CN106484511A - A kind of spectrum attitude moving method - Google Patents
A kind of spectrum attitude moving method Download PDFInfo
- Publication number
- CN106484511A CN106484511A CN201610872366.7A CN201610872366A CN106484511A CN 106484511 A CN106484511 A CN 106484511A CN 201610872366 A CN201610872366 A CN 201610872366A CN 106484511 A CN106484511 A CN 106484511A
- Authority
- CN
- China
- Prior art keywords
- attitude
- model
- grid
- spectrum
- source
- 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
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/48—Program initiating; Program switching, e.g. by interrupt
- G06F9/4806—Task transfer initiation or dispatching
- G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
- G06F9/485—Task life-cycle, e.g. stopping, restarting, resuming execution
- G06F9/4856—Task life-cycle, e.g. stopping, restarting, resuming execution resumption being on a different machine, e.g. task migration, virtual machine migration
- G06F9/4862—Task life-cycle, e.g. stopping, restarting, resuming execution resumption being on a different machine, e.g. task migration, virtual machine migration the task being a mobile agent, i.e. specifically designed to migrate
- G06F9/4875—Task life-cycle, e.g. stopping, restarting, resuming execution resumption being on a different machine, e.g. task migration, virtual machine migration the task being a mobile agent, i.e. specifically designed to migrate with migration policy, e.g. auction, contract negotiation
-
- 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
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of spectrum attitude moving method, including the reference model, the Laplacian Matrix of source model and its corresponding spectrum and the characteristic function that calculate input, obtain the mediation base of reference model and source model, and calculate the Laplacian Matrix controlling grid;Choose the feature corresponding point between reference model and source model, calculate respective function between reference model and source model, optimize the mediation base of reference model and source model using respective function;Calculate source model with regard to controlling the barycentric coodinates matrix of grid;The integrated sub-space technique based on generalized barycenter coordinate is in this energy function;Optimize energy function, the control grid after being deformed;Using barycentric coodinates matrix, calculate the vertex position of object module;Using layering spectrum attitude migration algorithm, the attitude of reference model is moved on source model;Carry out towards optimization processing, obtain object module.The present invention can obtain gratifying attitude migration effect.
Description
Technical field
The present invention relates to computer graphicss and three-dimensional animation production field are and in particular to a kind of compose attitude moving method.
Background technology
In computer animation, artist animation from a model move to need to spend substantial amounts of another model when
Between and energy, the development of mesh modeling (the Example-based mesh modeling) technology based on sample, is moving of animation
Move and provide a kind of conventional method.Deformation based on sample is that the existing animation of reuse model in computer animation provides one
Plant important means.As the deformation method based on sample, deformation migration (Deformation transfer) refers to, gives one
Two of model (reference model) different attitudes, extract the potential motion between this two attitudes as a geometric transformation, so
Another model (source model) is driven to do similar deformation using this geometric transformation afterwards.Usually require that attitude and the ginseng of source model
First attitude examining model is similar.
Expression cloning moves to another model facial expression from a model, and this is earliest deformation migration.Replicate
Being deformed on the grid of source of the grid of reference of two given different shapes (attitude), the annexation of source grid can be with reference net
Lattice are different. and they are expressed as the set of transformation matrix, the ginseng then specified by user the deformation between two grids of reference
Examine the correspondence between grid and source grid, these transformation matrixs are mapped on the grid of source, thus obtaining the target of a deformation
Grid.The semantic information of model is considered in deformation transition process.Ben-Chen etc. proposes geometric distortion moving method.They
Mapping (Variational Harmonic Maps) be in harmonious proportion using variation by controlling grid realization deformation migration.This algorithm can
Move for the attitude between the model of different expressions type (as tetrahedral grid, polygon soup (Polygon soup) etc.)
Move.Zhou etc. is generalized to deformation migrating technology on the complex model that any number of branches are constituted. based on the change controlling grid
Shape migration solves the deformation of a control grid first with the grid of reference, is then based on barycentric coodinates editor and carrys out the change of driving source grid
Shape.Accumulate operator (Intrinsic operator) reconstruct shape from interior, under Functional Mapping, they are become using shape difference
Shape migrates, by forcing the shape difference between object module and second reference model and source model and first reference model
Between shape difference identical solving object module.For second-rate grid, the cotangent weight that this algorithm is used
Laplace operator can produce negative value, and this will produce negative effect to the convergence optimizing energy function.In addition, this algorithm is made
Shape difference relies on the Functional Mapping between shape, and this leads to the quality of this algorithm to be equally subject to the shadow of Functional Mapping precision
Ring.It should be noted that for the deformation defining reference model, above-mentioned methodical input is required for reference model
Multiple with reference to attitudes.
