CN103177473A - Instance-based large-scale scene composition method - Google Patents
Instance-based large-scale scene composition method Download PDFInfo
- Publication number
- CN103177473A CN103177473A CN2013100759559A CN201310075955A CN103177473A CN 103177473 A CN103177473 A CN 103177473A CN 2013100759559 A CN2013100759559 A CN 2013100759559A CN 201310075955 A CN201310075955 A CN 201310075955A CN 103177473 A CN103177473 A CN 103177473A
- Authority
- CN
- China
- Prior art keywords
- node
- model
- candidate
- label
- tectonic element
- 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
- 238000000034 method Methods 0.000 title claims abstract description 95
- 239000000203 mixture Substances 0.000 title abstract description 3
- 230000008569 process Effects 0.000 claims abstract description 62
- 238000010276 construction Methods 0.000 claims abstract description 27
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 12
- 230000005012 migration Effects 0.000 claims description 48
- 238000013508 migration Methods 0.000 claims description 48
- 238000010189 synthetic method Methods 0.000 claims description 14
- 230000036039 immunity Effects 0.000 claims description 10
- 238000005457 optimization Methods 0.000 claims description 10
- 230000001195 anabolic effect Effects 0.000 claims description 9
- 230000035772 mutation Effects 0.000 claims description 9
- 238000004458 analytical method Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000001514 detection method Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 2
- 238000010586 diagram Methods 0.000 description 5
- 230000015572 biosynthetic process Effects 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 230000004069 differentiation Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000013459 approach Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000005429 filling process Methods 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 230000002452 interceptive effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- PXFBZOLANLWPMH-UHFFFAOYSA-N 16-Epiaffinine Natural products C1C(C2=CC=CC=C2N2)=C2C(=O)CC2C(=CC)CN(C)C1C2CO PXFBZOLANLWPMH-UHFFFAOYSA-N 0.000 description 1
- 235000007926 Craterellus fallax Nutrition 0.000 description 1
- 240000007175 Datura inoxia Species 0.000 description 1
- 239000000427 antigen Substances 0.000 description 1
- 102000036639 antigens Human genes 0.000 description 1
- 108091007433 antigens Proteins 0.000 description 1
- 230000003542 behavioural effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 239000012141 concentrate Substances 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 239000000945 filler Substances 0.000 description 1
- 239000012467 final product Substances 0.000 description 1
- 239000004615 ingredient Substances 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000005309 stochastic process Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 230000003612 virological effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses an instance-based large-scale scene composition method. The method automatically optimizing and grouping structure units to establish a large-scale scene includes the steps of giving a small scene model, splitting the model by three equally spaced orthogonal plane sets to establish a structure unit set of the model, and extracting geometrical characteristics of the structure units and adjacency relation of the structure units; defining a Petri net structure to establish parametric representation in target model construction process according to expansion factors in three directions; applying the defined Petri net and Parametric representations thereof as coded representation of a target model; and optimizing the construction process by artificial immune algorithm so as to obtain a model the most approximate to the instance model structure. The scale is expanded by structural characteristics of the instant model so as to develop the large-scale scene model, modeling efficiency is increased, and model resources are reused effectively.
Description
Technical field
The present invention relates to a kind of large scale scene synthetic method based on example, belong to computer software and computer graphics techniques field, specifically, be a kind ofly to use scale that computing machine carries out the processes such as model analysis, inference modeling and optimal combination expansion instance models to generate the method for more massive model of place.
Background technology
In the fields such as digital entertainment, virtual reality, the necessary means that structure large-scale complex geometric model such as city, mountain region have become the horn of plenty user to experience, the fields such as industrial circle, digital legacy protection also strengthen day by day to the demand of three-dimensional model.Existing modeling method often adopts standard C AD modeling tool, and its operating process complexity causes the user need to be through the training of plenty of time, and modeling efficiency is low thereby be difficult to satisfy the needs of practical application area when building complex model.Therefore, introduce automatic or semi-automatic mechanism at the complex geometry modeling process and can the simplified user interactive burden become a kind of new technological trend.
