CN110288696A - A kind of method for building up of complete consistent organism three-dimensional feature characterization model - Google Patents
A kind of method for building up of complete consistent organism three-dimensional feature characterization model Download PDFInfo
- Publication number
- CN110288696A CN110288696A CN201910511671.7A CN201910511671A CN110288696A CN 110288696 A CN110288696 A CN 110288696A CN 201910511671 A CN201910511671 A CN 201910511671A CN 110288696 A CN110288696 A CN 110288696A
- Authority
- CN
- China
- Prior art keywords
- model
- dimensional
- organism
- global
- shape
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/30—Determination of transform parameters for the alignment of images, i.e. image registration
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Graphics (AREA)
- Geometry (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Processing Or Creating Images (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of method for building up of complete consistent organism three-dimensional feature characterization model, by combining global/part suture shape (GLoSS) and Skeletal Skinned algorithm (LBS) to create a new biological body three-dimensional models.First with global/part suture shape (GLoSS), the parameter in its model is registrated again later, the parameter that registration obtains is obtained final biological body three-dimensional models in conjunction with Skeletal Skinned algorithm (LBS).The present invention guarantees that 3D data have certain authenticity, and the consistent organism three-dimensional feature characterization model of foundation can guarantee the deformation of posture and form, and the 3D animal model precision obtained is higher.
Description
Technical field
The present invention relates to a kind of method for building up of complete consistent organism three-dimensional feature characterization model, belong to computer vision
Technical field.
Background technique
The detection of animal, tracking and threedimensional model are reconstituted in biology, ecology, agricultural and Entertainment industry
There are many applications, but current most research direction is the foundation of human 3d model, this is because first, phase
Than in the mankind, the type of animal is more, and differing greatly between each animal species, threedimensional model is difficult to set up；Second, with people
Class is compared, it is difficult to obtain the 3D data of animal, most of animal is in field, and 3-D scanning tool is inconvenient to use, and animal
Unlike the mankind can make the movement under various instructions to obtain 3D data.
At present the method for building up of animal threedimensional model first is that being obtained by the threedimensional model of 3-D scanning manual manufacture
Model of the data of animal to establish.But the data source that this mode obtains is limited, and it is a lack of authenticity.In addition have
By going out key point by handmarking on animal picture, the 3D model of animal is established by these key points.But this
A little methods all deposit that model accuracy is not high, and animal model species are few, obtain 3D data inconvenience etc. limitation.
Summary of the invention
Goal of the invention: in order to overcome the deficiencies in the prior art, the present invention provides a kind of complete consistent organism three
The method for building up of dimensional feature characterization model, the present invention uses based on global/part suture shape (GLoSS), then ties
Skeletal Skinned algorithm (LBS) is closed to establish a kind of complete consistent organism three-dimensional feature characterization model.
Technical solution: to achieve the above object, the technical solution adopted by the present invention are as follows:
A kind of method for building up of complete consistent organism three-dimensional feature characterization model, comprising the following steps:
Step S1. establishes global/part suture shape GLoSS；
Step S11. obtains the three-dimensional feature of animal, determines a three dimensional animals template net according to the three-dimensional feature of animal
Lattice, and really
Its fixed covering weight；
S12. the parameter in global/part suture shape GLoSS is determined: including local positioning Ii, local three-dimensional
Spin matrix ri∈R3×1, internal morphological parametersAnd attitude parameterπi={ Ii,ri,si,diIt is office
The variables set of portion i, i is the partial indexes of model and i ∈ (1...N), Π={ I, r, s, d } are the variables set of all parts, then
Three-dimensional model gridding apex coordinate vector calculation formula is as follows:
Wherein, IiFor positioning local in model, riFor the three-dimensional rotation matrix of part, siFor internal morphological parameters, di
For appearance
State parameter,For three-dimensional model gridding apex coordinate, R (ri) it is whole three-dimensional rotation matrix, niIt is model
Part
Vertex quantity,It is the coordinate points in local coordinate system, calculation formula is as follows:
pi=ti+mp,i+Bs,isi+Bp,idi
Wherein,Indicate the template of part,It is the vector of average attitude offsets,Generation
Table form excursion matrix,It is the matrix for determining posture deforming；
Step S2. is global/part suture shape GLoSS and 3-D scanning figure initial registration；