Content of the invention
In order to overcome shortcoming and the deficiency of prior art presence, the present invention provides one kind to compose attitude moving method.
The present invention adopts the following technical scheme that:
A kind of spectrum attitude moving method, comprises the steps:
Step one calculates reference model, the Laplacian Matrix of source model and its corresponding spectrum and the characteristic function of input,
Obtain the mediation base of reference model and source model, and calculate the Laplacian Matrix controlling grid;
Step 2 chooses the feature corresponding point between reference model and source model, and it is right between reference model and source model to calculate
Answer function, optimize the mediation base of reference model and source model using respective function;Calculate source model M with regard to controlling the weight of grid C
Heart coordinates matrix Ψ;
Step 3 sets up the spectrum attitude migration energy function of details holding, the integrated subspace skill based on generalized barycenter coordinate
Art is in this energy function;Optimize energy function, the control grid after being deformedUsing barycentric coodinates matrix Ψ, calculate mesh
Mark modelVertex position
Step 4 is moved to the attitude of reference model on source model using layering spectrum attitude migration algorithm;
Step 5 is carried out towards optimization processing, obtains object module.
If two triangle gridding shapes M=with different annexations<V, E, F>With M '=<V ', E ', F '>, respectively
Source model and reference model, the control grid of source model M is designated as C.
Step 2 is specially:The described feature corresponding point chosen between reference model and source model, calculate reference model
The k- neighborhood indicator function based on area approximation and source model between is as respective function F between model and G;Optimize reference
The mediation base of model and source model, obtains coupling the accurate base { φ ' that is in harmonious proportioniAnd { φj};Calculate source model M with regard to its control grid
The barycentric coodinates matrix Ψ of C.
Described respective function F and G specifically obtain process and are:
Corresponding point are chosen to reference model3- neighborhood regionThe corresponding point of corresponding source modelSelect area
Approximate k- neighborhood regionK takes so that the triangle area of two corresponding regions and as approximately equalised as possible value are
Can;
F and G takes the indicator function of this two corresponding regions, that is,
5th, a kind of spectrum attitude moving method according to claim 1 is it is characterised in that described step 3 is specially:
S3.1 sets up the spectrum attitude migration energy function of details holding, and for reconstructing the deformed shape of M, this shape has
The overall attitude of M ' and the minutia of M;Described spectrum attitude migration energy function is as follows:
εPT=εLF+λεLC, whereinλ is power
Weight;
Wherein,For Laplce's energy,For low frequency energy
Flow function, εPT=εLF+λεLC. for attitude migration amount, λ is weight;
S3.2 is to reduce solution space scale, ensure stability of solution, the integrated sub-space technique based on generalized barycenter coordinate,
Obtain new energy function:Wherein λ1It is weight,Power
Weight,
Matrix,It is scaling constraint;
S3.3 minimizes this energy function, the control grid after being deformed
S3.4 utilizes barycentric coodinates matrix Ψ, calculates object moduleVertex position
Described layering spectrum attitude migration algorithm is specially:Obtain the layering knot of grid model using simple mesh segmentation method
Structure, is converted into secondary attitude the large scale attitude of regional area;In order to obtain permissible coupling quasi- mediation base, if office
The grid ratio in portion region is sparse, then it is carried out with one to segmenting twice;The attitude that localized region details of use feature keeps
Migration algorithm;If it is necessary, repeating this process, until obtaining gratifying attitude migration results.
Described towards optimization processing, specifically adopt frame to correct the direction of Local grid after deformation, and border put down
Sliding process.
Beneficial effects of the present invention:
(1) present invention layering spectrum attitude migration algorithm, for the grid mould of different shape difference attitude difference annexation
Type, this algorithm can obtain can obtaining deforming nature, error less attitude migration effect;
(2) inventive algorithm also highlights advantage in terms of the details holding of the seizure to reference model attitude and source model.
Brief description
Fig. 1 is the workflow diagram of the present invention.
