CN106446306A - Gauss sphere cluster based machine part reverse engineering modeling method - Google Patents

Gauss sphere cluster based machine part reverse engineering modeling method Download PDF

Info

Publication number
CN106446306A
CN106446306A CN201510470966.6A CN201510470966A CN106446306A CN 106446306 A CN106446306 A CN 106446306A CN 201510470966 A CN201510470966 A CN 201510470966A CN 106446306 A CN106446306 A CN 106446306A
Authority
CN
China
Prior art keywords
grid
gaussian sphere
sub
sphere
cluster
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN201510470966.6A
Other languages
Chinese (zh)
Inventor
丁展
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.)
Jinling Institute of Technology
Original Assignee
Jinling Institute of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jinling Institute of Technology filed Critical Jinling Institute of Technology
Priority to CN201510470966.6A priority Critical patent/CN106446306A/en
Publication of CN106446306A publication Critical patent/CN106446306A/en
Pending legal-status Critical Current

Links

Abstract

The invention discloses a reverse engineering modeling method for machine part data by using Gauss sphere cluster. The method includes the steps: allowing a user to scan a machine part into a three-dimensional point cloud by using a three-dimensional scanner; automatically converting the point cloud into a grid through a grid triangularization method; and performing rough decomposition on a whole grid model through a main curvature such that each secondary curved surface of the part independently falls into a relevant sub-zone. The method performs type detection on a sub-grid through a Gauss sphere cluster method, performs zone subdividing and recombination, performs secondary curved surface fitting on each subdivided sub-grid through a nonlinear least square method and a faithful distance, and cuts and sews a curved surface model so as to establish a digital entity model of the machine part. The user can perform parameter editing or curved surface deformation on the entity model so as to innovate a reverse design of the model. The modeling method is simple and efficient, can avoid user interference, and can automatically or semi-automatically complete reverse design modeling of the machine part.