Step S3. combines Skeletal Skinned algorithm LBS to establish most on the basis of global/part suture shape GLoSS
Whole organism three-dimensional feature characterization model；
The standardization of step S31. attitude parameter；After the initial registration of step S2, first biological bodily form is constructed
Shape model；According to the organism pose parameter of global/part suture shape GLoSS estimation, calculated using linear Skeletal Skinned
All registration templates are unified into identical posture by method LBS, to global/part suture shape GLoSS mirror image grid
Afterwards, opposite vertexes are averaging processing, to obtain the registration under initial posture；In the registration under initial posture, lower jaw and tongue
Point linear model learn from three dimensional animals template mesh；Laplce's smoothly entire grid is finally used, to complete
The standardization of posture；It realizes under initial posture, the foundation of the statistical model of change in shape in Euclidean space；Calculate average shape and
Principal component can capture the shape difference between organism；
Step S32. determines final biological body three-dimensional models；
Skeletal Skinned algorithm is used first, and equation is It is vertex all in grid model, J table
Show the position of artis,It is attitude parameter, Ω indicates covering weight；In each pointIt is obtained after deformationBecome
Shape formula is as follows:
Wherein, ωk,iIt is the element in covering weight matrix Ω, represents influence system of the vertex i by k-th of artis
Number,For the attitude parameter under stationary state, θjCorrespond to the spin matrix of artis for part,For null vector, jjFor in J
Each three-dimensional vector for corresponding to single joint center j；
Secondly, obtaining the equation for determining biological volume morphing after according to the registration of step S2Biological volume morphing with
Attitude parameterBetween relation equationAnd the relation equation between artis and morphological parametersPosture is joined
Number standardization, later by calculating average form parameter and its main component, the shape difference captured between animal has come
At the foundation of final biological body three-dimensional models, finally obtained biology body three-dimensional models areWhereinFor form ginseng
Number,For attitude parameter, γ is global translation matrix, and the deformation formula of each point is as follows in three-dimensional model gridding:
Wherein,RespectivelyWithIn element.
It is preferred: three dimensional animals template mesh to be divided into 33 parts in step S11, each part is more than one respectively
Side shape.
It is preferred: to optimize global/part suture shape GLoSS in step S2 using gradient descent method, so that entirely
Office/part sutures shape GLoSS closer to 3-D scanning figure.
It is preferred: it uses ARAP Regularization Technique to carry out model-free to grid vertex in step S2 and is registrated to capture details,
It completes and 3-D scanning figure initial registration.
The method for building up of a kind of complete consistent organism three-dimensional feature characterization model of view-based access control model proposed by the present invention and existing
Some technologies are compared to advantage:
1. being allowed to and the animal mould from scanning museum based on global/part suture shape (GLoSS)
Pattern is originally mutually registrated, and can guarantee that 3D data have certain authenticity in this way.
2. combining global/part suture shape (GLoSS) and Skeletal Skinned algorithm (LBS), what is be built such that is consistent
Organism three-dimensional feature characterization model can guarantee the deformation of posture and form, and the 3D animal model precision obtained is higher.
Detailed description of the invention
Fig. 1 is complete consistent organism three-dimensional feature characterization model method for building up flow chart；
Fig. 2 is that global/part suture shape (GLoSS) is registrated flow chart；
Specific embodiment
In the following with reference to the drawings and specific embodiments, the present invention is furture elucidated, it should be understood that these examples are merely to illustrate this
It invents rather than limits the scope of the invention, after the present invention has been read, those skilled in the art are to of the invention various
The modification of equivalent form falls within the application range as defined in the appended claims.
A kind of method for building up of complete consistent organism three-dimensional feature characterization model, first with global/part suture shape
Shape model (GLoSS), is later again registrated the parameter in its model, and the parameter that registration is obtained is calculated in conjunction with Skeletal Skinned
Method (LBS) is to obtain final biological body three-dimensional models, as shown in Figure 1, specific implementation method of the invention is as follows:
Step 1: establishing global/part suture shape (GLoSS).
Specific method for building up is as follows:
(1) a three dimensional animals template mesh is defined, chooses the three-dimensional model gridding of a lion here, and defines its illiteracy
Skin weight.Later artificially this lion three-dimensional template grid dividing at 33 parts, each part is a polygon respectively.
