CN108108513A - Face abstracting method in thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features - Google Patents
Face abstracting method in thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features Download PDFInfo
- Publication number
- CN108108513A CN108108513A CN201711145923.6A CN201711145923A CN108108513A CN 108108513 A CN108108513 A CN 108108513A CN 201711145923 A CN201711145923 A CN 201711145923A CN 108108513 A CN108108513 A CN 108108513A
- Authority
- CN
- China
- Prior art keywords
- face
- model
- muscle
- features
- thin
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/18—Manufacturability analysis or optimisation for manufacturability
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to CAD and CAE fields and data mining technology, face abstracting method in the thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features supports the efficiently and accurately extraction in face in the thin-wall part model with complicated muscle feature structure, and step includes：The definition of step 1 hierarchical structure semanteme；Efficient identification and the model segmentation of step 2 muscle feature；Face is extracted in step 3 hierarchical semantic；Step 4 mid-plane model is generated and reused, and based on this support model variant design when variable domain middle face result quickly generate, achieve the purpose that improve mid-plane model formation efficiency, geometric accuracy, topology availability, applicability and reusability.
Description
Technical field
The present invention relates to CAD and CAE fields and data mining technologies, and in particular to a kind of thin-walled based on muscle Image Segmentation Methods Based on Features
Face extracting method in part model hierarchy semanteme.
Background technology
In Modern Industry Products R＆D process, it is verification validity of products that Finite Element Simulation Analysis is carried out to deisgn product
One of important means.Before product model carries out Finite Element Simulation Analysis, it usually needs carry out minutia inhibition and drop to it
Dimension etc. simplifies operation to improve subsequent meshes formation efficiency and quality, and then improves simulation analysis efficiency.Wherein, thin-wall part is extensive
Applied in the product models such as complicated and high-precision automobile, aerospace, model simplification is to the performance point of thin-wall part model
It analyses most important.Middle face is extracted to be widely used as the most effective dimensionality reduction simplification method for thin-wall part model, important
Property is self-evident.
At present, have face in some and extract correlative study achievement, part achievement has been applied to business software, but multiple in processing
The middle face of miscellaneous thin-wall part model there are problems that when extracting, such as middle face formation efficiency, geometric accuracy, topology availability, applicable
The problems such as property and reusability.It is existing especially for thin-wall parts models such as the automobile with complicated muscle feature structure, aviation coverings
There is middle face abstracting method to simplify efficiency low, and can not effectively solve the mid-plane model adaptive updates after initial model design change
Problem.
The content of the invention
The present invention provides face abstracting method in a kind of thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features, with master
CAD software and CAE software are flowed for support, by the own characteristic of depth profiling thin-wall part model, are excavated and are efficiently used model
The information such as the connection semanteme in variant region and structure semantics support face in the thin-wall part model with complicated muscle feature structure
It efficiently and accurately extracts, reaches and improve mid-plane model formation efficiency, geometric accuracy, topology availability, applicability and reusability
Purpose.
Face abstracting method in thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features, including：
The definition of step 1 hierarchical structure semanteme：Define a new stratification language for being used for thin-wall part model and mid-plane model
Adopted structure supports subsequent operation to include model segmentation, the generation of middle face and the reuse of middle face etc.；
Efficient identification and the model segmentation of step 2 muscle feature：It identifies muscle feature and builds the hierarchical semantic knot based on muscle feature
Structure, it is then special according to the bottom-up successively segmentation muscle of hierarchical semantic structure of model using a kind of new model dividing method
Sign, to obtain the subregion with different levels structure semantics；
Face is extracted in step 3 hierarchical semantic：The muscle feature being partitioned into step 2 is extracted using adaptive plus mixing middle face
Mode, according to the bottom-up efficient and accurate extraction for carrying out mid-plane model of model hierarchy semantic structure；
Step 4 mid-plane model is generated and reused：Each Discrete facet is spliced by a complete layer according to hierarchical semantic information
Secondaryization semanteme mid-plane model, and based on this support model variant design when variable domain middle face quickly generate, in raising
The reusability of surface model.