Description

A kind of machine components reverse-engineering modeling method based on Gaussian sphere cluster
Technical field
The present invention relates to the computer Modeling Reverse Engineering field of machine components, more particularly, to a kind of machine based on Gaussian sphere cluster Tool part reverse-engineering modeling method.
Background technology
It is frequently encountered by plane, sphere, cylinder and the conical surface in the geometric shape design of the industrial products such as machine components, plastic Deng quadratic surface, a free-revving engine of reverse-engineering is exactly quickly and accurately to reduce these characteristic surfaces.It is true that big portion The quadratic surface in part is divided not to be self-existent, so the task first of quadratic surface identification is by region segmentation, that is, The data point belonging to these characteristic surfaces is isolated from measurement data.Quickly and accurately in locating element, quadratic surface region is One pith of reverse-engineering.
Region segmentation method can be generalized two classes, and a class is overall segmentation automatically, and another kind of is manual partial segmentation.From Dynamic overall region dividing method refers to put cloud by certain geological information (such as curvature, the geometrical invariants such as method arrow) or grid is drawn It is divided into multiple characteristic areas.And manual partial method refers to that user specifies seed point or seed region manually, by calculating certain Operator constantly expands feature quadratic surface region, and till the condition of characteristic area is unsatisfactory for, the method finally once generates one Individual characteristic area.At present, curved surface automated regional dividing method is mainly realized by curvature, that is, have curvature similarity and The continuous region in space is typically considered same characteristic area, and uses different curvature (as Gaussian curvature, principal curvatures, to put down All curvature), discomposing effect is also different.
Reverse engineering design being carried out to machine components, if directly carrying out surface fitting using the grid that curvature marks off, existing Problems with.(1) what curvature division methods can not be completely clean marks off characteristic area, if so directly entered to zoning Row Quadratic Surface Fitting, then fitting result and actual Surface Parameters can produce relatively large deviation, be unfavorable for that very much follow-up quadratic surface is joined Number constraint, or even be likely to cause part rejection.For example when plane and the face of cylinder are tangent, curvature would generally partly be justified when dividing Post stress and strain model is to plane grid region, or by part planar stress and strain model to cylindrical lattice region, thus results in surface fitting Error, further results in two curved surfaces simulating and can not meet tangent requirement.(2) measurement result of spatial digitizer is often held It is vulnerable to the impact of vibration, illumination and measured piece surface degree of roughness, the curvature value going out from grid computing suffers from this The impact of a little noise datas, causes numerical value unstable, affects the result of subsequent meshes region division further.
In order to solve above curvature partition problem, in succession curvature partitioning algorithm and related estimation method are optimized both at home and abroad, But all of curvature optimized algorithm all can only be directed to department pattern, therefore limitation is larger.
Content of the invention
For the problem that with existing reverse-engineering modeling method, machine components are carried out with presence during Model Reconstruction, the present invention solves Technical problem is:Propose a kind of method based on Gaussian sphere cluster, for detecting, segmenting after solution and the decomposition of reaggregation principal curvatures Feature quadratic surface subregion so that the division result in grid search-engine region is quite accurate, region fitting result precision pole Height, new method makes the reversal design mathematical model result of machine components more accurate.
The present invention proposes a kind of machine components reverse-engineering modeling method based on Gaussian sphere cluster, and technical scheme is as follows:
(1), when initially, true machine components are scanned into three-dimensional point cloud model with three-dimensional laser scanner by user.
(2) utilize mesh triangles method, point cloud model is converted into grid model.
(3) utilize the discrete principal curvatures of triangular plate, rough segmentation solution is carried out to full wafer grid, target is to make every in part secondary song What face was all independent falls in related subregion.
(4) using Gaussian sphere clustering method, type detection is carried out to sub-grid, and carry out region clustering, subdivision and restructuring simultaneously.
(5) using nonlinear least square method and loyal distance, plane, cylinder, circular cone and ball are carried out to each subdivision sub-grid Face matching.
(6) utilizing entity kernel-reduce and suture function, edlin being entered to surface model, thus being created that the digital entities of part Model.
(7) physical model can also enter line parameter editor or curved surface deformation, thus realizing the innovative design of reverse-engineering.
Brief description
The present invention is further described with reference to the accompanying drawings and examples.
Fig. 1 is the whole modeling method flow chart of the present invention.
Fig. 2 is the thick decomposition result schematic diagram to grid for the utilization principal curvatures of the present invention.
Fig. 3 is the plane grid Gaussian sphere feature schematic diagram of the summary of the present invention.
Fig. 4 is the cylindrical lattice Gaussian sphere feature schematic diagram of the summary of the present invention.
Fig. 5 is the circular cone grid Gaussian sphere feature schematic diagram of the summary of the present invention.
Fig. 6 is the Grid Gaussian sphere feature schematic diagram of the summary of the present invention.
Fig. 7 is that the Gaussian sphere of the plane class sub-grid of the present invention is clustered into a schematic diagram.
Fig. 8 is that the Gaussian sphere of the cylinder class sub-grid of the present invention is clustered into the schematic diagram of plane.
Specific embodiment
In conjunction with the accompanying drawings, the present invention is further detailed explanation.These accompanying drawings are practical operation schematic diagram, thus aobvious Show the composition relevant with the present invention.