(2) parameter in global/part suture shape (GLoSS) is determined:
Enabling i first is the partial indexes and i ∈ (1...N) of model.The parameter of this model includes local positioning Ii, local
Three-dimensional rotation matrix ri∈R3×1, internal morphological parametersAnd attitude parameterEnable πi={ Ii,ri,si,
diBe part i variables set, ∏={ I, r, s, d } for all parts variables set., then three-dimensional model gridding apex coordinate to
It is as follows to measure calculation formula:
Here niIt is the vertex quantity of model part.It is the coordinate points in local coordinate system, its calculation formula
It is as follows:
pi=ti+mp,i+Bs,isi+Bp,idi
Wherein,Indicate the template of part,It is the vector of average attitude offsets,Generation
Table form excursion matrix,It is the matrix for determining posture deforming.
Step 2: global/part suture shape (GLoSS) and 3-D scanning figure initial registration.Specific registration side
Method is divided into two steps:
(1) first using gradient descent method come Optimized model, this makes model closer to 3-D scanning figure
(2) ARAP Regularization Technique is then used to carry out model-free registration to grid vertex to capture details
Step 3: establishing final organism three-dimensional feature on the basis of global/part suture shape (GLoSS)
Characterization model, the specific steps are as follows:
(1) standardization of attitude parameter.By initial registration, first organism shape is constructed.According to GLoSS
All registration templates are unified into identical by the organism pose parameter of model estimation using linear Skeletal Skinned algorithm (LBS)
Posture.Finally obtained three dimensional biological volume mesh is asymmetric, this is because: the pose estimation of inaccuracy, linear hybrid
The limitation of covering, toy may not be symmetrical and the difference of body two sides causes different deformations.It is asked to solve this
Topic, after mirror image grid, opposite vertexes are averaging processing for we, to obtain the registration under initial posture.In addition, when animal closeds
When mouth, 3-D scanning can not observe the inside of mouth.In order to solve this problem, in registration, the point of lower jaw and tongue can
To be learnt from template with a simple linear model.Finally use Laplce's smoothly entire grid.To complete posture
Standardization.Posture standardization eliminates the non-linear effects of part rotation opposite vertexes.It realizes under initial posture, in Euclidean space
The foundation of the statistical model of change in shape.Average shape and principal component are calculated, the shape difference between organism can be captured.
(2) final biological body three-dimensional models are determined.
Skeletal Skinned algorithm is used first, and formula is as follows It is vertex all in grid model, J table
Show the position of artis,It is attitude parameter, Ω indicates covering weight.In each pointIt is obtained after deformationBecome
Shape formula is as follows
Wherein ωk,iIt is the element in covering weight matrix Ω, represents influence system of the vertex i by k-th of artis
Number.
Secondly, after according to the registration of step 2, the available equation for determining biological volume morphingThe biological bodily form
State and attitude parameterBetween relation equationAnd the relation equation between artis and morphological parameters
It further needs exist for standardizing attitude parameter, nonlinear influence is locally rotated on such vertex and is eliminated.It
We, which pass through, afterwards calculates average form parameter and its main component, these can capture the shape difference between animal and have come
At the foundation of final biological body three-dimensional models.It is finally obtained biology body three-dimensional models beWhereinFor form
Parameter,For attitude parameter, γ is global translation matrix., the deformation formula of each point is as follows in three-dimensional model gridding:
WhereinRespectivelyWithIn element.
In conclusion the present invention combines global/part suture shape (GLoSS) and Skeletal Skinned algorithm (LBS) wound
A new new biological body three-dimensional models out, suture shape (GLoSS) with global/part first, later again to its mould
Parameter in type is registrated, and the parameter that registration obtains is obtained final biology in conjunction with Skeletal Skinned algorithm (LBS)
Body three-dimensional models.
The above is only a preferred embodiment of the present invention, it should be pointed out that: for the ordinary skill people of the art
For member, various improvements and modifications may be made without departing from the principle of the present invention, these improvements and modifications are also answered
It is considered as protection scope of the present invention.