The present invention proposes face abstracting method in a kind of thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features.It is first
First, define a hierarchical semantic structure based on muscle feature, for support subsequent operation for example model segmentation, middle face generation with
And middle face reuse etc..Then, identify muscle feature and structure the hierarchical semantic structure based on muscle feature after, carry out model segmentation with
Obtain the subregion with different levels structure semantics.Then, according to model hierarchy semantic information, face offset is respectively adopted
The accurate extraction in face in different zones is realized with two methods of discrete point fitting；Finally, according to hierarchical semantic information will respectively from
Scattered dough sheet is spliced into a complete hierarchical semantic mid-plane model, and based on this support model variant design when fluctuation zone
The middle face result in domain quickly generates, and reaches and improves mid-plane model formation efficiency, geometric accuracy, topology availability, applicability and can
The purpose of reusability.
Preferably, step 1 middle-levelization structure semantics include two aspects：Connection semanteme and Layer semantics, mid-plane model
Connection semanteme be a simple two tuples relation, represent the connection relation of different zones, two types can be classified as：L-type
Connection and T-shaped connection；L-type connection is mainly shown as connection relation system at the same level, and T-shaped connection is then that the support between some face faces is closed
System, contains Layer semantics information, wherein stress surface part should be the last layer grade of force part；Layer semantics are described as two
The membership between subregion is connected, two kinds of relations can be divided into：Relation and relationship between superior and subordinate at the same level；Company between muscle feature at the same level
The relation of connecing can be divided into 3 kinds：Independence, weak rigidity relation and strong incidence relation；The dependence of the superior and the subordinate's muscle feature can be divided into 5
Kind：Simple dependence relies primarily on relation, complicated dependence, mixed dependence relation and multi-Dependency Relationship.
The present invention proposes the efficient identification of muscle feature and model dividing method, constructs the stratification language based on muscle feature
Adopted structure, and model segmentation is carried out accordingly to obtain the subregion with different levels structure semantics.
Preferably, the efficient identification method and step of muscle feature includes：
The generation of step 2-1-1 faces group：Homonymy side has identical bottom set in one muscle feature, and face group is exactly apart from phase
The near set of the face with homonymy geometry, the generation of face group are made of four steps：
（1）Surface geometry uniformity identifies；
（2）The curved surface that distance in potential face group set is more than threshold value is rejected in preliminary screening；
（3）The curved surface that normal direction is inconsistent in the group geometry of face is rejected in postsearch screening；
（4）Face collection to find builds a complete topology boundary；
The lookup of step 2-1-2 faces group pair：After face group is searched successfully, model topology structure just from original complex structure be converted to by
A series of relatively simple topological structure of face group compositions.With it is existing in the method compared with search, the lookup of face group pair
After the curved surface in original method to be exactly changed to newly-generated face group, before recycling in face of lookup method, based on distance,
Normal direction and overlapping rule search face group pair；
Muscle feature recognitions of the step 2-1-3 based on face group pair：Here use based on the method for figure to identify muscle feature, the first step is
Structure face group adjacent map is then based on FGAG and realizes muscle feature recognition.
Preferably, model dividing method step includes：
Step 2-2-1 lower floors muscle Image Segmentation Methods Based on Features：Lower floor's muscle Image Segmentation Methods Based on Features, the first step are realized using the method operated based on Euler
It is that detection belongs to muscle feature powder divisional plane from model, then rebuilds muscle feature and model rest part；
Step 2-2-2 is the same as layer muscle Image Segmentation Methods Based on Features：Include two steps with layer muscle Image Segmentation Methods Based on Features：When with upper sub-regions coupling part
Segmentation, belongs to complete parttion, using the identical method of lower floor's muscle Image Segmentation Methods Based on Features；Second is that with the redundancy between layer muscle feature point
It cuts；With, there may be overlapping relation, when a muscle intersects with other muscle, being divided into naturally by other muscle between layer muscle feature
Multiple sub- muscle regions, at this moment muscle side be also divided into multiple sub- sides, it is necessary to merge into a side, it is special convenient for subsequent child muscle
Sign merges.Muscle feature sides reconstruction process includes two steps：
（1）Using common low layer geometry as new face geometry；
（2）The border in new face is formed using all outer boundary sides.
