872490DEST_PATH_IMAGE050
And then:
the unit normal vector n of the curve is expressed as:
Figure 466282DEST_PATH_IMAGE051
from the second basic formula of the curved surface:
Figure 809539DEST_PATH_IMAGE052
in the formula:
Figure 779769DEST_PATH_IMAGE053
the normal curvature can be expressed as:
Figure 524871DEST_PATH_IMAGE054
process variables
Figure 289565DEST_PATH_IMAGE055
Can obtain k n Two radicles k of 1 ，k 2 From the nature of gaussian curvature, we can derive:
Figure 120117DEST_PATH_IMAGE056
s203, calculating single-point data D i The procedure of calculation of PFH descriptor F of (1) is as follows:
at a single point of data D i The coordinate system is defined as follows:
Figure 362880DEST_PATH_IMAGE057
at this time, the process of the present invention,
Figure 228068DEST_PATH_IMAGE058
is a point
Figure 665128DEST_PATH_IMAGE059
In the formula
Figure 982976DEST_PATH_IMAGE060
Is a point
Figure 29430DEST_PATH_IMAGE061
The normal vector of (a) is,
Figure 749124DEST_PATH_IMAGE062
is a point
Figure 855621DEST_PATH_IMAGE061
The neighborhood points of (c), then:
Figure 660765DEST_PATH_IMAGE063
by using
Figure 510910DEST_PATH_IMAGE064
Constituting PFH descriptor F.
S204, calculating the experience entropy of the training set D
Figure 85111DEST_PATH_IMAGE065
The calculation formula is as follows:
Figure 362508DEST_PATH_IMAGE066
in the above formula, the first and second carbon atoms are,
Figure 389370DEST_PATH_IMAGE067
for the number of samples, K classes are set
Figure 43205DEST_PATH_IMAGE068
The elements of a precision-machined component are divided into planar and cylindrical surfaces according to their structure, i.e. here
Figure 471913DEST_PATH_IMAGE069
To belong to the class
Figure 920211DEST_PATH_IMAGE070
The number of samples of (a) to (b),
Figure 699949DEST_PATH_IMAGE071
；
s205, recording the characteristics obtained in the steps S202-S203 as characteristics A, wherein the characteristics A have 2 different values of Gaussian curvature K and PFH descriptor F, and the characteristics A are respectively recorded as characteristics A
Figure 891896DEST_PATH_IMAGE072
That is, the training set D is divided into 2 subsets according to the value of A
Figure 175109DEST_PATH_IMAGE073
Calculating the empirical conditional entropy of the features A on the training set D
Figure 561354DEST_PATH_IMAGE074
The calculation formula is as follows:
Figure 562808DEST_PATH_IMAGE075
in the above-mentioned formula, the compound has the following structure,
Figure 292866DEST_PATH_IMAGE076
as a single point of data
Figure 696166DEST_PATH_IMAGE077
The number of samples of (a) to (b),
Figure 751846DEST_PATH_IMAGE078
n is the number of the characteristic A, namely n =2, and single point data is recorded
Figure 975017DEST_PATH_IMAGE077
In the middle of
Figure 774346DEST_PATH_IMAGE079
Is a set of samples of
Figure 32152DEST_PATH_IMAGE080
I.e. by
Figure 993155DEST_PATH_IMAGE081
Is composed of
Figure 969201DEST_PATH_IMAGE080
The number of samples of (a);
s206, entropy according to empirical condition
Figure 306642DEST_PATH_IMAGE082
Calculating information gain
Figure 418954DEST_PATH_IMAGE083
The calculation formula is as follows:
Figure 816437DEST_PATH_IMAGE084
in the above formula, the first and second carbon atoms are,
Figure 14201DEST_PATH_IMAGE085
in order to be the empirical entropy, the entropy is,
Figure 155332DEST_PATH_IMAGE082
the empirical conditional entropy of feature a on training set D.
S207, increasing according to the informationBenefit to
Figure 387730DEST_PATH_IMAGE083
Calculating an information gain ratio
Figure 203719DEST_PATH_IMAGE086
The calculation formula is as follows:
Figure 154357DEST_PATH_IMAGE087
wherein:
Figure 833601DEST_PATH_IMAGE088
s208, selecting the information gain ratio
Figure 920505DEST_PATH_IMAGE086
Central maximum feature
Figure 394212DEST_PATH_IMAGE089
To characteristics of
Figure 566567DEST_PATH_IMAGE089
Each possible value of (a)
Figure 49501DEST_PATH_IMAGE090
An
Figure 990912DEST_PATH_IMAGE091
Segmenting a training set D into a plurality of non-empty single-point data D i Taking the class with the maximum number of the examples as a mark to construct a child node, and forming a decision tree T by the node i and the child node thereof;
s209, aiming at the node i, using single-point data D i For the training set, the
Figure 901100DEST_PATH_IMAGE092
For the feature set, the steps S204-S208 are recursively called to loop until all instances in the training set D belong to the same class
Figure 560751DEST_PATH_IMAGE093
A decision tree T is obtained.
S3, obtaining a point cloud model M of the mechanical part, wherein the point cloud model M comprises a plurality of single points
Figure 581797DEST_PATH_IMAGE094
。