Specific embodiment
With reference to embodiment and accompanying drawing, the present invention is described in further detail, but embodiments of the present invention are not
It is limited to this.
Embodiment
As shown in figure 1, a kind of spectrum attitude moving method, comprise the steps:
The calculating of the Laplacian Matrix of step one grid model etc., calculates reference model M ', the drawing of source model M of input
This matrix L ' of pula, L and its corresponding spectrum and characteristic function, obtain the mediation base of reference model M ' and source model M;Calculate and control
The Laplacian Matrix of grid C.
If two triangle gridding shapes M=with different annexations<V, E, F>With M '=< V ', E ', F ' >, difference
For source grid and the grid of reference, the control grid of source grid M is designated as C.
We to set up the respective function between simplest indicator function tectonic model using the method for interaction. consider
Feature respective function F and choosing of G the result that be in harmonious proportion base accurate to coupling can produce certain impact, and we correspond in manual selection
After point, corresponding point are chosen to the grid of reference3- neighborhood regionThe corresponding point of corresponding source gridSelect area
Approximate k- neighborhood region(herein test in k take from 3,4,5 the triangle area making two corresponding regions and
As approximately equalised as possible it is worth), then F and G takes the indicator function of this two corresponding regions, that is,
Step 2 optimizes the mediation base of reference model M ' and source model M, obtains coupling the accurate base { φ ' that is in harmonious proportioniAnd { φj};Meter
Calculate the barycentric coodinates matrix Ψ with regard to its control grid C for the source model M.
Step 3 sets up the spectrum attitude migration energy function of details holding, the integrated subspace skill based on generalized barycenter coordinate
Art is in this energy function;Optimize energy function, the control grid after being deformedUsing barycentric coodinates matrix Ψ, calculate mesh
Mark modelVertex position
Specifically include following steps:
S3.1 sets up the spectrum attitude migration energy function of details holding, and for reconstructing the deformed shape of M, this shape has
The overall attitude of M ' and the minutia of M;Described spectrum attitude migration energy function is as follows:
Assume that Φ and Φ ' is optimized coupling quasi- mediation base, and apex coordinate vector has expanded into following form:
Merge the low frequency coefficient { α ' using M 'i(i=0,1 ..., K) and M Laplce's coordinate
δ=LV=(δ0, δ1..., δ|V|-1)TTo reconstruct the targeted attitude of MWherein
Due to the attitude of low frequency coefficient encoding model, and object module requires the overall attitude with M ', and this shows target
The vertex vector of attitudeIn basic function φiProjection on (0≤i≤K) should beWherein
() represents inner product, and so, we obtain following low frequency energy function:
On the other hand, in order to keep Laplce's coordinate of source model M, we are defined as follows Laplce's energy
WhereinIt is object moduleLaplce's coordinate,Being direction according to new attitude to convert drawing of source model
Pula this coordinate δiTransformation matrix. therefore, we have following attitude to migrate energy
εPT=εLF+λεLC(6)
Energy term ε defined in formula (4)LFIt is a global restriction, minimize formula (6)
εPT=εLF+λεLCIn energy function εPTA dense linear system will be led to.
In order to reduce computation complexity and seek stability of solution, we introduce the sub-space technique based on barycentric coodinates to S3.2
Solved. set the control grid that grid C is grid model M, our target is to find a new attitude of C, is designated as
So that fromObtained using barycentric coodinates editor reconstructThe low frequency attitude of M ' can be imitated well. hereinafter, Wo Menye
Represent the vertex vector controlling grid with same mark.
If Ψ is M with regard to controlling the barycentric coodinates matrix of grid C, that is, V=Ψ C. once it is determined thatWe just can utilizeCalculateVertex vector.WithReplace in formula (4) and formula (5)Respectively obtain
With
Wherein subscript M represents that this energy is that summit new position to source grid M enters row constraint. in the technical program,
Low frequency number is taken as K=5.
Minimum formula (7) is combined the energy obtaining and can cause to the Planar Mechanisms controlling grid with formula (8), and this is easy to
Produce the result of distortion. therefore we are with keeping controlling the holding source grid in Laplce's coordinate replacement formula (8) of grid C
Laplce's coordinate of M, suppresses to control grid with thisDeformation, that is,:
WhereinWithIt is to control grid C and its deformation respectivelyLaplce's coordinate.