The present invention proposes face abstracting method in a kind of hierarchical semantic, and the muscle feature being partitioned into step 2 is used certainly
It adapts to plus mode is extracted in the middle face of mixing, the efficient and smart of mid-plane model is carried out according to model hierarchy semantic structure is bottom-up
Really extract；Segmentation subregion generally comprises muscle feature and main basic subdomain；Muscle feature is typically expressed as plane or simple two
Secondary curved surface is suitble to use face in the offset operation realization of face to extract to ensure geometric accuracy and formation efficiency.And main basic sub-district
Domain geometry is increasingly complex, is more suitable for, using discrete point approximating method, first discrete model surface, finding both sides corresponding points,
Median point cloud is calculated, then median point cloud is fitted to plane.
The present invention proposes effective method for reusing of mid-plane model, is spliced each Discrete facet according to hierarchical semantic information
Into after a complete hierarchical semantic mid-plane model, the hierarchical structure semantic information based on structure, when to initial model into
During row part modification, existing middle face results model can be efficiently used, greatly improves mid-plane model formation efficiency.
Preferably, the reuse step of middle face results model includes：
Step 4-2-1 identification model variable domains：When local variation occurs for thin-wall part model, become according to model geometric and topology
Change, variable domain is identified, including direct variable domain and indirect variable domain；
Step 4-2-2 removes face dough sheet in the correlation in direct variable domain；Identify to remove it is related in direct variable domain
Middle face dough sheet；By the connection semantics recognition of middle face dough sheet and remove associated edge fit.
Step 4-2-3 regenerates face in direct variable domain, and is added in mid-plane model.
The present invention proposes face abstracting method in a kind of thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features, supports
The efficiently and accurately extraction in face in thin-wall part model with complicated muscle feature structure；Hierarchical semantic structure based on muscle feature
The middle face dough sheet of splicing can ensure the topology availability of mid-plane model, while can also be positioned according to the design alteration of source model
Variable domain, face is reached and is improved mid-plane model formation efficiency, several with face in quick local updating in being extracted only for variable domain
What precision, topology availability, the purpose of applicability and reusability.
Description of the drawings
Fig. 1 is that flow chart is extracted in the middle face based on muscle feature recognition；
Fig. 2 is L-type connection and T-shaped connection figure in mid-plane model；
Fig. 3 is a typical thin-wall part model hierarchy structure chart；
Fig. 4 is 3 kinds of connection figures of muscle feature at the same level；
5 kind connection figures of the Fig. 5 between the superior and the subordinate's muscle feature；
Fig. 6 is face group generating process figure；
Fig. 7 is two different model decomposing method figures；
Fig. 8 is three kinds of different types of lower floor's muscle Image Segmentation Methods Based on Features mode figures；
Fig. 9 is muscle feature sides reconstruction process figure；
Figure 10 efficiently produces figure for the middle face after model local variation；
Figure 11 is muscle recognition methods performance analysis chart；
Figure 12 is face automatic generation method flow chart in the hierarchical semantic based on muscle Image Segmentation Methods Based on Features；
Figure 13 is complex thin-walled member model and corresponding stratification result mid-plane model figure.
Specific embodiment
It is low for the middle face extraction efficiency of the automobile with complicated muscle feature structure, aviation covering model and can not solve
The quick self-adapted replacement problem of mid-plane model after certainly initial model localized design changes proposes a kind of based on muscle Image Segmentation Methods Based on Features
Face abstracting method in hierarchical semantic supports the efficiently and accurately extraction in face in the thin-wall part model with complicated muscle feature.It is fixed
After the hierarchical semantic structure of adopted muscle feature, identification model muscle feature and the hierarchical semantic structure of muscle feature is built first, led to
It crosses model segmentation and obtains the subregion with different levels structure semantics.Then, according to model hierarchy semantic information, difference
The accurate extraction in face in different zones is realized using two methods of face offset and discrete point fitting.Finally according to the level of muscle feature
Change semantic information and each Discrete facet is spliced into a complete hierarchical semantic mid-plane model, and support model based on this
The middle face results model of variable domain quickly generates during variant design, and then improves mid-plane model formation efficiency, geometry essence
Degree, topology availability, the purpose of applicability and reusability.Holistic approach realizes that flow chart is as shown in Figure 1.