Existing automatic Building modeling method comprises based on the method for reconstructing of scan-data and process modeling method.method for reconstructing based on scan-data comprises document 1:Nan L.L., Andrei S., Zhang H., Cohen-Or D., Chen B.Q., SmartBoxes for Interactive Urban Reconstruction, Proceedings SIGGRAPH2010, Article93 etc., these class methods need expensive equipment to carry out catching of data, and the processing of extensive cloud data has very high requirement to computer hardware, the ambiguousness that three dimensional point cloud is intrinsic and noise also have stronger impact to the quality of reconstruction model, thereby these class methods are not suitable for directly carrying out interactively scene modeling, process modeling method such as document 2:MullerP., Wonka P., Haegler, S., Ulmer, A., and Gool L.V., Procedural modeling of buildings.ACM Transaction of Graphics, the methods such as 25 (3) 614 – 623,2006. adopt inference method automatically to create complicated model according to the Construction of A Model rule of manual definition, and these class methods are little to the user interactions burden, and need not expensive catching and computing equipment, thereby become the powerful tool that creates complex scene.Yet complicacy and the abstractness of its formation rule definition make the user be difficult to visual control process modeling result.
In recent years, the process modeling method adopts the automatic tectonic model formation rule of model analysis technology, thereby has avoided loaded down with trivial details manual definition procedure.Document 3:Merrell P.Example-based model synthesis.Proceedings ofthe2007symposium on Interactive3D Graphics and Games.2007,105-112 has proposed the model synthetic method, and it can utilize discrete three dimensional example model self structure similarity to generate extensive three-dimensional scene models.This method only needs the user to provide little model of place can complete the automatically synthetic of any large scene, for the large-scale complex scene modeling provides a kind of approach easily.Existing model synthetic method only is satisfied with to seek and is met instance model in abutting connection with a feasible solution of constraint, other construction features that can't the routine model of complete body reality.This patent is introduced the global optimization composition mechanism and is satisfied the architectural feature of example that the user is provided to guarantee modeling result on the basis of existing model synthetic method.
Summary of the invention
Goal of the invention: technical matters to be solved by this invention is for the deficiencies in the prior art, and a kind of model synthetic method based on example is provided, and the method can more massive model automatically synthetic according to the instance model that the user provides or scene.
Technical scheme: the invention discloses a kind of large scale scene synthetic method based on example, comprise the following steps:
Step 1, instance analysis: a given example scenario model, the tectonic element set of calculated examples model of place and the restriction relation between tectonic element;
Step 2, the parametric modeling of object module construction process: the scale of given object module, generate the anabolic process that the Petri web frame represents the object module tectonic element, and the descriptive modelling process;
Step 3 is based on the Combinatorial Optimization of artificial immunity: use Artificial Immune Algorithm to be optimized to obtain the model the most similar to the architectural feature of example scenario model to parameterized construction process.