In order to prevent uniformly to scale, we constrain zoom factor, introduce one directly zoom factorConstraint
Penalty term for 1, that is,:
In this formulaCan also be expressed asThe function on middle summit, in formula (9)With formula (9)
InThis two can also be with controlling gridVertex position represent, concrete derive as follows:
Grid C is controlled to be deformed toWhen, vertex viConversion have following simple form:
Wherein s is zoom factor, other three element a, and b, c determine rotation. can be by solving following optimization problem
ParameterIt is expressed as controlling gridSummit function,
Wherein,WithIt isSummit,It is vertex vi1- neighborhood.
Note vertex viGeometric coordinate be (vi(x), vi(y), vi(z)), then have
Order
Wherein, ellipsis represents and takes all over all ofFormula (11) can be converted into following minimum
Change problem
min||Ai(S a b c)T-Bi||2.
Thus having
Therefore, the zoom factor s=U of the scaling constraint in formula (4)0Bi
BecauseThe Laplce's coordinates restriction controlling gridAs regular terms, obtain final energy letter
Number is as follows:
Wherein λ1With λ2It is weight, all items are all unknown control gridsSummit function, be not difficult to find out formula (12)
Energy is with regard to controlling gridCoordinate quadratic function, especially,
Wherein L is the Laplacian Matrix controlling grid C,WithRespectively be control grid C withLaplce's coordinate because Ψ and φiCan precalculate,
So the optimization problem of formula (4) is converted into a Linear least squares minimization problem. in our current experiments, in order to balance appearance
State study and details keep, weight λ1With λ2It is manual selection, such as λ1With λ2It is 0.05, or λ1=λ2=5 × 10-4,
If details has obvious deformation, increase this two weights, whereas if attitude learn not fully it has to reduce this two
Individual weight.By solution formula (12), it is possible to obtain have the low frequency attitude of reference model M ', the new mould of the details of source model M
Type
In experimentation, only make once it was noticed that carrying out the transmission of multistep low frequency using linear interpolation and can reducing
Force the distortion that low-frequency component brings, then when forcing low-frequency component identical, we are repeatedly divided using the method for linear interpolation
Step transmission, with t α 'i+(1-t)αiReplace α 'i, we obtain following energy function:
Wherein 0≤t≤1.
S3.3 minimizes this energy function, the control grid after being deformed
S3.4 utilizes barycentric coodinates matrix Ψ, calculates object moduleVertex position
Step 4 is moved to the attitude of reference model on source model using layering spectrum attitude migration algorithm;
In fact, the attitude of threedimensional model has multiple dimensioned property, and low frequency coefficient only encodes is the overall situation of model
Attitude, the spectrum attitude migration algorithm that details of use keep can only migrate the low frequency attitude of the entirety of reference model, and cannot be ginseng
The attitude examining other yardsticks of model moves on the grid of source.
In order to solve this problem, we adopt level attitude migration strategy, devise a simple layering attitude and move
Move framework, we utilize the low frequency coefficient of the overall situation first, and the overall attitude of migration reference model is on source model;Then obtain to new
To grid model split, so, the Small and Medium Sized attitude of overall attitude changes into low with respect to partial model
Frequency attitude. repeat this process, until the attitude of the different scale of reference model is hierarchically moved on source model be
Only.
In order to carry out layering spectrum attitude migration, after obtaining the target gridding of overall low frequency migration, we are to reference net
Lattice and target gridding are split accordingly, the partial model of the grid of reference that segmentation is obtained and the partial model of target gridding
Again respectively as reference model and source model, carry out the attitude migration of partial model.
More specifically it is assumed that M obtains after learning the low frequency attitude of M ' With P ' it is respectivelyRight with two in M '
The Local grid answered, we allow furtherP ' the attitude of study obtainsHere we determine attitude by the method for range estimation
Learn incomplete Local grid.
We utilize the different corresponding part of attitude in simple mesh segmentation interaction two models of Ground Split.
In experimentation, it was noticed that when grid than sparse when, the Local grid splitting is generally relatively thick
Rough, so that the accurate base that is in harmonious proportion of the coupling that cannot be able to use.For this reason, we adopt butterfly interpolation subdividing to excessively sparse office
Portion's grid is finely divided, the new point of insertion in subdivision templateDetermined by following formula:
We arrive twice than sparse Local grid subdivision one to these according to number of vertices. in an experiment, if top
Points are less than 500, and we just segment partial model twice, complete the appearance between reference model and the corresponding Local grid of source model
After state migration, we delete the summit of insertion so that new attitude has identical annexation with original mesh.