As shown in figure 1, the invention provides a kind of machine components reverse-engineering modeling method based on Gaussian sphere cluster, method Flow process is, first with laser 3 d scanner, machine components are scanned into point cloud model, will put cloud mould followed by Triangulation Algorithm Type is automatically converted into grid model, then using the discrete principal curvatures of tri patch whole grid is carried out with rough segmentation solution, described net Lattice rough segmentation solution, is using watershed algorithm, the principal curvatures of tri patch to be clustered, thus mesh segmentation is become some sub-grids; Target be make every quadratic surface in part all independent fall in related subregion, followed by Gaussian sphere clustering method pair Sub-grid carries out type detection and screening, and carries out sub-grid region segmentation and restructuring simultaneously, recycles nonlinear least square method With loyal distance, Quadratic Surface Fitting is carried out to each subdivision sub-grid, then again cutting and suture are carried out to surface model, thus It is created that the digital entities model of part, user can also enter line parameter editor or curved surface is deformed thus innovating to physical model The reversal design of model.Cutting and suture are the surface trimming being carried using entity kernel and suture function by some independent two Secondary curved surface is finally merged into a digital physical model.
As shown in Fig. 2 the present invention, first with the discrete principal curvatures of grid surface, the grid of machine components is roughly divided Solution becomes some sub-grids.
Due to plane, cylinder, circular cone and Grid, in Gaussian sphere, characteristic is different, and therefore the present invention is using different The different Gaussian sphere of curved surface type carry out type screening and subdivision.Gaussian sphere refers to be mapped to the unit normal direction of any tri patch The center of circle be (0,0,0), radius is on 1 sphere, and this ball is referred to as Gaussian sphere, and Gaussian sphere cluster refer to using plane, cylinder, The normal direction feature of circular cone and ball, the triangular plate per unit system arrow in its sub-grid is clustered into point respectively, crosses the plane, no of the centre of sphere Cross plane and the generalized spherical of the centre of sphere, thus realizing substantially quadric type detection and screening.Quadratic surface, refers to plane (i.e. Generalized Quadratic curved surface), cylinder, circular cone and ball.
Plane can be by 1 point of space P and a unit normal vectorDetermine, that is, be expressed as binary geometric parameter Defined according to Gaussian sphere, the coordinate in the corresponding Gaussian sphere of any point in plane is exactlyAnd in the grid model of physical plane, Due to grid coarse due to, the point after plane grid is mapped to Gaussian sphere is collected in a point region rather than is strictly located at Same point, this characteristic is as shown in Figure 3.
Cylinder can be with a ternary geometric parameterRepresent.Its midpoint P is any point in cylindrical axis,It is Axis unit vector, r is cylindrical radius.Obviously, on the face of cylinder normal vector of any point all with axis vectorVertically, according to Gaussian sphere defines, and on the face of cylinder, the Gauss spherical coordinates of any point were all located at initial point and normal direction isPlane on.This plane (n can be expressed asx,ny,nz,0).More definite, on the strict unit circle positioned at this plane of these points.And in reality Cylindrical lattice model in, coarse due to grid due to, approximate being gathered in of the coordinate in Gaussian sphere that usual all-pair is answered It is generally aligned in the same plane in one annular section rather than strictly.This annular section can be synthesized to a least square faceD is approximately 0.This characteristic is as shown in Figure 4.
Circular cone can be with a ternary geometric parameterRepresent.Its midpoint P is conical point,Axis unit to Amount,It is the 1/2 of circular cone subtended angle.It is not difficult to release, the method arrow of any point and axis on taper seatAngle be definite value: Defined according to Gaussian sphere, on circular cone, the Gauss spherical coordinates of any point were all located at pointAnd normal vector is Plane on.This plane can be expressed asMore definite, strict the putting down positioned at this of these points On face, radius isCircle on.Due to grid noise, the approximate gathering of the Gauss spherical coordinates that actual circular cone grid is mapped It is generally aligned in the same plane in an annular section rather than strictly, this region generally can be synthesized to planeThis Characteristic is as shown in Figure 5.
With unlike plane, cylinder, the conical surface, there is not vector component in the geometric parameter expression formula of sphere, it can be with table It is shown as binaryparameter<P,r>.Its midpoint P is the centre of sphere, and r is the radius of a ball.Obviously, any point P on sphereiMethod swear and be So according to the definition of Gaussian sphere, on sphere, the Gauss spherical coordinates of any two points are different from, more definite, these points are tight Lattice be filled in whole Gaussian sphere.This characteristic is as shown in Figure 6.
Because principal curvatures divides division characteristic area that can not be completely clean, if so directly subregion is carried out secondary Surface fitting, then fitting result can be with the curved surface larger error of generation in real data, and this is unfavorable for follow-up restriction on the parameters.And Gaussian sphere can be very good the region that cannot decompose in grid rough segmentation solution is subdivided, and can be with the differentiation grid of robust Quadratic surface type.The present invention with the addition of type sorting procedure on the basis of Gaussian sphere mapping, and cluster operation is arranged at net After lattice rough segmentation solution, before Quadratic Surface Fitting.The purpose of cluster is just to discriminate between quadratic surface type, deletes coarse sub-grid Noise region in region, segments adjacent grid subregion again, and transmits curved surface type information to Quadratic Surface Fitting module With the spatial sub-area being actually needed after the refinement of matching.Quadratic Surface Fitting, is after obtaining grid surface type, using non- Linear least square and loyal distance carry out Quadratic Surface Fitting to the summit of each subdivision sub-grid.
As shown in fig. 7, utilizing plane Gaussian sphere characteristic, the per unit system arrow of sub-grid is clustered in Gaussian sphere one by the present invention Individual, thus judging that net region is plane.Fig. 7 .a is initial Gaussian spherical model (corresponding net region such as Fig. 7 .f).Figure 7.b, 7.c, 7.d and 7.e, be through once, secondary, three times, the Gaussian sphere after four points cluster, corresponding grid chart is figure 7.g-7.j.
As shown in figure 8, using the Gaussian sphere characteristic on the face of cylinder, the per unit system arrow of sub-grid is clustered in Gaussian sphere the present invention A plane, and this plane is through the centre of sphere, thus judging that net region is cylinder.Fig. 8 .a is how first initial Gaussian ball is The plane (corresponding net region such as Fig. 8 .f) of beginning matching.Fig. 8 .b, Fig. 8 .c, Fig. 8 .d be through once, secondary and three times poly- Gaussian sphere after class, corresponding grid chart is Fig. 8 .g-8.j.Fig. 8 .e is the face of cylinder finally simulating.
The present invention, can extremely efficiently, simply become precision by a machine components reversal design by above-mentioned shown method High CAD Digitized surface model.