In Fig. 1,（a）Archetype is represented, is operated by the efficient identification of muscle feature,（b）The muscle feature of identification is marked,
（c）The Layer semantics division result of pattern reinforcing rib is given,（d）、（e）、（f）Face extraction process in representational level semanteme,
Wherein（d）Be in body region face extract as a result,（f）Green represent top layer muscle feature in face extract as a result,（e）Middle red table
Show that result is extracted in face in two level muscle feature.
The present invention is as follows：
Step 1, the hierarchical semantic structure of thin-wall part model and mid-plane model is defined
（1）Connect semantic definition
Connection semanteme in mid-plane model is a simple two tuples relation, represents the connection relation of different subregions, can return
Class is two types：L-type connects and T-shaped connection（Or the T-shaped connection of extension）.L-type connection is mainly shown as connection relation at the same level
System, T-shaped connection are then the supporting relations between some face faces, contain Layer semantics information, wherein stress surface part, which should be, applies
The last layer grade of power part, it is specific as shown in Figure 2.
（2）The definition of Layer semantics
Layer semantics describe the membership between two connection subregions, can be divided into two kinds of relations：Relation at the same level and the superior and the subordinate
Relation.Connection relation between muscle feature at the same level can be divided into 3 kinds：Independence, weak rigidity relation and strong incidence relation.The superior and the subordinate's muscle
The dependence of feature can be divided into 5 kinds：Simple dependence, rely primarily on relation, complicated dependence, mixed dependence relation and
Multi-Dependency Relationship.Fig. 3 gives a typical thin-wall part model hierarchy structure, and Fig. 4 and Fig. 5 represent that muscle at the same level is special respectively
5 kinds of connections between 3 kinds of connections of sign and the superior and the subordinate's muscle feature.
Efficient identification and the model segmentation of step 2 muscle feature
The efficient identification of muscle feature
The generation of first face group：Specific generating process four steps as shown in fig. 6, be mainly made of：（a）Surface geometry uniformity is known
Not；（b）The curved surface that distance in potential face group set is more than threshold value is rejected in preliminary screening；（c）Face group geometry is rejected in postsearch screening
The inconsistent curved surface of middle normal direction；（d）Face collection to find builds a complete topology boundary.
The lookup of second face group pair：The lookup of face group pair is exactly that the curved surface in original method is changed to newly-generated face group
Afterwards, lookup method is faced before recycling, face group pair is searched based on distance, normal direction and overlapping rule.
The 3rd muscle feature recognition based on face group pair：Here use based on the method for figure to identify muscle feature.The first step is
Structure face group adjacent map is then based on FGAG and realizes muscle feature recognition.
Model is split
Model segmentation is generally divided into two classes：Complete parttion and redundancy segmentation, the former by boolean operation progressively decomposition model until
Remainder can not decompose.In this case, all subregions are all independent to be connected between subregion only by face, such as
Fig. 7（b）It is shown.The latter is according to certain specific rule come decomposition model, and subregion is such as schemed there may be redundant area after decomposition
7（c）It is shown.Due to being to exist largely to intersect shared region between muscle and muscle, here using redundancy dividing method implementation model
Segmentation.
Model realizes hierarchical structure by muscle identification method and divides, and the subregion of different levels, which contains, to interdepend
Connection semantic relation.This connection semanteme is that a kind of bottom-up connection is semantic, that is, sub-regions is descended to rely on straton
Region exists, such as lower floor's muscle creates to strengthen upper strata muscle structure.At this point, lower floor's muscle is an independent individual, it can
With separated directly from model, and upper strata muscle is not an independent individual that can arbitrarily change, but a set
The presence of form, it is impossible to directly split.Therefore, the segmentation order of model should be bottom-up, introduce lower floor first below
Muscle Image Segmentation Methods Based on Features, then introduce same layer muscle Image Segmentation Methods Based on Features.