Using above mark it is assumed that the Local grid of the reference model M ' obtaining after segmentation is P ', in the migration of low frequency attitude
Target gridding afterwardsMiddle corresponding part isThen the intermediate frequency attitude of the overall situation with respect to grid model M ' is converted into Local grid
The low frequency attitude of P '.
Therefore, the attitude of P ' is moved to by we using spectrum attitude moving methodOn.
The method of similar overall situation attitude migration, we calculate first with reference to attitude P ' and source attitudeLaplacian MatrixAnd its corresponding spectrum and characteristic function;Calculating source attitudeControl grid(for convenience, during this, we make
The control grid being obtained with the first level is as source Local gridControl grid, be still designated as) La Pula operator. connect
Get off to choose partial model P ' withBetween corresponding point, and using these corresponding point area approximation k- neighborhood indicator function
As with reference to the respective function between attitude P ' and source attitude P, subsequently calculate and couple the accurate base that is in harmonious proportionWithCalculating source mould
TypeWith regard to controlling gridBarycentric coodinates matrixThen minimize following energy function:
Wherein,Force target griddingLow-frequency component with reference to attitude P ' identical,It is the low frequency coefficient with reference to attitude P ',Constraint controls the scaling of grid, λ3
With λ4It is weight, choosing method is similar to weight λ in overall situation low frequency attitude migration1With λ2.Solve the optimization minimizing formula (12)
Control grid after problem, after being deformedRecycleCalculate new localized target gridFinally handle
New localized target gridAdhere to the target gridding after the migration of low frequency attitudeUpper.
When being necessary, we can resolve into less Local grid these Local grids further, repeat on
State process, to obtain attitude study more fully target gridding model. that is this is the iterative process of a multilamellar.
Whole layering spectrum attitude migrates shown in the flow process accompanying drawing 1 of framework. and the overall low frequency attitude having executed the first level is moved
After shifting, respectively the left arm of the result of the first level, right arm and left lower limb have been carried out with the spectrum attitude migration of the second level. through two
After the attitude migration of level, we can obtain attitude and learn more sufficient target gridding, now, if it is also possible to deposited
Towards different problems, we can carry out one simply towards optimization processing.
Towards optimization processing, after multi-level spectrum attitude migrates, each source Local grid has been deformed into and reference
Local grid has the grid of similar attitude, but because comparatively low frequency and high frequency are, does not clearly define, in migration
Secondary attitude or details can be produced during low frequency certain affect, this lead to sometimes localized source grid direction may with corresponding
Reference partial model different.
In order to obtain more reasonably result, we adopt the simple method office adjacent to two from same layer
Portion's grid is carried out towards alignment. first, using (the i.e. equalization point drawn game of the equalization point of whole Local grid, partitioning boundary of 3 points
Apart from the point that border is farthest on portion's grid) respectively build a local frame in source Local grid with corresponding reference Local grid;
Then, the spin matrix between two local frames of calculating, and this spin matrix is applied to the source Local grid of deformation
On all summits, obtain towards identical target Local grid;Finally, the border vertices of the source Local grid being deformed by holding
Laplce's coordinate to smooth the internal vertex of source Local grid of the Laplce's coordinate before rotation and deformation after rotation
Borderline region.The present invention compared with prior art has the effect that:
There is the same object of different annexations:Same object (i.e. same object, though now attitude non-phase
With, but the geometry on surface be very close to) have between two grids of identical annexation and carry out attitude migration and be
Simplest situation.The base from second to the 6th of the such as Laplacian Matrix of the visualization grid of reference and source grid, because
Seem good enough for the correspondence between this two groups of bases, so, prior art directly can carry out attitude migration, institute using these bases
The result obtaining is successful on low frequency attitude migrates.The accurate base that is in harmonious proportion of the coupling of two models being used with our method
Functional matrix, from the results of view, the accurate correspondence being in harmonious proportion between base of coupling is also good enough, there is attitude and learn not in prior art
Sufficiently phenomenon. our attitude migration is more abundant in study in the attitude of head, tail and right rear leg.
There are the different objects of identical annexation:Have a case that the different objects of identical annexation are relatively conventional.