Claims (4)
1. a kind of method for building up of complete consistent organism three-dimensional feature characterization model, which comprises the following steps:
Step S1. establishes global/part suture shape GLoSS；
Step S11. obtains the three-dimensional feature of animal, determines a three dimensional animals template mesh according to the three-dimensional feature of animal, and
Determine its covering weight；
S12. the parameter in global/part suture shape GLoSS is determined: including local positioning Ii, local three-dimensional rotation square
Battle array ri∈R3×1, internal morphological parametersAnd attitude parameterπi={ Ii,ri,si,diBe part i change
Quantity set, i is the partial indexes of model and i ∈ (1...N), Π={ I, r, s, d } are the variables set of all parts, then threedimensional model
Mesh vertex coordinates vector calculation formula is as follows:
Wherein, IiFor positioning local in model, riFor the three-dimensional rotation matrix of part, siFor internal morphological parameters, diFor appearance
State parameter,For three-dimensional model gridding apex coordinate, R (ri) it is whole three-dimensional rotation matrix, niIt is the top of model part
Point quantity,It is the coordinate points in local coordinate system, calculation formula is as follows:
pi=ti+mp,i+Bs,isi+Bp,idi
Wherein,Indicate the template of part,It is the vector of average attitude offsets,Represent shape
State excursion matrix,It is the matrix for determining posture deforming；
Step S2. is global/part suture shape GLoSS and 3-D scanning figure initial registration；
Step S3. combines Skeletal Skinned algorithm LBS to establish finally on the basis of the overall situation/part suture shape GLoSS
Organism three-dimensional feature characterization model；
The standardization of step S31. attitude parameter；After the initial registration of step S2, first biological shape mould is constructed
Type；According to the organism pose parameter of global/part suture shape GLoSS estimation, using linear Skeletal Skinned algorithm
All registration templates are unified into identical posture by LBS, to global/part suture shape GLoSS mirror image grid
Afterwards, opposite vertexes are averaging processing, to obtain the registration under initial posture；In the registration under initial posture, lower jaw and tongue
Point linear model learn from three dimensional animals template mesh；Laplce's smoothly entire grid is finally used, to complete
The standardization of posture；It realizes under initial posture, the foundation of the statistical model of change in shape in Euclidean space；Calculate average shape and
Principal component can capture the shape difference between organism；
Step S32. determines final biological body three-dimensional models；
Skeletal Skinned algorithm is used first, and equation is It is vertex all in grid model, J indicates to close
The position of node,It is attitude parameter, Ω indicates covering weight；In each pointIt is obtained after deformationDeformation is public
Formula is as follows:
Wherein, ωk,iIt is the element in covering weight matrix Ω, represents influence coefficient of the vertex i by k-th of artis,
For the attitude parameter under stationary state, θjCorrespond to the spin matrix of artis for part,For null vector, jjIt is each right in J
It should be in the three-dimensional vector of single joint center j；
Secondly, obtaining the equation for determining biological volume morphing after according to the registration of step S2Biological volume morphing and posture
ParameterBetween relation equationAnd the relation equation between artis and morphological parametersTo attitude parameter mark
Standardization captures the shape difference between animal later by calculating average form parameter and its main component to complete most
The foundation of lifelong object dimensional model, finally obtained biology body three-dimensional models areWhereinFor morphological parameters,
For attitude parameter, γ is global translation matrix, and the deformation formula of each point is as follows in three-dimensional model gridding:
Wherein,RespectivelyWithIn element.
2. the method for building up of complete consistent organism three-dimensional feature characterization model according to claim 1, it is characterised in that: step
Three dimensional animals template mesh is divided into 33 parts in rapid S11, each part is a polygon respectively.
3. the method for building up of complete consistent organism three-dimensional feature characterization model according to claim 2, it is characterised in that: step
Optimize global/part suture shape GLoSS in rapid S2 using gradient descent method, so that global/part suture shape mould
Type GLoSS is closer to 3-D scanning figure.