First lower floor's muscle Image Segmentation Methods Based on Features：Lower floor's muscle Image Segmentation Methods Based on Features is realized using the method operated based on Euler.The first step
It is that detection belongs to muscle feature powder divisional plane from model, then rebuilds muscle feature and model rest part.Wherein muscle feature and master
Body subregion can be divided into 3 kinds of lower floor's muscle Image Segmentation Methods Based on Features modes, as shown in Figure 8 there are many kinds of connection mode.
Second the same as layer muscle Image Segmentation Methods Based on Features：Include two steps with layer muscle Image Segmentation Methods Based on Features：When with upper sub-regions coupling part
Segmentation, belongs to complete parttion, using the identical method of lower floor's muscle Image Segmentation Methods Based on Features；Second is that with the redundancy between layer muscle feature point
It cuts.With, there may be overlapping relation, when a muscle intersects with other muscle, being divided into naturally by other muscle between layer muscle feature
Multiple sub- muscle regions, at this moment muscle side be also divided into multiple sub- sides, it is necessary to merge into a side, it is special convenient for subsequent child muscle
Sign merges.Muscle feature sides reconstruction process includes two steps：（1）Using common low layer geometry as new face geometry；（2）Using institute
By outer boundary side form the border in new face, detailed process is as shown in Figure 9.
Face is extracted in step 3 hierarchical semantic
Mode is extracted using adaptive plus mixing middle face to the muscle feature being partitioned into step 2, it is semantic according to model hierarchyization
The bottom-up efficient and accurate extraction for carrying out mid-plane model of structure；Segmentation subregion generally comprises muscle feature and main son substantially
Region.Muscle feature is typically expressed as plane or simple quadratic surface, is suitble to use face in the offset operation realization of face to extract to ensure
Geometric accuracy and formation efficiency.And main basic subdomain geometry is increasingly complex, is more suitable for using discrete point approximating method,
Both sides corresponding points are found on discrete model surface first, calculate median point cloud, then median point cloud is fitted to plane.
Step 4, mid-plane model generation and reuse
After each Discrete facet is spliced into a complete hierarchical semantic mid-plane model according to hierarchical semantic information, based on structure
The hierarchical structure semantic information built when carrying out local modification to initial model, can efficiently use existing middle face results model,
Greatly improve mid-plane model formation efficiency.Middle face efficiently produces procedure chart, specific steps bag after Figure 10 represents model local variation
It includes：
First identification model variable domain：When local variation occurs for thin-wall part model, primary work is to determine the sub-district of variation
Domain.Figure 10（a）And Figure 10（b）An initial thin-wall part model and corresponding mid-plane model is set forth, when part occurs for model
During variation, two muscle become large-sized, such as Figure 10（c）Shown, variable domain is divided into two classes：Direct variable domain and indirect fluctuation zone
Domain.Here green represents that two muscle features belong to direct variable domain, and orange areas belongs to impacted indirect variable domain.
Second, remove face dough sheet in the correlation in direct variable domain.First, identify and remove in direct variable domain
Face dough sheet in correlation.Then, by the connection semantics recognition of middle face dough sheet and associated edge fit, Figure 10 are removed（d）It is shown.
3rd, face in direct variable domain is regenerated, and is added in mid-plane model, face result during final change changes
Model such as Figure 10（f）It is shown.
Figure 11 gives muscle recognition methods performance evaluation, and Figure 11 is test model, includes a main part and some letters
List muscle feature, wherein ' n ' a sub- muscle feature forms a complete muscle characteristic pattern 11 and corresponds to typical muscle characteristic recognition method and institute
The calculating time of proposition method corresponds to, and as seen from the figure, with the raising of model complexity, the method proposed is in recognition time
On have a clear superiority.
Exemplified by Figure 12 mono- has the Thin-walled Workpiece model of simple muscle feature, its layer based on muscle Image Segmentation Methods Based on Features is given
Face abstracting method flow chart during secondaryization is semantic.Figure 12（a）It is initial thin-wall part model, the first step is that muscle is identified from initial model
Feature, and according to muscle Image Segmentation Methods Based on Features model, archetype is organized as multiple subregions to generate hierarchical structure, such as Figure 12
（b）Shown, wherein different colours represent the different levels of structure.Then, different middle face extraction sides is used as the case may be
Method obtains stratification mid-plane model, such as Figure 12（c）Shown, this results model can aid in subsequent FEA.After FEA is completed,
Analyst may be found that the problem of model and change model, these usual variations are appeared in muscle feature, such as Figure 12（d）It is shown,
Some muscle features change, Figure 12（e）Involved area caused by variation is marked.Finally, Figure 12（f）Give model change
After dynamic, the intermediate result model that quickly generates.