If the grid of reference and source grid have identical topological structure.In this example, the method for prior art uses not
There is the original Laplce's base through optimizing, due to not having the correspondence between the feature base optimizing poor, attitude migration
Effect is unsatisfactory, our attitude migration study head, tail and front right-leg result all relatively good it is noted that I
Result be through two levels spectrum attitude migration obtain, in second level, we respectively to tail, head,
Left back pawl and right fore paw have carried out the migration of spectrum attitude.
The same object of same object with different annexations refers to the grid of reference and source grid in geometry
On be similar, but athletic posture is different.
Here we provide two more generally experimental results to general row --- -- different object difference annexation, its
In the grid of reference be different object from source grid, and their annexation is different, respectively the result of prior art and I
Arithmetic result, the first level attitude migrate after, our method is respectively to the arm of source grid model, lower limb and head
Portion performs the spectrum attitude migration of the second level it can be seen that the result of our methods is in the study direction of head and two legs
Attitude is better than prior art result.
The Shape Editing of threedimensional model is the important content of Digital Geometry Processing.By Shape Editing technology so that using
Family more easily creates various new shapes and new attitude.Provided for model reusability based on the mesh modeling technology of sample
A kind of important channel.As the mesh modeling technology based on sample, between deformation migration is using grid of reference model difference attitude
Deformation carry out driving source grid model and carry out deformation.Different from deformation migration, attitude migration only inputs an appearance of reference model
State, due to relatively difficult to the definition of attitude, this makes to study the attitude transport between the different grid model of annexation
Challenging.This technical scheme is according to the rough appearance of the low-frequency component coding grid model of the Laplacian Matrix of grid model
State, and Laplce's coordinate of grid keeps model detail characteristic, merges both and proposes the spectrum attitude that a details keeps
Moving method, and define corresponding nonlinear optimization energy.Due to solve this nonlinear optimal problem lead to one dense linear
System, this technical scheme is integrated with the sub-space technique based on generalized barycenter coordinate, and this can not only substantially reduce the rule of solution space
Mould, and due to decrease the degree of freedom of deformation to a certain extent and so that stability of solution is guaranteed.Observe three-dimensional mould
Type attitude has multiple dimensioned property, can only process the large scale attitude of model using the low frequency coefficient of model, and in cannot taking into account
Yardstick attitude or the attitude of regional area, lead to attitude migration insufficient.This technical scheme regards one point as reference attitude
Rotating fields, are not the attitude of large scale script using Model Segmentation, i.e. Small and Medium Sized attitude changes into regional area
Large scale attitude.Then, the attitude moving method same to the attitude application of these regional areas.By being layered attitude migration energy
Enough the attitude of the different scale of reference model is moved on source model step by step, obtain gratifying attitude migration effect.
Above-described embodiment is the present invention preferably embodiment, but embodiments of the present invention are not subject to described embodiment
Limit, other any spirit without departing from the present invention and the change made under principle, modification, replacement, combine, simplify,
All should be equivalent substitute mode, be included within protection scope of the present invention.
Claims (7)
1. a kind of spectrum attitude moving method is it is characterised in that comprise the steps:
Step one calculates reference model, the Laplacian Matrix of source model and its corresponding spectrum and the characteristic function of input, obtains
Reference model and the mediation base of source model, and calculate the Laplacian Matrix controlling grid;
Step 2 chooses the feature corresponding point between reference model and source model, calculates corresponding letter between reference model and source model
Number, optimizes the mediation base of reference model and source model using respective function;Calculate source model M with regard to controlling the center of gravity of grid C to sit
Mark matrix Ψ;
Step 3 sets up the spectrum attitude migration energy function of details holding, integrated is arrived based on the sub-space technique of generalized barycenter coordinate
In this energy function;Optimize energy function, the control grid after being deformedUsing barycentric coodinates matrix Ψ, calculate target
ModelVertex position
Step 4 is moved to the attitude of reference model on source model using layering spectrum attitude migration algorithm;
Step 5 is carried out towards optimization processing, obtains object module.
2. a kind of spectrum attitude moving method according to claim 1 it is characterised in that
If two triangle gridding shapes M=with different annexations<V, E, F>With M '=<V ', E ', F '>, respectively source mould
Type and reference model, the control grid of source model M is designated as C.