Collecting information of mechanical parts to be detected by using a laser scanner, generating high-precision point cloud data with complete surface, and forming a point cloud model of the parts under the point cloud data, namely acquiring a point cloud model M of the mechanical parts, wherein the point cloud model M comprises a plurality of point clouds of single points
Figure 643294DEST_PATH_IMAGE094
Point cloud of single points
Figure 724382DEST_PATH_IMAGE094
The parameters contained in (1) are
Figure 605750DEST_PATH_IMAGE095
Wherein, in the process,
Figure 430487DEST_PATH_IMAGE096
point cloud for each single point
Figure 346490DEST_PATH_IMAGE094
The coordinates of (a).
Referring to fig. 2, the present embodiment takes a transverse shock absorber bearing as an example, and fig. 2 shows a schematic diagram of a point cloud model of the transverse shock absorber bearing;
s4, point cloud of each single point in the point cloud model M
Figure 99945DEST_PATH_IMAGE094
Performing near point query and calculating point cloud
Figure 734189DEST_PATH_IMAGE094
Is denoted as set N, the specific steps include: each list of point cloud model MThe point cloud Mi of the point is queried about the nearby point, the point Gaussian curvature K and the PFH descriptor F are calculated based on the nearby point and are recorded as a set N, the specific calculation process for calculating the point Gaussian curvature K and the PFH descriptor F is the same as the steps S202-S203, the steps disclose related calculation formulas in detail, and in order to avoid repeated description, redundant description is not repeated here, and the calculation result is recorded as the set N;
s5, inputting the set N into a decision tree T for element extraction to obtain a point cloud group P after element extraction;
the detailed process of obtaining the point cloud group P after primitive extraction in step S5 is as follows: inputting the set N into a decision tree T to obtain the point cloud of each single point in the set N
Figure 97037DEST_PATH_IMAGE094
The sample classification result is a primitive segmentation result, and a segmented point cloud group P is output.
And S6, segmenting the point cloud group P by adopting a region growing algorithm to obtain a segmented point cloud group Q.
Referring to fig. 3, the present embodiment takes the transverse damper seat as an example, and fig. 3 shows several point cloud groups that are presented after the point cloud model is segmented after primitive extraction, which correspond to the point cloud group Q in the present embodiment.
In step S6, the specific steps of segmenting the point cloud group P by using a region growing algorithm to obtain a segmented point cloud group Q are as follows:
s601, randomly selecting a point P in the point cloud group P as a seed point, and setting the curvature of the seed point P as
Figure 867547DEST_PATH_IMAGE097
；
S602, carrying out near point search on the seed point p to obtain a near point p i Let a near point p i Has a curvature of
Figure 290438DEST_PATH_IMAGE098
Setting a curvature change threshold value of
Figure 146398DEST_PATH_IMAGE099
；
If it is
Figure 312937DEST_PATH_IMAGE100
Then will be close to point p i Polymerizing with the seed point p and adding the proximity point p i As new seed points, until all the points are clustered to obtain a segmented point cloud group Q, in the process, the cylindrical surface is marked as
Figure 203533DEST_PATH_IMAGE101
The number of the cylindrical surfaces, the planes
Figure 797325DEST_PATH_IMAGE102
b is the number of planes.
S7, calculating, identifying and grouping the geometrical characteristics of the target surface of the point cloud group Q to obtain measurement data of the geometrical characteristics of the part;
in step S7, the step of calculating, identifying and grouping the geometric features of the target surface for the point cloud group Q to obtain the measurement data of the geometric features of the part specifically includes the following steps:
s701, using RANSAC to align the cylindrical surfaces
Figure 140582DEST_PATH_IMAGE103
Fitting to obtain the center line of the fitted cylindrical surface
Figure 845233DEST_PATH_IMAGE104
Center line of
Figure 590335DEST_PATH_IMAGE104
The linear equation of (c) is:
Figure 355029DEST_PATH_IMAGE105
according to the cylindrical surface
Figure 185582DEST_PATH_IMAGE103
Height of (2)
Figure 693923DEST_PATH_IMAGE106
Radius, radius
Figure 559111DEST_PATH_IMAGE107
Calculating a centerline
Figure 730592DEST_PATH_IMAGE104
The distance between the center line of the cylindrical surface and the center line of other cylindrical surfaces is selected as the nearest distance and the farthest distance
Figure 48441DEST_PATH_IMAGE108
And
Figure 360473DEST_PATH_IMAGE109
building up a cylindrical surface
Figure 80167DEST_PATH_IMAGE103
Characteristic operator of
Figure 186664DEST_PATH_IMAGE110