In step 1 of the present invention, comprise the following steps:
Use the symmetrical cell detection algorithm to extract the self similarity unit of example scenario model, according to the size of self similarity unit and the equally spaced three groups of orthogonal plane groups of positional information structure to cut apart the example scenario model, each organizes the plane respectively take X-axis, Y-axis and Z-direction as the method direction, and its number of planes is respectively N
x, N
yAnd N
z, set up thus the tectonic element set { M of example scenario model
[i, j, k]| 1≤i≤N
x, 1≤j≤N
y, 1≤k≤N
z, wherein, M
[i, j, k]Be a tectonic element, i, j and k represent that respectively this tectonic element is along the ordinal number of X-axis, Y-axis and Z axis;
Use the triangular mesh dough sheet that each tectonic element is expressed as: M
[i, j, k]={ V
[i, j, k], F
[i, j, k], l
[i, j, k], wherein, V
[i, j, k]Be vertex set, F
[i, j, k]Be tri patch collection, l
[i, j, k]Be label;
Calculate each tectonic element M
[i, j, k]Shape facility, comprise grid table area S, rectangular parallelepiped convex closure feature { x
0, x
1, y
0, y
1, z
0, z
1And grid vertex overlay marks { f
uvw| 0≤u, v, w≤1}, wherein, subscript u, v and w represent grid vertex overlay marks f
uvwThe rectangular parallelepiped convex closure summit sequence number of indication, described grid table area S=∑ A
F, A
FBe the area of a triangle surface on grid, described rectangular parallelepiped convex closure feature { x
0, x
1, y
0, y
1, z
0, z
1Be convex closure at X, the coordinate extreme value on three main shafts of Y and Z, grid vertex overlay marks f
uvwExpression rectangular parallelepiped convex closure summit [x
u, y
v, z
w] whether belong to vertex set V
[i, j, k]If belong to f
uvw=1, on the contrary f
uvw=0;
Use shape facility to the self similarity relation of tectonic element cluster with the extraction tectonic element, if two tectonic elements
With
Feature in full accord (wherein,
With
Expression ordinal number group is the reduced representation form of [i, j, k]), use same label L to carry out mark to these two unit, namely
Incorporate label L used into label set U
LIn;
Use following Boolean function represent example scenario Construction of A Model unit in abutting connection with constrain set:
Wherein, L
1And L
2Label set U
LIn two labels,
Be a Boolean function, its field of definition is U
L* U
L, codomain be 0,1},
It is the vector of unit length on three main shafts.
In step 2, the parametrization that adopts colored Petri network to set up the tectonic element anabolic process of object module represents, and then completes the description to construction process, comprises the following steps:
Scale parameter S according to user's setting
x, S
yAnd S
z, set up a three-dimensional grid space, S
x, S
yAnd S
zRepresent respectively this lattice space along the subspace number of X-axis, Y-axis and Z axis, span is positive integer;
Set up local Petri net in each lattice space in three-dimensional grid space, each local Petri net comprises candidate and two storehouse institute's nodes of filling and the candidate upgrades and the candidate selects two migration nodes, being followed successively by the candidate along flow relation upgrades migration node, candidate storehouse institute node, candidate and selects to move node and filling reservoir institute node, and connect arbitrary candidate storehouse institute node and upgrade the migration node with filling reservoir institute node with the candidate of adjacent tectonic element and be connected, thereby set up the Petri web frame of object module;
The operation of migration node in the Petri web frame of objective definition model:
The candidate upgrades the migration node: agree information and calculate the Tuo Ken of this migration node in abutting connection with constrain set according to the holder of candidate storehouse institute's node of adjacent tectonic element and filling reservoir institute node;
The candidate selects to move node: select a label L to make up from candidate storehouse institute node is random, the execution priority that the candidate selects to move node is lower than the execution priority that the candidate upgrades the migration node;
Define two class mutation operators and realize variation to the tectonic element anabolic process:
Label is selected variation: this variation does not change the node sequence of reasoning, only changes the candidate and selects to move the label that node is selected;
Node is selected variation: this variation selects other feasible migration nodes to carry out carrying out the candidate when selecting to move the node execution.