4. the method for building up of complete consistent organism three-dimensional feature characterization model according to claim 3, it is characterised in that: step
It uses ARAP Regularization Technique to carry out model-free registration to grid vertex to capture details in rapid S2, completes at the beginning of with 3-D scanning figure
Begin registration.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910511671.7A CN110288696B (en) | 2019-06-13 | 2019-06-13 | Establishment method of complete and consistent organism three-dimensional characteristic representation model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910511671.7A CN110288696B (en) | 2019-06-13 | 2019-06-13 | Establishment method of complete and consistent organism three-dimensional characteristic representation model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110288696A true CN110288696A (en) | 2019-09-27 |
| CN110288696B CN110288696B (en) | 2023-01-06 |
Family
ID=68005270
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910511671.7A Active CN110288696B (en) | 2019-06-13 | 2019-06-13 | Establishment method of complete and consistent organism three-dimensional characteristic representation model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110288696B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112785712A (en) * | 2021-01-25 | 2021-05-11 | 新东方教育科技集团有限公司 | Three-dimensional model generation method and device and electronic equipment |
| CN113160418A (en) * | 2021-05-10 | 2021-07-23 | 上海商汤智能科技有限公司 | Three-dimensional reconstruction method, device and system, medium and computer equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102467753A (en) * | 2010-11-04 | 2012-05-23 | 中国科学院深圳先进技术研究院 | Method and system for reconstructing time-varying point cloud based on framework registration |
| CN105006016A (en) * | 2015-05-21 | 2015-10-28 | 北京航空航天大学 | Component level three dimensional model building method of bayesian network constraint |
| CN108320326A (en) * | 2018-01-12 | 2018-07-24 | 东南大学 | A kind of three-dimensional modeling method for human hand |
| CN108629294A (en) * | 2018-04-17 | 2018-10-09 | 华南理工大学 | Human body based on deformation pattern and face net template approximating method |
-
2019
- 2019-06-13 CN CN201910511671.7A patent/CN110288696B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102467753A (en) * | 2010-11-04 | 2012-05-23 | 中国科学院深圳先进技术研究院 | Method and system for reconstructing time-varying point cloud based on framework registration |
| CN105006016A (en) * | 2015-05-21 | 2015-10-28 | 北京航空航天大学 | Component level three dimensional model building method of bayesian network constraint |
| CN108320326A (en) * | 2018-01-12 | 2018-07-24 | 东南大学 | A kind of three-dimensional modeling method for human hand |
| CN108629294A (en) * | 2018-04-17 | 2018-10-09 | 华南理工大学 | Human body based on deformation pattern and face net template approximating method |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112785712A (en) * | 2021-01-25 | 2021-05-11 | 新东方教育科技集团有限公司 | Three-dimensional model generation method and device and electronic equipment |
| CN113160418A (en) * | 2021-05-10 | 2021-07-23 | 上海商汤智能科技有限公司 | Three-dimensional reconstruction method, device and system, medium and computer equipment |
| WO2022237249A1 (en) * | 2021-05-10 | 2022-11-17 | 上海商汤智能科技有限公司 | Three-dimensional reconstruction method, apparatus and system, medium, and computer device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110288696B (en) | 2023-01-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Yu et al. | Doublefusion: Real-time capture of human performances with inner body shapes from a single depth sensor | |
| Zuffi et al. | Lions and tigers and bears: Capturing non-rigid, 3d, articulated shape from images | |
| CN109584353B (en) | Method for reconstructing three-dimensional facial expression model based on monocular video | |
| CN102999942B (en) | Three-dimensional face reconstruction method | |
| WO2019219013A1 (en) | Three-dimensional reconstruction method and system for joint optimization of human body posture model and appearance model | |
| CN103733226B (en) | Quickly there is the tracking of joint motions | |
| CN102971768B (en) | Posture state estimation unit and posture state method of estimation | |
| CN113012282B (en) | Three-dimensional human body reconstruction method, device, equipment and storage medium | |
| CN107967463B (en) | Virtual face recognition method based on synthetic image and deep learning | |
| CN109697688A (en) | A kind of method and apparatus for image procossing | |
| US20100259538A1 (en) | Apparatus and method for generating facial animation | |
| CN106355147A (en) | Acquiring method and detecting method of live face head pose detection regression apparatus | |
| CN108629294A (en) | Human body based on deformation pattern and face net template approximating method | |
| WO2006049147A1 (en) | 3d shape estimation system and image generation system | |
| CN108363973A (en) | A kind of unconfined 3D expressions moving method | |
| CN108154104A (en) | A kind of estimation method of human posture based on depth image super-pixel union feature | |
| CN110288696A (en) | A kind of method for building up of complete consistent organism three-dimensional feature characterization model | |
| CN114422832A (en) | Anchor virtual image generation method and device | |
| CN112365589B (en) | Virtual three-dimensional scene display method, device and system | |
| CN106408654B (en) | A kind of creation method and system of three-dimensional map | |
| CN102682473A (en) | Virtual clothing real-time physical modeling method | |
| Hu et al. | A dense point-to-point alignment method for realistic 3D face morphing and animation | |
| Lee et al. | Noniterative 3D face reconstruction based on photometric stereo | |
| CN113112586B (en) | Real-time human body three-dimensional reconstruction method based on prior model | |
| CN115457171A (en) | Efficient expression migration method adopting base expression space transformation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |