CN112164144B - Casting three-dimensional model classification method combining D2 operator and normal operator - Google Patents

Casting three-dimensional model classification method combining D2 operator and normal operator Download PDF

Info

Publication number
CN112164144B
CN112164144B CN202011017464.5A CN202011017464A CN112164144B CN 112164144 B CN112164144 B CN 112164144B CN 202011017464 A CN202011017464 A CN 202011017464A CN 112164144 B CN112164144 B CN 112164144B
Authority
CN
China
Prior art keywords
casting
dimensional
castings
operator
dimensional model
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
CN202011017464.5A
Other languages
Chinese (zh)
Other versions
CN112164144A (en
Inventor
周建新
万鹏
孙晓龙
张志鹏
计效园
刘迎辉
殷亚军
沈旭
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.)
Huazhong University of Science and Technology
Original Assignee
Huazhong University of Science and 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 Huazhong University of Science and Technology filed Critical Huazhong University of Science and Technology
Priority to CN202011017464.5A priority Critical patent/CN112164144B/en
Publication of CN112164144A publication Critical patent/CN112164144A/en
Application granted granted Critical
Publication of CN112164144B publication Critical patent/CN112164144B/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
    • 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
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/04—Manufacturing
    • 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
    • G06T7/62—Analysis of geometric attributes of area, perimeter, diameter or volume
    • 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
    • G06T7/66—Analysis of geometric attributes of image moments or centre of gravity
    • 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
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2200/00—Indexing scheme for image data processing or generation, in general
    • G06T2200/08—Indexing scheme for image data processing or generation, in general involving all processing steps from image acquisition to 3D model generation
    • 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
    • G06T2207/30116—Casting
    • 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)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Geometry (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • Business, Economics & Management (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Multimedia (AREA)
  • Manufacturing & Machinery (AREA)
  • Databases & Information Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Medical Informatics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Evolutionary Biology (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Graphics (AREA)
  • Computing Systems (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • Tourism & Hospitality (AREA)
  • General Business, Economics & Management (AREA)
  • Image Analysis (AREA)

Abstract

The invention belongs to the field of casting three-dimensional model classification, and particularly discloses a casting three-dimensional model classification method combining a D2 operator and a normal operator, which comprises the following steps: triangular surface tiling is carried out on the three-dimensional models of various castings; respectively adopting a D2 operator and a normal operator to obtain a D2 characteristic descriptor and an N2 characteristic descriptor of the casting three-dimensional model; then obtaining D2 and N2 category feature descriptors of various castings according to the initial categories of the castings; and finally, calculating the similarity of the three-dimensional model of the casting to be classified and the three-dimensional models of all classes by using a similarity measurement function, and classifying the three-dimensional model of the casting according to the similarity. The method can identify the category of the new casting based on the established database, improves the intelligent degree of casting process design, has good characteristic identification effect on common casting categories, can effectively judge the casting types, and solves the process reuse problem of the same type of casting.