；
S702, using RANSAC to plane
Figure 726229DEST_PATH_IMAGE111
Fitting is carried out, and the fitted plane equation is obtained as follows:
Figure 576374DEST_PATH_IMAGE112
according to the position of the center point of the fitted plane
Figure 150575DEST_PATH_IMAGE113
Normal vector of plane
Figure 427972DEST_PATH_IMAGE114
Computing
Figure 720413DEST_PATH_IMAGE115
Area of
Figure 108669DEST_PATH_IMAGE116
Building a plane
Figure 599693DEST_PATH_IMAGE115
Characteristic operator of
Figure 251255DEST_PATH_IMAGE117
。
S703, constructing each cylindrical surface in part standard parts
Figure 562150DEST_PATH_IMAGE118
Characteristic operator of
Figure 691780DEST_PATH_IMAGE119
The specific construction steps are the same as S701, and each cylindrical surface is divided into two parts
Figure 804355DEST_PATH_IMAGE118
Characteristic operator of
Figure 361238DEST_PATH_IMAGE120
Pushing into conventional hash table, extracting key words by using digital analysis method
Figure 425009DEST_PATH_IMAGE121
The first two digits of (c) and
Figure 358330DEST_PATH_IMAGE122
the first two digits of
Figure 823947DEST_PATH_IMAGE118
The hash address of, complete hash table
Figure 551731DEST_PATH_IMAGE123
Is created.
It should be noted that the component standard component is a known mechanical component, and each dimension and index of the component standard component meet the design requirement, each dimension and index of the component are used as known data, and since each component is different from another component, the related parameters are not unique, so that redundant description is not repeated in this embodiment, the related parameters can be determined according to the actual mechanical component before measurement, and in the scheme, the geometric characteristics of the mechanical components produced in batches are measured and compared with the related parameters of the standard mechanical component, so as to measure whether the produced mechanical components meet the standard requirement, thereby improving the detection efficiency and accuracy, and further improving the inspection efficiency and the delivery schedule of the product.
S704, constructing each plane in the part standard component
Figure 102798DEST_PATH_IMAGE124
Characteristic operator of
Figure 839810DEST_PATH_IMAGE125
The specific construction steps are the same as S702, each characteristic operator is pushed into a conventional hash table, and a numerical analysis method is used for extracting keywords
Figure 159933DEST_PATH_IMAGE126
The first two digits of
Figure 58619DEST_PATH_IMAGE127
The first two digits of
Figure 96982DEST_PATH_IMAGE128
The hash address of, complete hash table
Figure 372106DEST_PATH_IMAGE129
Is created.
S705, characteristic operator
Figure 546735DEST_PATH_IMAGE130
And a hash table
Figure 881901DEST_PATH_IMAGE123
Matching to obtain feature pairs
Figure 141981DEST_PATH_IMAGE131
The identification of the geometric characteristic cylindrical surface of the part is realized;
s706, performing feature operator
Figure 220796DEST_PATH_IMAGE132
And a hash table
Figure 751397DEST_PATH_IMAGE129
Matching to obtain feature pairs
Figure 257464DEST_PATH_IMAGE133
And realizing the identification of the geometric feature plane of the part.
S707, obtaining the geometric dimension to be checked according to the design standard of the part
Figure 4840DEST_PATH_IMAGE134
Wherein
Figure 621767DEST_PATH_IMAGE135
Represents the distance between the axis of the cylindrical surface i and the axis of the cylindrical surface j to be tested, where x represents the nominal dimension, r1 represents the upper deviation, r2 represents the lower deviation,
Figure 770988DEST_PATH_IMAGE136
the distance between the two planes m, n to be detected is indicated, where y denotes the nominal dimension, e1 denotes the upper deviation and e2 denotes the lower deviation.
The part design standard is known data and has the same meaning as a part standard, and the part design standard is labeled in detail in a corresponding part design drawing, illustratively, according to the design standard of a certain part, the geometric dimension to be checked is obtained, such as the distance between the axes of the two top surface circular holes is 90mm, and the distance between the two side surfaces of the top surface is 60 mm.
S708, calculating the distance between the two planes of the mechanical part to be inspected respectively
Figure 447957DEST_PATH_IMAGE137
Distance between them
Figure 682629DEST_PATH_IMAGE138
Referring to FIG. 4, the present embodiment takes the transverse damper seat as an example, and shows the actual state of the plane distance measurement in combination with the actual stateTwo planes in the example step
Figure 103246DEST_PATH_IMAGE139
Distance between them
Figure 106975DEST_PATH_IMAGE138