3. a kind of spectrum attitude moving method according to claim 1 is it is characterised in that step 2 is specially:Described choosing
Take the feature corresponding point between reference model and source model, calculate the k- based on area approximation between reference model and source model
Neighborhood indicator function is as respective function F between model and G;Optimize the mediation base of reference model and source model, coupled
Accurate base { the φ ' that is in harmonious proportioniAnd { φj};Calculate the barycentric coodinates matrix Ψ with regard to its control grid C for the source model M.
4. a kind of spectrum attitude moving method according to claim 3 is it is characterised in that described respective function F and G specifically obtain
Process be:
Corresponding point are chosen to reference model3- neighborhood regionThe corresponding point of corresponding source modelSelect area approximation
K- neighborhood regionK takes so that the triangle area of two corresponding regions and as approximately equalised be as possible worth;
F and G takes the indicator function of this two corresponding regions, that is,
5. a kind of spectrum attitude moving method according to claim 1 is it is characterised in that described step 3 is specially:
S3.1 sets up the spectrum attitude migration energy function of details holding, and for reconstructing the deformed shape of M, this shape has M's '
Overall attitude and the minutia of M;Described spectrum attitude migration energy function is as follows:
εPT=εLF+λεLC, whereinλ is weight;
Wherein,For Laplce's energy,For low frequency energy letter
Number, εPT=εLF+λεLC. for attitude migration amount, λ is weight;
S3.2 is to reduce solution space scale, ensure stability of solution, and the integrated son based on generalized barycenter coordinate is emptySquare
Battle array,It is scaling constraint;
S3.3 minimizes this energy function, the control grid after being deformed
S3.4 utilizes barycentric coodinates matrix Ψ, calculates object moduleVertex position
6. a kind of spectrum attitude moving method according to claim 1 is it is characterised in that attitude migration algorithm is composed in described layering
It is specially:Obtain the hierarchy of grid model using simple mesh segmentation method, secondary attitude is converted into regional area
Large scale attitude;In order to obtain permissible coupling quasi- mediation base, if the grid ratio of regional area is sparse, it is entered
Row one is to segmenting twice;The attitude migration algorithm that localized region details of use feature keeps;If it is necessary, repeating this mistake
Journey, until obtaining gratifying attitude migration results.
7. a kind of spectrum attitude moving method according to claim 1 it is characterised in that described towards optimization processing, specifically
Correct the direction of Local grid after deformation using frame, and border is smoothed.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610872366.7A CN106484511A (en) | 2016-09-30 | 2016-09-30 | A kind of spectrum attitude moving method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610872366.7A CN106484511A (en) | 2016-09-30 | 2016-09-30 | A kind of spectrum attitude moving method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106484511A true CN106484511A (en) | 2017-03-08 |
Family
ID=58268305
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610872366.7A Pending CN106484511A (en) | 2016-09-30 | 2016-09-30 | A kind of spectrum attitude moving method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106484511A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109859322A (en) * | 2019-01-22 | 2019-06-07 | 广西大学 | A kind of spectrum posture moving method based on deformation pattern |
CN111626218A (en) * | 2020-05-28 | 2020-09-04 | 腾讯科技(深圳)有限公司 | Image generation method, device and equipment based on artificial intelligence and storage medium |
CN112530016A (en) * | 2020-10-30 | 2021-03-19 | 北京字跳网络技术有限公司 | Method, device, equipment and storage medium for adsorbing road fittings |
CN113223152A (en) * | 2021-05-14 | 2021-08-06 | 浙江大学 | Method for automatic pose and wrinkle migration for three-dimensional garment model aesthetic display |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101944238A (en) * | 2010-09-27 | 2011-01-12 | 浙江大学 | Data driving face expression synthesis method based on Laplace transformation |
CN104616347A (en) * | 2015-01-05 | 2015-05-13 | 掌赢信息科技(上海)有限公司 | Expression migration method, electronic equipment and system |
CN105427360A (en) * | 2015-11-11 | 2016-03-23 | 华南理工大学 | Error-controllable CAGE sequence representation algorithm for dynamic grid |
-
2016
- 2016-09-30 CN CN201610872366.7A patent/CN106484511A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101944238A (en) * | 2010-09-27 | 2011-01-12 | 浙江大学 | Data driving face expression synthesis method based on Laplace transformation |
CN104616347A (en) * | 2015-01-05 | 2015-05-13 | 掌赢信息科技(上海)有限公司 | Expression migration method, electronic equipment and system |
CN105427360A (en) * | 2015-11-11 | 2016-03-23 | 华南理工大学 | Error-controllable CAGE sequence representation algorithm for dynamic grid |
Non-Patent Citations (3)
Title |
---|
MENGXIAO YIN等: ""Spectral pose transfer"", 《COMPUTER AIDED GEOMETRIC DESIGN》 * |
S. WUHRER等: ""Posture-invariant statistical shape analysis using Laplace operator"", 《COMPUT. GRAPH》 * |
李欧州: ""基于频谱变换的三维模型检索和编辑的研究与实现"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109859322A (en) * | 2019-01-22 | 2019-06-07 | 广西大学 | A kind of spectrum posture moving method based on deformation pattern |
CN109859322B (en) * | 2019-01-22 | 2022-12-06 | 广西大学 | Spectral attitude migration method based on deformation graph |
CN111626218A (en) * | 2020-05-28 | 2020-09-04 | 腾讯科技(深圳)有限公司 | Image generation method, device and equipment based on artificial intelligence and storage medium |
WO2021238595A1 (en) * | 2020-05-28 | 2021-12-02 | 腾讯科技(深圳)有限公司 | Image generation method and apparatus based on artificial intelligence, and device and storage medium |
CN111626218B (en) * | 2020-05-28 | 2023-12-26 | 腾讯科技(深圳)有限公司 | Image generation method, device, equipment and storage medium based on artificial intelligence |
CN112530016A (en) * | 2020-10-30 | 2021-03-19 | 北京字跳网络技术有限公司 | Method, device, equipment and storage medium for adsorbing road fittings |
CN113223152A (en) * | 2021-05-14 | 2021-08-06 | 浙江大学 | Method for automatic pose and wrinkle migration for three-dimensional garment model aesthetic display |
CN113223152B (en) * | 2021-05-14 | 2022-07-12 | 浙江大学 | Method for automatic pose and wrinkle migration for three-dimensional garment model aesthetic display |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Sumner et al. | Mesh-based inverse kinematics | |
CN104268934B (en) | Method for reconstructing three-dimensional curve face through point cloud | |
Ijiri et al. | Floral diagrams and inflorescences: interactive flower modeling using botanical structural constraints | |
CN106484511A (en) | A kind of spectrum attitude moving method | |
CN103714577B (en) | Three-dimensional model simplification method suitable for model with textures | |
CN103886603B (en) | Left ventricle nuclear magnetic resonance image segmentation and three-dimensional reconstruction method | |
CN105427360B (en) | A kind of controllable CAGE sequence expression algorithm of the error of dynamic grid | |
CN104183008B (en) | Shader classification method and device based on surface signal fitting and tessellation and graphics rendering method | |
CN109472870B (en) | Model matching method based on grid reconstruction and multi-influence-domain correction | |
Zhao et al. | IGA-based point cloud fitting using B-spline surfaces for reverse engineering | |
CN109544666A (en) | A kind of full automatic model deformation transmission method and system | |
CN111028335B (en) | Point cloud data block surface patch reconstruction method based on deep learning | |
CN110889893B (en) | Three-dimensional model representation method and system for expressing geometric details and complex topology | |
CN105006016A (en) | Component level three dimensional model building method of bayesian network constraint | |
CN108710906A (en) | Real-time point cloud model sorting technique based on lightweight network LightPointNet | |
Zollhöfer et al. | GPU based ARAP Deformation using Volumetric Lattices. | |
CN110619681A (en) | Human body geometric reconstruction method based on Euler field deformation constraint | |
CN101833785A (en) | Controllable dynamic shape interpolation method with physical third dimension | |
CN108230452B (en) | Model hole filling method based on texture synthesis | |
CN106682286A (en) | Geometric analysis method based functionally gradient material part modeling method | |
CN102339473B (en) | Genetic algorithm-based method for design of cartoon model | |
CN110757806A (en) | 3D printing-oriented double-type-shape close-laying method and system | |
Manson et al. | Hierarchical deformation of locally rigid meshes | |
CN109859322A (en) | A kind of spectrum posture moving method based on deformation pattern | |
CN114329320A (en) | Partial differential equation numerical solution method based on heuristic training data sampling |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170308 |
|
WD01 | Invention patent application deemed withdrawn after publication |