In order to further verify the applicability of proposition method, Figure 13 give a complex thin-walled member model as an example,
Figure 13（a）It is initial thin-wall part model, wherein including many reinforcing rib features with complicated connection relation, Figure 13（b）It is pair
Face results model in the stratification answered, the dough sheet of different levels are identified using different colours.Here, face uses in main part
NURBS approximating methods are realized, and face is extracted using face offset method in other muscle features.Compared to existing method, institute's extracting method is taken out
Take it is more efficient, especially the middle face of muscle characteristic area extract precision higher.
The efficiency that extracting method face in complicated muscle characteristic model is extracted in order to verify, uses two methods respectively：It is based on
In the complex thin-walled member model of virtual dividing in face abstracting method and thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features
Face abstracting method is that face extraction in progress on thin-wall part model is tested at 5, and time such as table 1 are extracted in model master data and middle face
It is shown.As seen from the table, more in model muscle feature quantity, the middle face extraction efficiency of institute's extracting method is higher, fully demonstrates
High efficiency of institute's extracting method in complicated muscle characteristic model in face extraction problem.
The efficiency comparative of face abstracting method in 1 two kinds of table
Claims (8)
1. face abstracting method in the thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features, which is characterized in that including：
The definition of step 1 hierarchical structure semanteme：
A new hierarchical semantic structure for thin-wall part model and mid-plane model is defined, subsequent operation is supported to include model
Segmentation, the generation of middle face and the reuse of middle face etc.；
Efficient identification and the model segmentation of step 2 muscle feature：
It identifies muscle feature and builds the hierarchical semantic structure based on muscle feature, then pressed using a kind of new model dividing method
According to the bottom-up son successively split muscle feature, there are different levels structure semantics with acquisition of hierarchical semantic structure of model
Region；
Face is extracted in step 3 hierarchical semantic：
Mode is extracted using adaptive plus mixing middle face to the muscle feature being partitioned into step 2, it is semantic according to model hierarchyization
The bottom-up efficient and accurate extraction for carrying out mid-plane model of structure；
Step 4：Mid-plane model is generated and reused：
Each Discrete facet is spliced by a complete hierarchical semantic mid-plane model according to hierarchical semantic information, and as
The middle face that infrastructural support variation sets timing variation region quickly generates, and improves the reusability of mid-plane model.
2. face abstracting method in the thin-wall part model hierarchy semanteme according to claim 1 based on muscle Image Segmentation Methods Based on Features,
It is characterized in that, step 1 specifically includes：
The definition of step 1-1 connection semantemes：
Connection semanteme in mid-plane model is a simple two tuples relation, represents the connection relation of different subregions, can return
Class is two types：L-type connects and T-shaped connection；
The definition of step 1-2 Layer semantics：Layer semantics describe the membership between two connection subregions, can be divided into two kinds
Relation：Relation and relationship between superior and subordinate at the same level.
3. face abstracting method in the thin-wall part model hierarchy semanteme according to claim 1 based on muscle Image Segmentation Methods Based on Features,
It is characterized in that, step 2 specifically includes：
The efficient identification of step 2-1 muscle features：Face group information identification muscle feature based on thin-wall part model；
Step 2-2 models are split：Model segmentation is typically simplified model topology most effective way, is generally divided into two classes：Completely
Segmentation and redundancy segmentation, largely intersect shared region due to existing between muscle and muscle, realize mould using redundancy partitioning scheme here
Type is split.