Claims (7)

1. a kind of machine components reverse-engineering modeling method based on Gaussian sphere cluster it is characterised in that
Comprise the following steps:
S1:Using laser 3 d scanner, machine components are scanned into point cloud model;
S2:Using Triangulation Algorithm, point cloud model is automatically converted into grid model;
S3:Then using curvature, rough segmentation solution is carried out to whole grid, target is all independent the falling in phase of every quadratic surface making in part In the subregion closing;
S4:Followed by Gaussian sphere clustering method, type detection and screening are carried out to sub-grid, and carry out simultaneously sub-grid region segmentation and Restructuring;
S5:Using nonlinear least square method and loyal distance, Quadratic Surface Fitting is carried out to each subdivision sub-grid;
S6:Then again cutting and suture are carried out to surface model, thus being created that the digital entities model of part.
2. a kind of machine components reverse-engineering modeling method based on Gaussian sphere cluster according to claim 1 it is characterised in that: Described curvature, refers to the discrete principal curvatures of tri patch.
3. a kind of machine components reverse-engineering modeling method based on Gaussian sphere cluster according to claim 1 it is characterised in that: Described grid rough segmentation solution, is using watershed algorithm, the principal curvatures of tri patch to be clustered, thus by mesh segmentation Cheng Ruo Dry sub-grid.
4. a kind of machine components reverse-engineering modeling method based on Gaussian sphere cluster according to claim 1 it is characterised in that: Described Gaussian sphere clustering method, Gaussian sphere refers to that the unit normal direction of any tri patch is mapped to the center of circle is (0,0,0), half Footpath is on 1 sphere, and this ball is referred to as Gaussian sphere, and Gaussian sphere cluster refers to the method using plane, cylinder, circular cone and ball To feature, the plane of the triangular plate per unit system arrow in its sub-grid is clustered into point respectively, crossing the centre of sphere but the plane of the centre of sphere with And generalized spherical, thus realize substantially quadric type detection and screening.
5. a kind of machine components reverse-engineering modeling method based on Gaussian sphere cluster according to claim 1 it is characterised in that: Described quadratic surface, refers to plane, cylinder, circular cone and ball.
6. a kind of machine components reverse-engineering modeling method based on Gaussian sphere cluster according to claim 1 it is characterised in that: Described Quadratic Surface Fitting, is after obtaining grid surface type, using nonlinear least square method and loyal distance to each The summit of subdivision sub-grid carries out Quadratic Surface Fitting.
7. a kind of machine components reverse-engineering modeling method based on Gaussian sphere cluster according to claim 1 it is characterised in that: Described cutting and suture, are will be final for some independent quadratic surfaces with suture function using the surface trimming that entity kernel carries It is merged into a digital physical model.
CN201510470966.6A 2015-08-04 2015-08-04 Gauss sphere cluster based machine part reverse engineering modeling method Pending CN106446306A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510470966.6A CN106446306A (en) 2015-08-04 2015-08-04 Gauss sphere cluster based machine part reverse engineering modeling method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510470966.6A CN106446306A (en) 2015-08-04 2015-08-04 Gauss sphere cluster based machine part reverse engineering modeling method

Publications (1)

Publication Number Publication Date
CN106446306A true CN106446306A (en) 2017-02-22

Family

ID=59216509

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510470966.6A Pending CN106446306A (en) 2015-08-04 2015-08-04 Gauss sphere cluster based machine part reverse engineering modeling method

Country Status (1)

Country Link
CN (1) CN106446306A (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107273639A (en) * 2017-07-10 2017-10-20 海南中新塑料有限公司 A kind of method for the air inlet pipe that utilization reverse-engineering is built
CN110223378A (en) * 2019-05-28 2019-09-10 中国科学院自动化研究所 Quadratic surface primitive extracting method, system, device based on hierarchical clustering
CN112182718A (en) * 2020-10-10 2021-01-05 中国建筑第二工程局有限公司 Component design method and device based on BIM and 3D technology and storage medium
CN112017290B (en) * 2020-10-27 2021-02-19 杭州汇萃智能科技有限公司 Conical surface fitting method and device, computer equipment and storage medium
CN112465755A (en) * 2020-11-18 2021-03-09 熵智科技(深圳)有限公司 Initial sub-area subdivision method and device, computer equipment and storage medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101071451A (en) * 2006-05-09 2007-11-14 乐必峰软件公司 System and method for mesh and body hybrid modeling using 3d scan data
CN101257858A (en) * 2005-07-15 2008-09-03 美特瑞牙科有限公司 Method for (semi-) automatic dental implant planning
CN104504693A (en) * 2014-12-16 2015-04-08 佛山市诺威科技有限公司 Neck-edge line extraction method based on simple crown prosthesis mesh model of false tooth
CN104700398A (en) * 2014-12-31 2015-06-10 西安理工大学 Point cloud scene object extracting method