Comprise following steps in step 3 of the present invention:
The example scenario model of note input is M
E, carry out the architectural feature of instance model and calculate, obtain the tectonic element number vector of instance model
Wherein,
Expression example scenario model M
EContained label is the tectonic element number of L;
According to the Petri web frame that step 2 is set up, the inference method that Reusability transmits based on signal selects operation to generate N by the random label that the candidate selects to move node
AisIndividual different new model is designated as initial population Abs;
The population step of updating:
Carry out the individual configurations feature calculation, obtain each individual tectonic element number vector in initial population
Wherein,
Represent the tectonic element number that each individual contained label is L;
Calculated examples model of place M
EWith each individual M
OBetween affinity, its computing formula is:
Wherein, aff (M
E, M
O) expression example scenario model M
EWith individual M
OBetween affinity;
Select K the highest individuality of affinity as new population A bs
2
Calculate new population A bs
2In each individual M '
OConcentration:
Wherein, density (M '
O) be individual M '
OConcentration, h is the sum formula ordinal number, M '
hBe new population A bs
2In h individual, aff (M '
O, M '
h) be individual M '
OAnd M '
hBetween affinity;
For population A bs
2In each individuality, the clone obtains new individuality according to concentration and affinity, to consist of new population set Abs
3
According to the mutation operator of step 2 definition, each individual execution is made a variation, and obtain new population Abs in conjunction with initial population
3∪ Abs;
Calculate new population Abs
3Each individual and affinity instance model in ∪ Abs;
Select the N of affinity maximum
AisIndividuality returns to the population step of updating as new initial population, until the gained population no longer changes.
Beneficial effect: the present invention has the following advantages: 1, allow the user instance model is provided and automatically carries out the Expansion of model, improved modeling efficiency, and realized effectively reusing of data with existing resource; 2, adopt the construction process of Petri net expression model, implicitly represent on the one hand structure constraint and the feature of model, realized on the other hand the management to Construction of A Model, support recalling and make a variation and then realize whole optimizing process construction process; 3, in the present invention for the high-dimensional characteristic of the parameter of construction process, adopt artificial immunity as optimization method, effectively avoided optimum results to be absorbed in local optimum.
Description of drawings
Below in conjunction with the drawings and specific embodiments, the present invention is done further illustrating, above-mentioned and/or otherwise advantage of the present invention will become apparent.
Fig. 1 is treatment scheme schematic diagram of the present invention.
Fig. 2 a is the input model collection example schematic diagram of embodiment.
Fig. 2 b is the output model collection example schematic diagram of embodiment.
Fig. 3 a is the quadrature segmentation plane group schematic diagram of embodiment.
Fig. 3 b is some tectonic element schematic diagram that embodiment is cut apart gained.
Fig. 4 is the Petri web frame of embodiment.
Embodiment:
The invention discloses a kind of large scale scene synthetic method based on example, comprise the following steps:
Step 1, the instance model analysis: a given instance model, build model the tectonic element set and between restriction relation;
Step 2, the parametric modeling of object module construction process: the scale of given object module generates the anabolic process that the Petri web frame represents tectonic element, and adopts the reasoning process realization to effective description of modeling process;
Step 3 is based on the Combinatorial Optimization of artificial immunity: use Artificial Immune Algorithm to be optimized to obtain the model the most similar to the instance model feature to parameterized construction process.
More particularly, the present invention allows the user that an instance model is provided, the alms giver will relate to model analysis, construction process parametric modeling and based on the large gordian technique of Combinatorial Optimization three of artificial immunity in fact, the Expansion of realization to instance model, its treatment scheme as shown in Figure 1, Fig. 2 is an example of modeling result, and Fig. 2 a is input model, and Fig. 2 b is output model.Model analysis is to cut apart the tectonic element set of extraction model by model, and the syntople between operation parameter the Representation Equation unit; The construction process parametric modeling is to use the Petri net to carry out parametric modeling to the anabolic process of tectonic element, and definition variation and back tracking operation realization are to the management of modeling process; Combinatorial Optimization based on artificial immunity uses clonal selection algorithm to select to reflect that the model of instance model shape and architectural feature is as modeling result at corresponding solution space.The below introduces respectively the main embodiment of each several part.