4. face abstracting method in the thin-wall part model hierarchy semanteme according to claim 3 based on muscle Image Segmentation Methods Based on Features,
It is characterized in that, step 2-1 includes the following steps：
The generation of step 2-1-1 faces group, homonymy side has identical bottom set in a muscle feature, and face group is exactly apart from phase
The near set of the face with homonymy geometry；
The generation of face group is made of four steps：
（1）Surface geometry uniformity identifies；
（2）The curved surface that distance in potential face group set is more than threshold value is rejected in preliminary screening；
（3）The curved surface that normal direction is inconsistent in the group geometry of face is rejected in postsearch screening；
（4）Face collection to find builds a complete topology boundary；
The lookup of step 2-1-2 faces group pair：
After face group is searched successfully, model topology structure is just converted to a series of more letter being made of face groups from original complex structure
Single topological structure；With it is existing in the method compared with search, the lookup of face group pair is exactly by the curved surface in original method
After being changed to newly-generated face group, lookup method is faced before recycling, face group is searched based on distance, normal direction and overlapping rule
It is right；
Muscle feature recognitions of the step 2-1-3 based on face group pair：Using muscle feature is identified based on the method for figure, the first step is structure
Face group adjacent map is then based on FGAG and realizes muscle feature recognition.
5. face abstracting method in the thin-wall part model hierarchy semanteme according to claim 3 based on muscle Image Segmentation Methods Based on Features,
It is characterized in that, step 2-2 includes the following steps：
Step 2-2-1 lower floors muscle Image Segmentation Methods Based on Features：Lower floor's muscle Image Segmentation Methods Based on Features, the first step are realized using the method operated based on Euler
It is that detection belongs to muscle feature powder divisional plane from model, then rebuilds muscle feature and model rest part；
Step 2-2-2 is the same as layer muscle Image Segmentation Methods Based on Features：
Include two steps with layer muscle Image Segmentation Methods Based on Features：First, the segmentation with upper sub-regions coupling part, belongs to complete parttion, under
The identical method of layer muscle Image Segmentation Methods Based on Features；Second is that with the redundancy segmentation between layer muscle feature.
6. face abstracting method in the thin-wall part model hierarchy semanteme according to claim 1 based on muscle Image Segmentation Methods Based on Features,
It is characterized in that, segmentation subregion generally comprises muscle feature and main basic subdomain, distinct methods need to be used to realize different zones
The accurate extraction in middle face, step 3 specifically include：
It takes out in the middle face for the muscle characteristic area that step 3-1 is usually made of using the realization of face offset operation plane or simple quadratic surface
It takes, to ensure geometric accuracy and formation efficiency；
Step 3-2 is using the increasingly complex main basic subdomain of discrete point process of fitting treatment geometry；
Both sides corresponding points are found on discrete point approximating method discrete model surface first, calculate median point cloud, then median point cloud is intended
Synthesize plane.
7. face abstracting method in the thin-wall part model hierarchy semanteme according to claim 1 based on muscle Image Segmentation Methods Based on Features,
It is characterized in that, step 4 specifically includes：
Each Discrete facet is spliced into the hierarchical semantic mid-plane model of a completion according to hierarchical semantic information by step 4-1；
The reuse of face result during step 4-2 is realized based on stratified semantic model, during support model variant design in face result
It quickly generates.
8. face abstracting method in the thin-wall part model hierarchy semanteme according to claim 7 based on muscle Image Segmentation Methods Based on Features,
It is characterized in that, step 4-2 includes the following steps：
Step 4-2-1 identification model variable domains：
When local variation occurs for thin-wall part model, according to model geometric and change in topology, variable domain is identified, including directly becoming
Dynamic region and indirect variable domain；
Step 4-2-2 removes face dough sheet in the correlation in direct variable domain, identify to remove it is related in direct variable domain
Middle face dough sheet by the connection semantics recognition of middle face dough sheet and removes associated edge fit；
Step 4-2-3 regenerates face in direct variable domain, and is added in mid-plane model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711145923.6A CN108108513A (en) | 2017-11-17 | 2017-11-17 | Face abstracting method in thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711145923.6A CN108108513A (en) | 2017-11-17 | 2017-11-17 | Face abstracting method in thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108108513A true CN108108513A (en) | 2018-06-01 |
Family
ID=62206434
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711145923.6A Pending CN108108513A (en) | 2017-11-17 | 2017-11-17 | Face abstracting method in thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108108513A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110263426A (en) * | 2019-06-18 | 2019-09-20 | 珠海格力电器股份有限公司 | Medium surface extraction method and device of thin-wall structure model and storage medium |
| CN114741753A (en) * | 2022-03-28 | 2022-07-12 | 大连理工大学 | Thin-wall reinforcement structure optimization method and device, computer and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090271156A1 (en) * | 2008-04-28 | 2009-10-29 | Canon Kabushiki Kaisha | Apparatus and method for generating analysis model |
| CN106326573A (en) * | 2016-08-26 | 2017-01-11 | 武汉船用机械有限责任公司 | Design method of suspension arm of crane |
| CN106874545A (en) * | 2017-01-05 | 2017-06-20 | 武汉船用机械有限责任公司 | A kind of method for designing of flat board with holes |
-
2017
- 2017-11-17 CN CN201711145923.6A patent/CN108108513A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090271156A1 (en) * | 2008-04-28 | 2009-10-29 | Canon Kabushiki Kaisha | Apparatus and method for generating analysis model |
| CN106326573A (en) * | 2016-08-26 | 2017-01-11 | 武汉船用机械有限责任公司 | Design method of suspension arm of crane |
| CN106874545A (en) * | 2017-01-05 | 2017-06-20 | 武汉船用机械有限责任公司 | A kind of method for designing of flat board with holes |
Non-Patent Citations (1)
| Title |
|---|
| HUAWEI ZHU ET AL.: "Automatic hierarchical mid-surface abstraction of thin-walled model based on rib decomposition", 《ADVANCES IN ENGINEERING SOFTWARE》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110263426A (en) * | 2019-06-18 | 2019-09-20 | 珠海格力电器股份有限公司 | Medium surface extraction method and device of thin-wall structure model and storage medium |
| CN110263426B (en) * | 2019-06-18 | 2020-12-04 | 珠海格力电器股份有限公司 | Medium surface extraction method and device of thin-wall structure model and storage medium |
| CN114741753A (en) * | 2022-03-28 | 2022-07-12 | 大连理工大学 | Thin-wall reinforcement structure optimization method and device, computer and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101807308B (en) | Three-dimensional model segmenting device and method | |
| CN109118500A (en) | A kind of dividing method of the Point Cloud Data from Three Dimension Laser Scanning based on image | |
| CN103106265B (en) | Similar image sorting technique and system | |
| CN101197045A (en) | Image solid matching method and device thereof | |
| CN106803253A (en) | Rock 3-D view crack extract method | |
| CN104299008A (en) | Vehicle type classification method based on multi-feature fusion | |
| CN104992143B (en) | A kind of Chinese-character stroke extraction method towards vector font | |
| CN109359355B (en) | Design implementation method of standard structure module | |
| CN104299009A (en) | Plate number character recognition method based on multi-feature fusion | |
| CN104966279A (en) | Image synthesis restoration method based on local structure features | |
| CN109726424A (en) | A kind of design implementation method of punching module | |
| CN108108513A (en) | Face abstracting method in thin-wall part model hierarchy semanteme based on muscle Image Segmentation Methods Based on Features | |
| Nnaji et al. | Feature reasoning for sheet metal components | |
| CN105427252A (en) | Mesh model multi-scale geometric detail repair method based on empirical mode decomposition | |
| CN107194402B (en) | Parallel refined skeleton extraction method | |
| CN104573722A (en) | Three-dimensional face race classifying device and method based on three-dimensional point cloud | |
| CN107862133A (en) | A kind of method of Mass production bridge pier BIM models | |
| CN109388875A (en) | A kind of design implementation method of elastic element module | |
| Qian et al. | Spatial contextual noise removal for post classification smoothing of remotely sensed images | |
| CN107908880A (en) | Face abstracting method in the complex thin-walled member model of guarantor's topology based on virtual dividing | |
| CN105653622A (en) | PRN data integrating method and system | |
| CN101101668A (en) | Medical image cutting method based on oscillatory network | |
| CN102426708B (en) | Texture design and synthesis method based on element reorganization | |
| Takashima et al. | Recognition of free-form features for finite element meshing using deep learning | |
| CN105139452B (en) | A kind of Geologic Curve method for reconstructing based on image segmentation |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180601 |