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101257858A (en) * 2005-07-15 2008-09-03 美特瑞牙科有限公司 Method for (semi-) automatic dental implant planning
CN101071451A (en) * 2006-05-09 2007-11-14 乐必峰软件公司 System and method for mesh and body hybrid modeling using 3d scan data
CN104504693A (en) * 2014-12-16 2015-04-08 佛山市诺威科技有限公司 Neck-edge line extraction method based on simple crown prosthesis mesh model of false tooth
CN104700398A (en) * 2014-12-31 2015-06-10 西安理工大学 Point cloud scene object extracting method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
丁展等: "基于Gauss Ball+的二次曲面细分解与识别", 《计算机辅助设计与图形学学报》 *
史阳: "三维空间索引结构-LABB树的研究", 《硕士学位论文》 *

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107273639A (en) * 2017-07-10 2017-10-20 海南中新塑料有限公司 A kind of method for the air inlet pipe that utilization reverse-engineering is built
CN110223378A (en) * 2019-05-28 2019-09-10 中国科学院自动化研究所 Quadratic surface primitive extracting method, system, device based on hierarchical clustering
CN112182718A (en) * 2020-10-10 2021-01-05 中国建筑第二工程局有限公司 Component design method and device based on BIM and 3D technology and storage medium
CN112182718B (en) * 2020-10-10 2021-06-15 中国建筑第二工程局有限公司 Component design method and device based on BIM and 3D technology and storage medium
CN112017290B (en) * 2020-10-27 2021-02-19 杭州汇萃智能科技有限公司 Conical surface fitting method and device, computer equipment and storage medium
CN112465755A (en) * 2020-11-18 2021-03-09 熵智科技(深圳)有限公司 Initial sub-area subdivision method and device, computer equipment and storage medium
CN112465755B (en) * 2020-11-18 2021-09-10 熵智科技(深圳)有限公司 Initial sub-area subdivision method and device, computer equipment and storage medium

Similar Documents

Publication Publication Date Title
CN106446306A (en) Gauss sphere cluster based machine part reverse engineering modeling method
Pound et al. Automated recovery of three-dimensional models of plant shoots from multiple color images
CN101105396B (en) System and method for automatic 3D scan data alignment
CN106199557B (en) A kind of airborne laser radar data vegetation extracting method
CN104573705B (en) A kind of clustering method of building Point Cloud of Laser Scanner
CN105513127A (en) Rod-shaped object regular three-dimensional modeling method and rod-shaped object regular three-dimensional modeling system based on density peak clustering
CN102306396A (en) Three-dimensional entity model surface finite element mesh automatic generation method
CN102938161B (en) A kind of 3D shape automatic division method based on Mean Shift
CN104134234A (en) Full-automatic three-dimensional scene construction method based on single image
CN109118500A (en) A kind of dividing method of the Point Cloud Data from Three Dimension Laser Scanning based on image
Pound et al. A patch-based approach to 3D plant shoot phenotyping
CN101515168B (en) Method for generating ring-cutting roughing NC tool path of product triangular mesh model
CN105404898A (en) Loose-type point cloud data segmentation method and device
CN111062915A (en) Real-time steel pipe defect detection method based on improved YOLOv3 model
Zhang et al. Region segmentation and shape characterisation for tessellated CAD models
Xu et al. Machining feature recognition from in-process model of NC simulation
Men et al. Hue-assisted automatic registration of color point clouds
CN112070769A (en) Layered point cloud segmentation method based on DBSCAN
Álvarez et al. Junction assisted 3d pose retrieval of untextured 3d models in monocular images
Beirão et al. Convex-and solid-void models for analysis and classification of public spaces
Pound et al. Surface reconstruction of plant shoots from multiple views
CN109359680B (en) Explosion sillar automatic identification and lumpiness feature extracting method and device
CN105184854A (en) Quick modeling method for cloud achievement data of underground space scanning point
CN107358655B (en) Identification method of hemispherical surface and conical surface models based on discrete stationary wavelet transform
Ji et al. Point cloud segmentation for complex microsurfaces based on feature line fitting

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into 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: 20170222