1. instance model analysis
The instance model M that the present invention processes
EBe triangular mesh model, can be expressed as V, F}, wherein,
Be vertex set, wherein, i ' is the summit ordinal number,
Be summit number, v
Ei '=[x
Ei ', y
Ei ', z
Ei '] be a summit, x
Ei ', y
Ei ', z
Ei 'Be a v
Ei 'D coordinates value,
Be the triangle surface set, wherein, j ' is the dough sheet ordinal number,
Be triangle surface number, f
Ej '=[t
1j ', t
2j ', t
3j '] be a triangle surface, this triangle is by t in vertex set V
1j ', t
2j 'And t
3j 'Individual summit forms.A given instance model M
EThe present invention is the symmetrical cell on computation model at first, and utilize the size of symmetrical cell and position to set up equally spaced three groups of orthogonal plane components and cut model, segmentation result is the tectonic element set of instance model, then calculate the restriction relation between tectonic element, namely according to the shape facility of each unit to tectonic element set classify, and by Boolean function represent between tectonic element in abutting connection with restriction relation.
1.1 the tectonic element set builds
Thereby this step utilizes equally spaced three groups of orthogonal plane components to cut the tectonic element set that model obtains model, and the present invention is according to the three group orthogonal plane group { Planes of instance model symmetrical cell testing result structure in order to cut apart
X, Plane
Y, Plane
Z, wherein, Plane
X, Plane
YAnd Plane
ZRepresent respectively three groups of planes take X, Y and Z axis as normal direction.
Given two three-dimensional model M
S1={ V
S1, F
S1And M
S2={ V
S2, F
S2, if model M
S1And M
S2Satisfy following condition:
(1). model M
S1And M
S2It is model M
E=V, the submodel of F}, namely
Wherein, l ' is the model ordinal number;
(2). model M
S1And M
S2Has consistent mesh topology, i.e. vertex set V
S1And V
S2Set sizes identical, and have vertex set V
S1An arrangement, to vertex set V
S1With triangle surface collection F
S1In after summit sequence number in each triangle surface rearranges, the new triangle surface set F ' of gained
S1With F
S2In full accord;
(3). there is an affined transformation T, makes T (V
S1)=V
S2, wherein, T (V
S1) be to V
S1In each summit carry out the vertex set that consists of after conversion T;
Model M
S1And M
S2Be called model M
EOne group of symmetrical cell, can be found out one group of submodel with affine unchangeability on symmetrical cell finger print type by above definition.Because the model building-up process is only carried out translation transformation to tectonic element, therefore, the present invention only considers to have the symmetrical cell of translation invariant shape.
The present invention adopts based on the symmetrical cell set on the symmetrical cell detection method extraction instance model of ballot conversion, and it is defined as { M
Si ' '| i ' '=1 ..., N
Ms, wherein, M
Si ' 'Be model M
EOn the individual symmetrical cell of i ' ', the symmetrical cell set sizes is N
Ms, computing method are seen document 4:Pauly M., Mitra N.J., Wallner J., Pottmann H., andGuibas L.J.Discovering structural regularity in3D geometry.ACM Transactions onGraphics, 2008,27 (3), Article43.