The calculation is carried out according to the following formula:
Figure 954845DEST_PATH_IMAGE140
wherein the content of the first and second substances,
Figure 676813DEST_PATH_IMAGE141
is a plane
Figure 635542DEST_PATH_IMAGE142
At any point on the upper surface
Figure 759356DEST_PATH_IMAGE143
The equation is
Figure 512548DEST_PATH_IMAGE144
The equation is obtained by fitting in step S702, and a, B, and C are equation coefficients, respectively.
S709, calculating the distance between the axes of the two cylinders to be tested
Figure 721813DEST_PATH_IMAGE145
The calculation formula is as follows:
Figure 484232DEST_PATH_IMAGE146
wherein the two-line equations are respectively
Figure 964017DEST_PATH_IMAGE147
Here, the equation is fitted in S701,
Figure 153690DEST_PATH_IMAGE148
are all equation coefficients.
And S8, judging whether the design standard of the part is met according to the measurement data of the geometric characteristics of the part.
Specifically, the judgment logic is as follows: comparison
Figure 584672DEST_PATH_IMAGE149
And
Figure 681941DEST_PATH_IMAGE150
，
Figure 718030DEST_PATH_IMAGE151
and
Figure 140921DEST_PATH_IMAGE152
if, if
Figure 996881DEST_PATH_IMAGE153
Or
Figure 163420DEST_PATH_IMAGE154
Then, the first step is executed,
Figure 54016DEST_PATH_IMAGE155
does not meet the requirements, if
Figure 647808DEST_PATH_IMAGE156
Then, the first step is executed,
Figure 991065DEST_PATH_IMAGE155
the requirements are met; if it is
Figure 695716DEST_PATH_IMAGE157
Or
Figure 440818DEST_PATH_IMAGE158
Then, the first step is executed,
Figure 205512DEST_PATH_IMAGE159
does not meet the requirements, if
Figure 36064DEST_PATH_IMAGE160
Then, if the number of the first time zone is less than the first threshold value,
Figure 45871DEST_PATH_IMAGE159
the requirements are met.
According to the measuring method provided by the embodiment, the point cloud data is processed and analyzed by utilizing the point cloud model of the precision machining component, the geometric characteristics of the part are measured, the intelligent measurement of the precision machining component is realized, the inspection efficiency and the delivery progress of the product are improved, the practicability is better, and the detection efficiency and the accuracy of the geometric precision are improved.
The embodiment of the geometric feature measuring method of the part is also applicable to the geometric feature measuring device of the part provided by the embodiment, and the detailed description is omitted in the embodiment since the geometric feature measuring device of the part provided by the embodiment corresponds to the geometric feature measuring method of the part provided by the embodiment.
Referring to fig. 5, a block diagram of a geometric feature measuring device provided in this embodiment is shown, where the geometric feature measuring device includes:
an apparatus for implementing the method for measuring the geometric characteristics of the part comprises:
a first obtaining module 10, wherein the first obtaining module 10 is used for obtaining a training set D of a point cloud model of a mechanical part with marked elements, and the training set D comprises a plurality of single-point data D i ；
A decision tree T generating module 20, wherein the decision tree T generating module 20 is used for generating a decision tree T for element extraction according to the training set D by adopting a C4.5 algorithm;
a second obtaining module 30, where the second obtaining module 30 is configured to obtain a point cloud model M of the mechanical component, where the point cloud model M includes a plurality of single points i ；
A local feature calculation module 40, wherein the local feature calculation module 40 is used for point cloud M of each single point in the point cloud model M i Performing a near point query, anCalculating point cloud M i The local features of (a) are recorded as a set N;
the extraction module 50 is used for inputting the set N into the decision tree T for element extraction to obtain a point cloud group P after element extraction;
the segmentation module 60 is configured to segment the point cloud group P by using a region growing algorithm to obtain a segmented point cloud group Q;
the measurement data obtaining module 70 is used for calculating, identifying and grouping the geometric features of the target surface of the point cloud group Q to obtain measurement data of the geometric features of the parts;
and the judging module 80, wherein the judging module 80 is used for judging whether the design standard of the part is met according to the measurement data of the geometric characteristics of the part.
It should be noted that, when the apparatus provided in the foregoing embodiment implements the functions thereof, only the division of the functional modules is illustrated, and in practical applications, the functions may be distributed by different functional modules according to needs, that is, the internal structure of the apparatus may be divided into different functional modules to implement all or part of the functions described above. In addition, the apparatus and method embodiments provided by the above embodiments belong to the same concept, and specific implementation processes thereof are described in the method embodiments for details, which are not described herein again.
The part geometric characteristics measuring device of this embodiment is through the point cloud model that utilizes the precision finishing component, and the analysis is handled to the counter point cloud data, carries out the part geometric characteristics and measures, has realized carrying out the intelligent measurement to the precision finishing component, handles the analysis through the counter point cloud data and carries out the geometric accuracy and detect, improves detection efficiency and rate of accuracy to promote the efficiency of inspection and the delivery progress of product, have better practicality, improve detection efficiency and the rate of accuracy to the geometric accuracy.
Referring to fig. 6, a block diagram of a geometric feature measurement system of a part according to the present embodiment is shown, where the geometric feature measurement system includes:
a system for implementing the method for measuring geometric characteristics of a part, the system comprising:
the point cloud model forming unit 100 is used for acquiring point cloud data of the precision machining component to form a point cloud model M;
the point cloud model segmentation unit 200 is used for carrying out element classification and segmentation on a point cloud model M of the precision machining component, realizing element classification by adopting a decision tree C4.5 method, realizing element segmentation by adopting a region growing algorithm and obtaining a segmentation result point cloud group Q;
a geometric feature recognition unit 300, the geometric feature recognition unit 300 being configured to match each segmented region of the segmentation result point cloud group Q with a design structure;
and the geometric accuracy detection unit 400 is used for calculating a numerical value of the geometric feature accuracy of the required precision machining component on the point cloud, comparing the numerical value with a design value and finally outputting a quality inspection report.
The part geometric feature measuring system of this embodiment has realized the intelligent measurement to the precision finishing component, through with three-dimensional laser scanning technique, measurement error theory and adjustment technique, intelligent analysis technique's organic integration, improves geometric accuracy detection efficiency and rate of accuracy to the precision finishing component, has better practicality.
Those skilled in the art will appreciate that the architecture shown in fig. 6 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the systems in which the disclosed aspects may be implemented, and that a particular system suitable for computer operations may include more or fewer components than those shown, or some of the components may be combined, or have a different arrangement of components.
The method solves the technical problems of large error, low efficiency and high technical requirement on detection personnel in the traditional contact type part detection method, realizes intelligent measurement on the precision machining component, improves the geometric precision detection efficiency and accuracy of the precision machining component through organic integration of a three-dimensional laser scanning technology, a measurement error theory, an adjustment technology and an intelligent analysis technology, and has better practicability.
The above-mentioned serial numbers of the embodiments of the present application are merely for description and do not represent the merits of the embodiments. Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solution of the present application may be substantially or partially embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) as described above and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present application.
The present invention has been described in detail with reference to the foregoing embodiments, and the principles and embodiments of the present invention have been described herein with reference to specific examples, which are provided only to assist understanding of the methods and core concepts of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.

Claims (7)

1. A part geometric feature measuring method based on point cloud is characterized by comprising the following processes:
obtaining a training set D of a mechanical part point cloud model with marked elements, wherein the training set D comprises a plurality of single-point data D i ；
Generating a decision tree T for element extraction according to the training set D by adopting a C4.5 algorithm;
obtaining a point cloud model M of mechanical parts, wherein the point cloud M comprises a plurality of single points i ；
Point cloud M of each single point in point cloud model M i Performing near point query and calculating point cloud M i The local features of (a) are recorded as a set N;
inputting the set N into a decision tree T for element extraction to obtain a point cloud group P after element extraction;
segmenting the point cloud group P by adopting a region growing algorithm to obtain a segmented point cloud group Q;
calculating, identifying and grouping the geometrical characteristics of the target surface of the point cloud group Q to obtain measurement data of the geometrical characteristics of the part;
in the process of calculating, identifying and grouping the geometric features of the target surface of the point cloud group Q to obtain the measurement data of the geometric features of the part, the method further comprises the following steps:
s701, using RANSAC to align the cylindrical surfaces
Figure 136442DEST_PATH_IMAGE001
Fitting to obtain the center line of the fitted cylindrical surface
Figure 240664DEST_PATH_IMAGE002
According to the cylindrical surface
Figure 927998DEST_PATH_IMAGE003
Height of (2)
Figure 459473DEST_PATH_IMAGE004
Radius, of the shaft
Figure 366511DEST_PATH_IMAGE005
Calculating a centerline
Figure 274425DEST_PATH_IMAGE002
The distance between the center line of the other cylindrical surfaces is selected as the nearest distance and the farthest distance
Figure 81844DEST_PATH_IMAGE006
And
Figure 784220DEST_PATH_IMAGE007
building up a cylindrical surface
Figure 411511DEST_PATH_IMAGE001
The feature operator of (2);
s702, using RANSAC to plane
Figure 919852DEST_PATH_IMAGE008
Fitting to obtain a fitted plane, and positioning the center point of the fitted plane
Figure 519461DEST_PATH_IMAGE009
Plane normal vector
Figure 455056DEST_PATH_IMAGE010
Calculating out
Figure 772905DEST_PATH_IMAGE008
Area of
Figure 819358DEST_PATH_IMAGE011
Building a plane
Figure 539053DEST_PATH_IMAGE008
The feature operator of (3);
s703, constructing each cylindrical surface in part standard parts
Figure 645549DEST_PATH_IMAGE012
According to the characteristic operator of each cylindrical surface
Figure 450694DEST_PATH_IMAGE012
Characteristic operator completion hash table
Figure 300838DEST_PATH_IMAGE013
Creating;
s704, constructing each plane in the part standard component
Figure 875039DEST_PATH_IMAGE014
And according to each plane
Figure 653901DEST_PATH_IMAGE014
The feature operator of (1) completes the hash table
Figure 680763DEST_PATH_IMAGE015
Creating;
s705, characteristic operator
Figure 334599DEST_PATH_IMAGE016
And a hash table
Figure 763306DEST_PATH_IMAGE013
Matching to obtain feature pairs
Figure 211605DEST_PATH_IMAGE017
The identification of the geometric characteristic cylindrical surface of the part is realized;
s706, performing feature operator
Figure 991342DEST_PATH_IMAGE018
And a hash table
Figure 183289DEST_PATH_IMAGE019
Matching to obtain feature pairs
Figure 466503DEST_PATH_IMAGE020
Realizing the identification of the geometric feature plane of the part;
s707, obtaining a geometric dimension to be checked according to a part design standard;
s708, calculating the distance between the two planes of the mechanical part to be inspected respectively
Figure 351282DEST_PATH_IMAGE021
Distance between them
Figure 352736DEST_PATH_IMAGE022
；
S709, calculating the distance between the two cylindrical axes to be checkedAxial distance between cylindrical surfaces
Figure 348374DEST_PATH_IMAGE023
；
And judging whether the design standard of the part is met or not according to the measurement data of the geometric characteristics of the part.
2. The method of measuring geometric features of a part according to claim 1, wherein: in the process of generating the decision tree T for primitive extraction according to the training set D by adopting the C4.5 algorithm, the method also comprises the following steps:
s201, for each single point data D in the training set D i Using a kd-tree to perform near point query;
s202, calculating each single point data D i Gaussian curvature K;
s203, calculating single point data D i PFH descriptor F of (1);
s204, calculating the empirical entropy of the training set D
Figure 282832DEST_PATH_IMAGE024
；
S205, recording the features obtained in the steps S202-S203 as features A, and calculating the empirical condition entropy of the features A on the training set D
Figure 276196DEST_PATH_IMAGE025
；
S206, entropy according to empirical condition
Figure 827263DEST_PATH_IMAGE025
Calculating information gain
Figure 298695DEST_PATH_IMAGE026
；
S207, gain according to information
Figure 120283DEST_PATH_IMAGE026
Calculating an information gain ratio
Figure 284548DEST_PATH_IMAGE027
；
S208, selecting the information gain ratio
Figure 57332DEST_PATH_IMAGE027
Central maximum feature
Figure 332456DEST_PATH_IMAGE028
To characteristic(s)
Figure 772664DEST_PATH_IMAGE028
Each possible value of
Figure 842251DEST_PATH_IMAGE029
In accordance with
Figure 367911DEST_PATH_IMAGE030
Segmenting a training set D into a plurality of non-empty single-point data D i Taking the class with the maximum number of examples as a mark to construct a child node, and constructing a decision tree T by the node i and the child node thereof;
s209, aiming at the node i, using single-point data D i For the training set, the
Figure 181146DEST_PATH_IMAGE031
For the feature set, the steps S204-S208 are recursively called to loop until all instances in the training set D belong to the same class C k A decision tree T is obtained.
3. The part geometry measuring method according to claim 1, characterized in that: point cloud M of each single point in point cloud model M i Performing near point query and calculating point cloud M i In the process of recording the local features as a set N, the detailed process includes:
and (4) carrying out near point query on the point cloud Mi of each single point of the point cloud model M, calculating the point Gaussian curvature K and the PFH descriptor F based on the near points, and recording as a set N.
4. The part geometry measuring method according to claim 1, characterized in that: in the process of inputting the set N into the decision tree T for element extraction to obtain the point cloud group P after element extraction, the detailed process comprises the following steps:
inputting the set N into a decision tree T to obtain a point cloud M of each single point in the set N i The sample classification result is a primitive segmentation result, and a segmented point cloud group P is output.
5. The method of measuring geometric features of a part according to claim 1, wherein: in the process of segmenting the point cloud group P by adopting a region growing algorithm to obtain a segmented point cloud group Q, the method also comprises the following steps:
s601, randomly selecting a point P in the point cloud group P as a seed point, and setting the curvature of the seed point P as
Figure 475861DEST_PATH_IMAGE032
；
S602, carrying out near point search on the seed point p to obtain a near point p i Let a near point p i Has a curvature of
Figure 716349DEST_PATH_IMAGE033
Setting a curvature change threshold value of
Figure 729305DEST_PATH_IMAGE034
；
If it is
Figure 346231DEST_PATH_IMAGE035
Then will be close to point p i Polymerizing with the seed point p and adding the proximity point p i And as new seed points, clustering all the points to obtain a segmented point cloud group Q.
6. An apparatus for implementing the point cloud-based part geometric feature measurement method of any one of claims 1-5, comprising:
a first acquisition module (10), wherein the first acquisition module (10) is used for acquiring a training set D of a point cloud model of a mechanical part marked with elements, and the training set D comprises a plurality of single-point data D i ；
A decision tree T generating module (20), wherein the decision tree T generating module (20) is used for generating a decision tree T for element extraction according to the training set D by adopting a C4.5 algorithm;
a second acquisition module (30), wherein the second acquisition module (30) is used for acquiring a mechanical part point cloud model M, and the point cloud M of a plurality of single points is contained in the cloud model M i ；
A local feature calculation module (40), the local feature calculation module (40) being configured to calculate a point cloud M for each single point in the point cloud model M i Performing near point query and calculating point cloud M i The local features of (a) are recorded as a set N;
the extraction module (50) is used for inputting the set N into the decision tree T for element extraction to obtain a point cloud group P after the element extraction;
the segmentation module (60) is used for segmenting the point cloud group P by adopting a region growing algorithm to obtain a segmented point cloud group Q;
the measurement data obtaining module (70), the measurement data obtaining module (70) is used for calculating, identifying and grouping the geometrical characteristics of the target surface of the point cloud group Q to obtain the measurement data of the geometrical characteristics of the parts;
the judging module (80) is used for judging whether the design standard of the part is met or not according to the measurement data of the geometric characteristics of the part.
7. A system for implementing the point cloud-based part geometric feature measurement method of any one of claims 1 to 5, comprising:
the point cloud model forming unit (100) is used for obtaining point cloud data of the precision machining component to form a point cloud model M;
the point cloud model segmentation unit (200) is used for carrying out element classification and segmentation on a point cloud model M of the precision machining component, realizing element classification by adopting a decision tree C4.5 method, realizing element segmentation by adopting a region growing algorithm and obtaining a segmentation result point cloud group Q;
a geometric feature recognition unit (300), the geometric feature recognition unit (300) being configured to match each segmented region of the segmentation result point cloud group Q with a design structure;
and the geometric precision detection unit (400) is used for calculating a numerical value of the geometric feature precision of the required precision machining component on the point cloud, comparing the numerical value with a design value and finally outputting a quality inspection report.
CN202211360827.4A 2022-11-02 2022-11-02 Part geometric feature measuring method, device and system based on point cloud Active CN115409886B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211360827.4A CN115409886B (en) 2022-11-02 2022-11-02 Part geometric feature measuring method, device and system based on point cloud

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211360827.4A CN115409886B (en) 2022-11-02 2022-11-02 Part geometric feature measuring method, device and system based on point cloud

Publications (2)

Publication Number Publication Date
CN115409886A CN115409886A (en) 2022-11-29
CN115409886B true CN115409886B (en) 2023-02-21

Family

ID=84169292

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211360827.4A Active CN115409886B (en) 2022-11-02 2022-11-02 Part geometric feature measuring method, device and system based on point cloud

Country Status (1)

Country Link
CN (1) CN115409886B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116046017B (en) * 2023-01-18 2024-05-07 上汽通用五菱汽车股份有限公司 Calibration method and device for measuring path, storage medium and computer equipment
CN116778260B (en) * 2023-08-17 2023-11-17 南京航空航天大学 Aviation rivet flushness detection method, device and system based on AdaBoost ensemble learning
CN116817771B (en) * 2023-08-28 2023-11-17 南京航空航天大学 Aerospace part coating thickness measurement method based on cylindrical voxel characteristics
CN117495868A (en) * 2024-01-03 2024-02-02 南京航空航天大学 Point cloud deep learning-based mechanical part assembly feature measurement method

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112435239A (en) * 2020-11-25 2021-03-02 南京农业大学 Scindapsus aureus leaf shape parameter estimation method based on MRE-PointNet and self-encoder model
CN113781553A (en) * 2021-09-08 2021-12-10 北京理工大学 Virtual detection method for geometric precision of mechanical product based on error modeling
CN115239951A (en) * 2022-06-08 2022-10-25 广东领慧建筑科技有限公司 Wall surface segmentation and identification method and system based on point cloud data processing

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109255833A (en) * 2018-09-30 2019-01-22 宁波工程学院 Based on semantic priori and the wide baseline densification method for reconstructing three-dimensional scene of gradual optimization

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112435239A (en) * 2020-11-25 2021-03-02 南京农业大学 Scindapsus aureus leaf shape parameter estimation method based on MRE-PointNet and self-encoder model
CN113781553A (en) * 2021-09-08 2021-12-10 北京理工大学 Virtual detection method for geometric precision of mechanical product based on error modeling
CN115239951A (en) * 2022-06-08 2022-10-25 广东领慧建筑科技有限公司 Wall surface segmentation and identification method and system based on point cloud data processing

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
三维点云数据的几何特性估算与特征识别;安毅;《中国博士学位论文全文数据库》;20120515;全文 *

Also Published As

Publication number Publication date
CN115409886A (en) 2022-11-29

Similar Documents

Publication Publication Date Title
CN115409886B (en) Part geometric feature measuring method, device and system based on point cloud
CN108107444B (en) Transformer substation foreign matter identification method based on laser data
CN107703480B (en) Mixed kernel function indoor positioning method based on machine learning
CN106373118B (en) The complex curved surface parts point cloud compressing method of border and local feature can be effectively retained
CN105528503B (en) A kind of large-scale component dynamic optimization design method based on STRUCTURE DECOMPOSITION
CN108303037B (en) Method and device for detecting workpiece surface shape difference based on point cloud analysis
Keselman et al. Many-to-many graph matching via metric embedding
CN105654483A (en) Three-dimensional point cloud full-automatic registration method
CN113344019A (en) K-means algorithm for improving decision value selection initial clustering center
CN111027140B (en) Airplane standard part model rapid reconstruction method based on multi-view point cloud data
CN114972459A (en) Point cloud registration method based on low-dimensional point cloud local feature descriptor
CN111914480B (en) Processing feature intelligent recognition method based on point cloud semantic segmentation
CN113961738A (en) Multi-feature casting three-dimensional model retrieval method and device
CN114663373A (en) Point cloud registration method and device for detecting surface quality of part
CN111504191A (en) Aviation part automatic rapid measurement method based on three-dimensional laser scanning
CN105574265B (en) Entire assembly model quantitative description towards model index
CN116008671A (en) Lightning positioning method based on time difference and clustering
CN117495891B (en) Point cloud edge detection method and device and electronic equipment
CN113255677B (en) Method, equipment and medium for rapidly extracting rock mass structural plane and occurrence information
Demir Automated detection of 3D roof planes from Lidar data
CN112561989A (en) Method for identifying hoisting object in construction scene
Xu et al. Binocular measurement method for the continuous casting slab model based on the improved BRISK algorithm
Liu et al. A fast weighted registration method of 3d point cloud based on curvature feature
CN116582309A (en) GAN-CNN-BiLSTM-based network intrusion detection method
CN103853817B (en) Based on the space singular point method of excavation of the magnanimity statistics of GIS

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