Description

Casting three-dimensional model classification method combining D2 operator and normal operator
Technical Field
The invention belongs to the field of casting three-dimensional model classification, and particularly relates to a casting three-dimensional model classification method combining a D2 operator and a normal operator.
Background
The intelligent production mode is the target pursued by the casting enterprises, wherein the casting process design is used as a precedent link, and the realization of the intellectualization of the casting process design has very important significance.
Most of the product design modes of casting enterprises at the present stage are driven by experience, and the quality of process design depends on the experience of designers; the knowledge retrieval method relying on manual work has low reuse rate of mature process schemes of enterprises, and causes the inefficiency of process design. With the increasingly wide application of three-dimensional models in casting production, the number and the types of the models are increasingly abundant, the model reuse technical advantages in the process of process design become obvious day by day, and a database of an enterprise has a large amount of design data, historical processes, three-dimensional casting models and other professional data, which is a wealth summarizing historical experience and results, but is not efficiently managed and reused. Therefore, how to identify the category of a new casting through the established database and improve the intelligent degree of casting process design is a technical problem to be solved urgently.
Disclosure of Invention
Aiming at the defects or improvement requirements of the prior art, the invention provides a casting three-dimensional model classification method combining a D2 operator and a normal operator, and aims to perform casting three-dimensional model classification combining a D2 operator and a normal operator, accurately and effectively identify the category of a new casting based on an established database and improve the intelligent degree of casting process design.
In order to achieve the purpose, the invention provides a casting three-dimensional model classification method combining a D2 operator and a normal operator, which comprises the following steps:
s1, obtaining three-dimensional models of various castings, and performing triangular surface tiling on the three-dimensional models;
s2, randomly selecting a plurality of point pairs on the surface of the three-dimensional model of the casting, calculating the distance between two points of each point pair to obtain an array containing the distances of all the point pairs, normalizing the array, setting a statistical interval, and counting the times of the point pair distances in the array appearing in each interval to form a feature vector, namely a D2 feature descriptor of the three-dimensional model of the casting;
s3, randomly selecting a plurality of patch pairs on the three-dimensional model of the casting, calculating cosine values of included angles of normal vectors of two patches of each patch pair, obtaining an array containing all the cosine values, setting a statistical interval, and counting the times of the cosine values of the included angles in the array appearing in each interval to form a feature vector, namely an N2 feature descriptor of the three-dimensional model of the casting;
s4, according to the initial type of the casting, respectively obtaining D2 type characteristic descriptors and N2 type characteristic descriptors of various types of castings through the D2 characteristic descriptors and the N2 characteristic descriptors of the three-dimensional casting model;
s5, acquiring D2 feature descriptors of the three-dimensional models of the castings to be classified, and calculating the similarity Sim between the D2 feature descriptors and various D2 category feature descriptors of the castingsD2(ii) a Acquiring N2 feature descriptors of the three-dimensional model of the casting to be classified, and calculating the similarity Sim between the N2 feature descriptors and the N2 category feature descriptors of various castingsN2；
S6, according to similarity SimD2And SimN2And obtaining the total similarity of the three-dimensional model of the casting to be classified and the three-dimensional models of the castings of all classes, and further judging the class of the casting to be classified to realize the classification of the three-dimensional model of the casting.
Preferably, in the step S2, the obtaining of the D2 feature descriptor of the three-dimensional casting model specifically includes the following steps:
s21, calculating the areas of all triangular patches on the three-dimensional model of the casting according to the Helen formula, and storing the areas in a list S [ S ]1,S2,S3…Sn]Then, an ordered area array T [ T ] is obtained from the following formula according to the list S1,T2,T3…Tn]，x＝1,2,3…n：
Figure BDA0002699543290000021
S22, using random number function, at 0-TnGenerating a plurality of random number pairs, determining subscripts of the random number pairs in the ordered area array by utilizing binary search, and determining a plurality of sampling patch pairs;
s23, determining a plurality of point pairs on the sampling patch pair, calculating Euclidean distances among the point pairs to form an array containing the distances of all the point pairs, and standardizing the array by using the maximum value and the minimum value of the distances in the array;
and S24, determining the number of the statistical intervals so as to determine the range of each statistical interval, wherein the number of times of point pairs in the statistical group of point pairs and the distance in each interval range is used as a D2 characteristic descriptor of the casting three-dimensional model.
Preferably, in step S23, a plurality of point pairs are determined on the sampling patch pair according to a point-taking formula, where the point-taking formula is as follows:
Figure BDA0002699543290000031
wherein A, B, C is the three vertex coordinates of the sample patch, P1Is a sampling point, r1、r2Is [0,1 ]]The random number of (2).
Preferably, in the step S3, the obtaining of the N2 feature descriptor of the three-dimensional casting model specifically includes the following steps:
s31, calculating the three-dimensional model of the casting according to the Helen formulaThe areas of all triangle patches are stored in a list S [ S ]1,S2,S3…Sn]Then, an ordered area array T [ T ] is obtained from the following formula according to the list S1,T2,T3…Tn]，x＝1,2,3…n：
Figure BDA0002699543290000032
S32, using random number function, at 0-TnGenerating a plurality of random number pairs, determining subscripts of the random number pairs in the ordered area array by utilizing binary search, and determining a plurality of sampling patch pairs;
s33, calculating the cosine value of the included angle between the normal vectors of the sampling patch pair to form an array of cosine values of the included angle of the sampling patch pair;
and S34, determining the number of the statistical intervals so as to determine the range of each statistical interval, and counting the number of times of the cosine value of the included angle in the array appearing in each interval range to be used as the N2 characteristic descriptor of the three-dimensional casting model.
More preferably, in the step S4, in different casting categories, the D2 feature descriptors and the N2 feature descriptors of the three-dimensional casting model are averaged to obtain the D2 and N2 category feature descriptors of each casting category.
Preferably, in step S5, the euclidean distance between the D2 feature descriptor of the casting to be classified and the D2 category feature descriptor of each type of casting is calculated to obtain a plurality of distance values, and the distance values are normalized to obtain the similarity SimD2(ii) a Calculating Euclidean distances between the N2 feature descriptors of the castings to be classified and the N2 category feature descriptors of various castings to obtain a plurality of distance values, normalizing the distance values to obtain similarity SimN2。
More preferably, in step S6, the similarity Sim is compounded using a compounding formulaD2And SimN2Obtaining the total similarity SimK of the three-dimensional model of the casting to be classified and the three-dimensional models of all classes of castings, and obtaining the corresponding casting with the maximum total similarityThe category is used as the category of the casting to be classified; the composite formula is as follows:
SimK＝K*SimD2+(1-K)*SimN2
wherein, the weighting factor K is 0.8.
More preferably, the initial category includes a pulley-type casting, a bearing seat-type casting, a link-type casting, a spoke-type casting, a tire outer-mold-type casting, and a mine frame-type casting.
Generally, compared with the prior art, the above technical solution conceived by the present invention mainly has the following technical advantages:
1. the method combines a D2 operator and a normal operator (N2 operator) to extract a feature descriptor of the three-dimensional model of the casting, then calculates the similarity between the three-dimensional model of the casting and the three-dimensional models of various categories by using a similarity measurement function, and efficiently and accurately classifies the three-dimensional model of the casting according to the similarity; the method has good characteristic recognition effect on common casting types, can accurately and effectively judge the casting types, and solves the process reuse problem of the same type of castings.
2. The three-dimensional model classification algorithm takes the shape distribution of the three-dimensional model as a core, and the adopted D2 operator is one of the shape distribution algorithms with better distinguishing effect and has good robustness for model scaling; the surface characteristic information of the three-dimensional model is extracted through the normal operator, and the N2 operator characteristic descriptor of the casting three-dimensional model is extracted, so that the defect that the D2 operator is insensitive to the characteristics when part of complex casting three-dimensional models are extracted is overcome.
3. When a sampling surface patch and a sampling point are determined, the model characteristics for determining the category of the casting can be better extracted by adopting an area weighting mode, and the algorithm is ensured to have good robustness to partial surface patch loss; meanwhile, the D2 and N2 feature descriptors have good model rotation robustness.
Drawings
FIG. 1 is a flow chart of a method for classifying three-dimensional models of castings according to an embodiment of the present invention in combination with a D2 operator and a normal operator;
FIGS. 2 (a) - (f) show three-dimensional models of various new castings to be classified according to the embodiment of the present invention;
FIG. 3 is a D2 shape distribution histogram of a three-dimensional model of a casting to be classified according to an embodiment of the invention;
FIG. 4 is a histogram of the distribution of N2 shapes of a three-dimensional model of a casting to be classified according to an embodiment of the present invention;
FIG. 5 is a diagram of initial categories of three-dimensional models of castings according to an embodiment of the present invention;
fig. 6 is a three-dimensional model of a typical casting of each type according to the initial classification of the embodiment of the present invention, wherein (a) (b) is a pulley type, (c) (d) is a bearing seat type, (e) (f) is a connecting rod type, (g) (h) is a spoke type, (i) (j) is a tire outer mold type, and (k) (l) is a mining machine frame type.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
The casting three-dimensional model classification method combining the D2 operator and the normal operator provided by the embodiment of the invention is shown in FIG. 1; the three-dimensional model of the casting shown in (b) in fig. 2 is used as a new casting to be classified for explanation, the number of the statistical intervals is 180, and the number of the sampling points (sampling surface patches) is 400000. The classification method specifically comprises the following steps:
step 1: reading various three-dimensional model files of the casting from the model library, and performing triangular surface tiling on the three-dimensional model to obtain an STL text file of the casting.
Step 1-1, analyzing and reading a three-dimensional model file of the casting by using three-dimensional modeling software.
And 1-2, obtaining a triangular tiling file of the three-dimensional model of the casting by using a three-dimensional model tiling function in software, and storing the triangular tiling file by using an STL text file.
Specifically, after the three-dimensional model of the casting is analyzed and read and the triangular patches are formed by using three-dimensional modeling software, the three-dimensional vertex coordinates and the three-dimensional normal vector coordinates of the model patches are respectively stored by using 6 list data structures, and meanwhile, subscripts of a list are used for corresponding to the serial numbers of the triangular patches.
Step 2: reading a three-dimensional model STL file of the casting, randomly selecting a plurality of point pairs on the surface of the three-dimensional model, calculating the distance between two points of each point pair to obtain an array containing the distances of all the point pairs, normalizing the array of the distances, setting a statistical interval, and counting the times of the point pair distances in the array appearing in all the intervals to form a characteristic vector which is used as a D2 characteristic descriptor of the three-dimensional model of the casting.
Step 2-1, sequentially calculating the area S of each triangular patch according to a Helen formula, and storing the area S in a list S [ S ]1,S2,S3…Sn]The Helen formula is as follows:
p＝(a+b+c)/2
Figure BDA0002699543290000061
wherein, a, b and c are three side lengths of the triangular patch.
Step 2-2, according to the list S, successively accumulating the areas of the patches to obtain an ordered area array, and recording the ordered area array as T [ T ]1,T2,T3…Tn]Wherein T isnAll triangular surface patches have the sum of areas, and the subscripts of the array correspond to the accumulated serial numbers of the triangular surface patches, TxThe calculation formula of (a) is as follows:
Figure BDA0002699543290000062
step 2-3 utilizes random number function, at 0-TnAnd generating a plurality of random number pairs, determining subscripts of the random number pairs in the area array by utilizing binary search, and determining a plurality of sampling patch pairs.
Step 2-4, determining a plurality of point pairs on the sampling surface patch pair according to a point taking formula, wherein the point taking formula is as follows:
Figure BDA0002699543290000063
where A, B, C is the three vertex coordinates of the sample patch, P1Is a sampling point, r1、r2Is [0,1 ]]The random number of (2).
And 2-5, calculating Euclidean distances among the selected point pairs to form a distance array of the point pairs.
Steps 2-6 utilize the maximum value of distance Dis in the distance arraymaxAnd minimum value DisminThe distance array is normalized by the following formula:
Figure BDA0002699543290000071
therein, DisiIs the distance value before normalization, disiIs the normalized distance value.
And 2-7, determining the number of the statistical intervals, thereby determining the range of each statistical interval, and counting the number of elements in the distance normalized array of the points in each interval range to serve as the D2 feature descriptor of the three-dimensional model.
And step 3: reading an STL file of a three-dimensional model of the casting, randomly selecting a plurality of dough sheet pairs, calculating cosine values of included angles of normal vectors of two dough sheets of each dough sheet pair, obtaining an array containing all the cosine values, setting a statistical interval, and counting the times of the cosine values of the included angles in the array appearing in all the intervals to form a characteristic vector which is used as an N2 characteristic descriptor of the three-dimensional model of the casting.
Step 3-1, calculating the area of all triangular patches in sequence according to the Helen formula, and storing the calculated area in a list S [ S ]1,S2,S3…Sn]Performing the following steps;
step 3-2, according to the list S, successively accumulating the areas of the patches to obtain an ordered area array, and recording the ordered area array as T [ T ]1,T2,T3…Tn]Wherein T isnAll the triangular surface patches are summed, and the subscripts of the array correspond to the accumulated serial numbers of the triangular surface patches.
Step 3-3 utilizes a random number function between 0 and TnAnd generating a plurality of random number pairs, determining subscripts of the random number pairs in the area array by utilizing binary search, and determining a plurality of sampling patch pairs.
And 3-4, calculating the cosine values of included angles between the normal vectors of the selected plurality of sampling surface patch pairs to form an array of the cosine values of the included angles of the sampling surface patch pairs.
And 3-5, determining the number of the statistical intervals so as to determine the range of each statistical interval, and counting the number of elements in the cosine value array of the included angle of the patch pair in each interval range to serve as the N2 feature descriptor of the three-dimensional model.
And 4, step 4: and determining the initial classification category of the three-dimensional casting model, and calculating to obtain D2 category characteristic descriptors representing different casting categories.
Step 4-1, determining classification categories of the three-dimensional casting models, as shown in fig. 5, wherein the initial categories comprise pulleys, bearing seats, connecting rods, spokes, external tire molds and mine frames, and two typical three-dimensional casting models are respectively given from the six categories, as shown in fig. 6.
And 4-2, extracting D2 characteristic descriptors of all typical casting three-dimensional models.
Step 4-3, averaging the D2 feature descriptors of the typical casting three-dimensional model in FIG. 6 in different categories by using a mean formula to obtain D2 category feature descriptors representing the categories, wherein the mean formula is as follows:
Figure BDA0002699543290000081
wherein, SDiD2 feature descriptors for a typical casting model, and SD is the corresponding category feature descriptors.
For various types of typical casting three-dimensional models, the D2 category feature descriptors are respectively as follows: pulleys [0.0000615,0.000143,0.000241,0.000335,0.000454,0.000671,0.0008345, · ], bearing seats [0.00005599,0.000143,0.0002185,0.000308,0.000402,0.0005149, · ], connecting rods [0.000149,0.0004115,0.000702,0.001014,0.001294,0.001614, ·, spokes [0.00006001,0.000152,0.000255,0.0003673,0.0004439,0.0005435, · ], tire outer molds [0.0000509,0.000148,0.000233,0.0003275,0.000434,0.000544,0.000667,0.0007305, · ], mine frames [0.0000475,0.00010999,0.0001765,0.000300,0.0004325,0.0005455,0.000686, · ].
And 5: and determining the initial classification category of the three-dimensional casting model, and calculating to obtain N2 category characteristic descriptors representing different casting categories.
Step 5-1, determining classification categories of the three-dimensional casting models, as shown in fig. 5, wherein the initial categories comprise pulleys, bearing seats, connecting rods, spokes, external tire molds and mine frames, and two typical three-dimensional casting models are respectively given from the six categories, as shown in fig. 6.
Step 5-2, extracting N2 feature descriptors of all typical casting three-dimensional models in different categories;
and 5-3, averaging the N2 feature descriptors of all the typical casting three-dimensional models in the graph 6 in different categories by using a mean formula to obtain N2 category feature descriptors representing the categories.
For various types of typical casting three-dimensional models, the N2 category feature descriptors are respectively as follows: pulleys [0.006608,0.002759,0.007200,0.006417,0.007148,0.010161, bearing seats [0.001417,0.002343,0.005572,0.004429,0.003232,0.003788, connecting rods [0.001261,0.00252,0.0066,0.004947,0.004112,0.004038, 0.013864,0.005872 ], spokes [0.000068,0.00131,0.006784,0.005993,0.013864,0.005872 ], tire molds [0.003193,0.005554,0.005202,0.005407,0.005371,0.0049229, 0.0055130.007177 ], mine frames [0.001521,0.002818,0.006161,0.006242, 0.0055130.007177 ].
Furthermore, D2 and N2 category feature descriptors of various casting three-dimensional models can be stored in a feature library so as to be directly used in subsequent classification.
Step 6: and (3) taking the casting three-dimensional model shown in (b) in fig. 2 as a casting three-dimensional model to be classified, acquiring the D2 feature descriptor of the casting, and obtaining the similarity between the D2 feature descriptor of the new casting and the D2 category feature descriptors of various types by utilizing an Euclidean distance formula.
Step 6-1 shows the three-dimensional casting model shown in fig. 2 (b) with D2 distribution characteristics [0.0000675, 0.0001425, 0.0001825, 0.0003125, 0.0004075, 0.0005425, 0.0005725, 0.0006825, 0.0008425, 0.0009525, 0.0010725, 0.00129, 0.0013325, · ], and frequency distribution histogram as shown in fig. 3.
Step 6-2, calculating D2 characteristic descriptor SD of castings to be classified by using Euclidean distance formulaA＝[SDA1,SDA2,SDA3,…SDAM]And D2 category feature descriptor SD ═ SD for each category1,SD2,SD3,…SDM]The euclidean distance between them, which yields a plurality of distance values, the euclidean distance formula is as follows:
Figure BDA0002699543290000091
step 6-3, uniformizing the distance values by utilizing a uniformization formula to obtain the D2 feature descriptor similarity Sim of the castings to be classified and the castings of each class, wherein the uniformization formula is as follows:
Figure BDA0002699543290000101
wherein, the larger the Sim value is, the more similar the casting to be classified and the casting of the category are.
Under the D2 operator, the similarity of the three-dimensional model of the casting to be classified and each class of casting model is pulley class: 0.978575, respectively; bearing seats: 0.985982, respectively; connecting rods: 0.980085, respectively; spoke type: 0.990085, respectively; external tire molds: 0.974192, respectively; mine machine frame: 0.989920.
and 7: and acquiring the N2 feature descriptors of the three-dimensional model of the casting to be classified, and obtaining the similarity between the N2 feature descriptors of the casting to be classified and the various N2 type feature descriptors by utilizing an Euclidean distance formula.
Step 7-1 the distribution characteristics of the N2 shape of the three-dimensional model of the casting to be classified are [0.07709, 0.00831, 0.00574, 0.00625, 0.00392, 0.00414, 0.00441, 0.00379, 0.00327, 0.00354,. cndot.. distribution histogram of the frequency distribution histogram of the three-dimensional model of the casting to be classified in step 4 is shown in FIG. cndot..
7-2, calculating Euclidean distances between the N2 feature descriptors of the castings to be classified and the N2 category feature descriptors representing various categories by using an Euclidean distance formula to obtain a plurality of distance values;
and 7-3, homogenizing the distance values by utilizing a homogenization formula to obtain the similarity of the D2 feature descriptors of the castings to be classified and the castings of all classes.
Under the N2 operator, the similarity of the three-dimensional model of the casting to be classified and the casting models of all classes is pulley class: 0.823963, respectively; bearing seats: 0.894464, respectively; connecting rods: 0.869926, respectively; spoke type: 0.851242, respectively; external tire molds: 0.863209, respectively; mine machine frame: 0.827487.
and 8: and obtaining the total similarity of the three-dimensional model to be classified and various casting three-dimensional models by using the weight factors, and realizing the classification of the casting three-dimensional models by judging the similarity.
Step 8-1, compounding the similarity of the D2 feature descriptors and the similarity of the N2 feature descriptors of the castings to be classified and the castings of all classes by using a weight factor K to obtain the total similarity SimK of the castings to be classified and the castings of all classes, wherein a compound operator similarity calculation formula is as follows:
SimK＝K*SimD2+(1-K)*SimN2
wherein, for the current initial category, K is 0.8, so as to better utilize the robustness of the D2 operator on model scaling, and simultaneously improve the defect that the D2 operator is insensitive to the characteristics when extracting a part of complex casting three-dimensional model through the N2 operator.
And 8-2, retrieving the corresponding casting category when the total similarity is maximum, namely the category of the casting to be classified, and realizing the classification of the three-dimensional model of the casting.
The total similarity of the three-dimensional model of the casting to be classified and the casting models of all classes is respectively as follows:
pulleys: 0.946554, respectively; bearing seats: 0.978811, respectively; connecting rods: 0.948969, respectively; spoke type: 0.951002, respectively; external tire molds: 0.953141, respectively; mine machine frame: 0.952243. and searching the corresponding casting category when the similarity is searched, so that the category of the casting three-dimensional model to be classified is the bearing seat category.
In order to better represent the classification capability of the algorithm of the invention on the common categories of the steel castings, the classification results of the three-dimensional models of various new castings to be classified shown in the figure 2 in the algorithm of the invention are given in the table 1, and it can be seen that the classification results are all accurate.
TABLE 1 results of three-dimensional model classification of new castings to be classified
Figure BDA0002699543290000111
The invention combines a casting three-dimensional model classification algorithm of a D2 operator and a normal operator, and provides a method for respectively extracting the surface characteristics and point characteristic information of a three-dimensional model by the normal operator and the D2 operator and carrying out similarity measurement of the characteristic information on the basis of the 'tiling' of the model, thereby realizing the classification retrieval of the three-dimensional model. The algorithm has a good classification effect on common steel casting models such as mine frames, bearing seats and connecting rods.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.

Claims (8)

1. A three-dimensional casting model classification method combining a D2 operator and a normal operator is characterized by comprising the following steps:
s1, obtaining three-dimensional models of various castings, and performing triangular surface tiling on the three-dimensional models;
s2, randomly selecting a plurality of point pairs on the surface of the three-dimensional model of the casting, calculating the distance between two points of each point pair to obtain an array containing the distances of all the point pairs, normalizing the array, setting a statistical interval, and counting the times of the point pair distances in the array appearing in each interval to form a feature vector, namely a D2 feature descriptor of the three-dimensional model of the casting;
s3, randomly selecting a plurality of patch pairs on the three-dimensional model of the casting, calculating cosine values of included angles of normal vectors of two patches of each patch pair, obtaining an array containing all the cosine values, setting a statistical interval, and counting the times of the cosine values of the included angles in the array appearing in each interval to form a feature vector, namely an N2 feature descriptor of the three-dimensional model of the casting;
s4, according to the initial type of the casting, respectively obtaining D2 type characteristic descriptors and N2 type characteristic descriptors of various types of castings through the D2 characteristic descriptors and the N2 characteristic descriptors of the three-dimensional casting model;
s5, acquiring D2 feature descriptors of the three-dimensional models of the castings to be classified, and calculating the similarity Sim between the D2 feature descriptors and various D2 category feature descriptors of the castingsD2(ii) a Acquiring N2 feature descriptors of the three-dimensional model of the casting to be classified, and calculating the similarity Sim between the N2 feature descriptors and the N2 category feature descriptors of various castingsN2；
S6, according to similarity SimD2And SimN2And obtaining the total similarity of the three-dimensional model of the casting to be classified and the three-dimensional models of the castings of all classes, and further judging the class of the casting to be classified to realize the classification of the three-dimensional model of the casting.
2. The method for classifying the three-dimensional casting model by combining the D2 operator and the normal operator as claimed in claim 1, wherein the step S2 of obtaining the D2 feature descriptor of the three-dimensional casting model comprises the following steps:
s21, calculating the areas of all triangular patches on the three-dimensional model of the casting according to the Helen formula, and storing the areas in a list S [ S ]1,S2,S3…Sn]Then, an ordered area array T [ T ] is obtained from the following formula according to the list S1,T2,T3…Tn]，x＝1,2,3…n：
Figure FDA0002699543280000021
S22, using random number function, at 0-TnGenerating a plurality of random number pairs, determining subscripts of the random number pairs in the ordered area array by utilizing binary search, and determining a plurality of sampling patch pairs;
s23, determining a plurality of point pairs on the sampling patch pair, calculating Euclidean distances among the point pairs to form an array containing the distances of all the point pairs, and standardizing the array by using the maximum value and the minimum value of the distances in the array;
and S24, determining the number of the statistical intervals so as to determine the range of each statistical interval, wherein the number of times of point pairs in the statistical group of point pairs and the distance in each interval range is used as a D2 characteristic descriptor of the casting three-dimensional model.
3. The method for classifying a three-dimensional model of a casting using a combination of a D2 operator and a normal operator as claimed in claim 2, wherein in step S23, a plurality of point pairs are determined on the pairs of sampling patches according to a point-taking formula as follows:
Figure FDA0002699543280000022
where A, B, C is the three vertex coordinates of the sample patch, P1Is a sampling point, r1、r2Is [0,1 ]]The random number of (2).
4. The method for classifying the three-dimensional casting model by combining the D2 operator and the normal operator as claimed in claim 1, wherein the step S3 of obtaining the N2 feature descriptor of the three-dimensional casting model specifically comprises the following steps:
s31, calculating the areas of all triangular patches on the three-dimensional model of the casting according to the Helen formula, and storing the areas in a list S [ S ]1,S2,S3…Sn]Then, an ordered area array is obtained according to the list S by the following formulaT[T1,T2,T3…Tn]，x＝1,2,3…n：
Figure FDA0002699543280000023
S32, using random number function, at 0-TnGenerating a plurality of random number pairs, determining subscripts of the random number pairs in the ordered area array by utilizing binary search, and determining a plurality of sampling patch pairs;
s33, calculating the cosine value of the included angle between the normal vectors of the sampling patch pair to form an array of cosine values of the included angle of the sampling patch pair;
and S34, determining the number of the statistical intervals so as to determine the range of each statistical interval, and counting the number of times of the cosine value of the included angle in the array appearing in each interval range to be used as the N2 characteristic descriptor of the three-dimensional casting model.
5. The method for classifying the three-dimensional casting models by combining the D2 operator and the normal operator as claimed in claim 1, wherein in the step S4, the D2 feature descriptors and the N2 feature descriptors of the three-dimensional casting models are respectively averaged to obtain the D2 and N2 feature descriptors of all types of castings in different casting types.
6. The method for classifying the three-dimensional casting models by combining the D2 operator and the normal operator as claimed in claim 1, wherein in the step S5, Euclidean distances between D2 feature descriptors of the castings to be classified and D2 category feature descriptors of various types of castings are calculated to obtain a plurality of distance values, and the distance values are normalized to obtain the similarity SimD2(ii) a Calculating Euclidean distances between the N2 feature descriptors of the castings to be classified and the N2 category feature descriptors of various castings to obtain a plurality of distance values, normalizing the distance values to obtain similarity SimN2。
7. The method for classifying a three-dimensional model of a casting using a combination of the D2 operator and the normal operator as defined in claim 1The method is characterized in that in the step S6, the similarity Sim is compounded by using a compound formulaD2And SimN2Obtaining the total similarity SimK of the three-dimensional model of the casting to be classified and the three-dimensional models of all classes of castings, and taking the corresponding casting class with the maximum total similarity as the class of the casting to be classified; the composite formula is as follows:
SimK＝K*SimD2+(1-K)*SimN2
wherein, the weighting factor K is 0.8.
8. The method for three-dimensional model classification of castings according to the combination of the D2 operator and the normal operator, according to any of the claims 1-7, wherein said initial categories include pulley-type castings, bearing-housing-type castings, connecting-rod-type castings, spoke-type castings, external-tire-type castings, mine-rack-type castings.
CN202011017464.5A 2020-09-24 2020-09-24 Casting three-dimensional model classification method combining D2 operator and normal operator Active CN112164144B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011017464.5A CN112164144B (en) 2020-09-24 2020-09-24 Casting three-dimensional model classification method combining D2 operator and normal operator

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011017464.5A CN112164144B (en) 2020-09-24 2020-09-24 Casting three-dimensional model classification method combining D2 operator and normal operator

Publications (2)

Publication Number Publication Date
CN112164144A CN112164144A (en) 2021-01-01
CN112164144B true CN112164144B (en) 2022-04-19

Family

ID=73863886

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011017464.5A Active CN112164144B (en) 2020-09-24 2020-09-24 Casting three-dimensional model classification method combining D2 operator and normal operator

Country Status (1)

Country Link
CN (1) CN112164144B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113393145B (en) * 2021-06-25 2023-06-30 广东利元亨智能装备股份有限公司 Model similarity obtaining method and device, electronic equipment and storage medium

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105118044A (en) * 2015-06-16 2015-12-02 华南理工大学 Method for automatically detecting defects of wheel-shaped cast product
CN110006907A (en) * 2019-04-10 2019-07-12 清华大学深圳研究生院 A kind of die casting detection method of surface flaw and system based on machine vision
CN110909697A (en) * 2019-11-28 2020-03-24 华中科技大学 Casting three-dimensional feature extraction and similarity measurement method combined with technological parameters

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8655476B2 (en) * 2011-03-09 2014-02-18 GM Global Technology Operations LLC Systems and methods for computationally developing manufacturable and durable cast components

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105118044A (en) * 2015-06-16 2015-12-02 华南理工大学 Method for automatically detecting defects of wheel-shaped cast product
CN110006907A (en) * 2019-04-10 2019-07-12 清华大学深圳研究生院 A kind of die casting detection method of surface flaw and system based on machine vision
CN110909697A (en) * 2019-11-28 2020-03-24 华中科技大学 Casting three-dimensional feature extraction and similarity measurement method combined with technological parameters

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
A Method for Modeling and Extracting 3D Structural Features of Castings Considering Siz;Zhipeng Zhang等;《Procedia Manufacturing》;20200106;第563-570页 *
基于法向分类的三维模型形状分布检索算法;王洪申等;《计算机集成制造系统》;20090630;第1187-1193页 *
基于聚类分析的三维模型检索综述;毛瑞等;《微型机与应用》;20120525(第10期);第1-3页 *
法向约束的多幅点云数据融合算法;梁新合等;《西安交通大学学报》;20090510(第05期);第71-75页 *

Also Published As

Publication number Publication date
CN112164144A (en) 2021-01-01

Similar Documents

Publication Publication Date Title
CN109887015B (en) Point cloud automatic registration method based on local curved surface feature histogram
CN110443281B (en) Text classification self-adaptive oversampling method based on HDBSCAN (high-density binary-coded decimal) clustering
CN109523518B (en) Tire X-ray defect detection method
CN106228185B (en) A kind of general image classifying and identifying system neural network based and method
CN110826618A (en) Personal credit risk assessment method based on random forest
CN105046694A (en) Quick point cloud registration method based on curved surface fitting coefficient features
CN104731916A (en) Optimizing initial center K-means clustering method based on density in data mining
CN113344019A (en) K-means algorithm for improving decision value selection initial clustering center
CN110021028A (en) A kind of automatic clothing method based on garment fashion drawing
CN111898443A (en) Flow monitoring method for wire feeding mechanism of FDM type 3D printer
CN112164144B (en) Casting three-dimensional model classification method combining D2 operator and normal operator
CN116523320A (en) Intellectual property risk intelligent analysis method based on Internet big data
CN108537249B (en) Industrial process data clustering method for density peak clustering
CN110909697A (en) Casting three-dimensional feature extraction and similarity measurement method combined with technological parameters
CN110909792A (en) Clustering analysis method based on improved K-means algorithm and new clustering effectiveness index
CN117495891B (en) Point cloud edge detection method and device and electronic equipment
CN111553410B (en) Point cloud identification method based on key point local curved surface feature histogram and spatial relationship
CN116432052B (en) Quality detection method for clamp for new energy automobile die
CN109035311A (en) A kind of curved bone fracture autoregistration and internal fixation steel plate pre-bending modeling method
CN108376266A (en) One-class support vector machines Optimization Method of Kernel Parameter based on sample edge point internal point
CN102262682A (en) Rapid attribute reduction method based on rough classification knowledge discovery
CN104036232A (en) Image edge feature analysis-based necktie pattern retrieval method
CN113361616A (en) K-means algorithm for optimizing clustering center
CN111985823B (en) Crystal bar quality assessment method for roller mill orientation instrument
CN110059353B (en) Automatic modal parameter identification, simplification and practical method

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