According to symmetrical cell set { M
Si ' '| i ' '=1 ..., N
Ms, the present invention calculates each unit M
Si ' 'Three-dimensional rectangle convex closure size, can be expressed as [l
Hxi ' ', l
Hyi ' ', l
Hzi ' '], wherein, l
Hxi ' ', l
Hyi ' 'And l
Hzi ' 'Represent respectively unit M
Si ' 'Each limit length of side of three-dimensional rectangle convex closure.In order to make the more symmetrical cell of tectonic element set-inclusion of model, the present invention is according to position and the equally spaced three groups of orthogonal plane groups of convex closure length of side structure of the symmetrical cell of convex closure volume minimum, and the symmetrical cell of supposing convex closure volume minimum is M '
S, its three-dimensional rectangle convex closure size be expressed as [l '
Hx, l '
Hy, l '
Hz], the D coordinates value at convex closure center be [C '
Mx, C '
My, C '
Mz], instance model M
EThe size of three-dimensional rectangle convex closure be expressed as [l
Hx, l
Hy, l
Hz], the D coordinates value at its center is [C
Mx, C
My, C
Mz], three groups of orthogonal plane group { Plane
X, Plane
Y, Plane
ZConstruction process is as follows:
Step 1: calculated examples model three-dimensional rectangle convex closure is along the minimal value of all directions coordinate
Computing formula is
Step 2: computing unit M '
SThe three-dimensional rectangle convex closure is along the minimal value of all directions coordinate
Computing formula is
Step 3: calculate the min coordinates value of respectively organizing orthogonal plane and corresponding normal direction coordinate axis intersection point
Wherein,
With
Represent respectively set of planes Plane
XMidplane and X-axis, set of planes Plane
YMidplane and Y-axis and set of planes Plane
ZThe intersection point minimal value of midplane and Z axis, computing formula is
Step 4: calculate the number [N that respectively organizes orthogonal plane
x, N
y, N
z], wherein, N
x, N
yAnd N
zRepresent respectively set of planes Plane
X, Plane
YAnd Plane
ZIn number of planes, computing formula is
Wherein,
Be Gaussian function;
Step 5: formation level group Plane
X, Plane
YAnd Plane
Z, computing formula is
These three groups of orthogonal planes can be constructed a three-dimensional grid space, comprise altogether (N
x+ 1) * (N
y+ 1) * (N
z+ 1) sub spaces, Fig. 3 a(is based on the characteristic of effect of the present invention, can only use gray-scale map to show) in the coarse net ruling represented the cut-off rule of three-dimensional grid space on embodiment, every sub spaces can be used a parameterized three-dimensional coordinate
Represent, wherein, 1≤i≤N
x, 1≤j≤N
y, 1≤k≤N
zThe present invention is according to three groups of orthogonal plane group { Plane that construct
X, Plane
Y, Plane
ZThe cutting instance model, (N can be obtained
x+ 1) * (N
y+ 1) * (N
z+ 1) individual tectonic element, Fig. 3 b(is based on the characteristic of effect of the present invention, can only use gray-scale map to show) be 3 tectonic elements wherein, each tectonic element is all corresponding to the sub spaces in building three-dimensional grid space, the present invention is defined as { M with the tectonic element set as model of the tectonic element of gained
[i, j, k]| 1≤i≤N
x, 1≤j≤N
y, 1≤k≤N
z, M wherein
[i, j, k]It is the corresponding tectonic element in subspace of [i, j, k] for the parametrization coordinate.
1.2 structure constraint calculates
After the tectonic element set that obtains model, the present invention calculates the shape facility of each tectonic element, and according to shape facility, tectonic element is classified, belong to of a sort unit and will use same integer to carry out mark to it, at last by Boolean function represent between the tectonic element of instance model in abutting connection with restriction relation.
At first, calculate tectonic element M
[i, j, k]Shape facility vector
Be followed successively by grid table area S, three-dimensional rectangle convex closure feature F
H={ x
0, x
1, y
0, y
1, z
0, z
1And grid vertex overlay marks { f
uvw| 0≤u, v, w≤1}, wherein, grid table area S=∑ A
F, the area A of all triangle surfaces on the expression grid
FSum, three-dimensional rectangle convex closure feature { x
0, x
1, y
0, y
1, z
0, z
1Be convex closure at X, the coordinate extreme value on three main shafts of Y and Z, grid vertex overlay marks f
uvwExpression rectangular parallelepiped convex closure summit [x
u, y
v, z
w] whether belong to vertex set V, if belong to f
uvw=1, on the contrary f
uvw=0;
Then, tectonic element is classified, if tectonic element
With
The shape facility vector in full accord, with the unit
With
Be divided into a class, suppose at last tectonic element to be divided into N
SClass is used set of integers U
L=1 ..., N
SIn a certain integer the tectonic element in all kinds of is carried out mark, of a sort tectonic element uses same integer.
At last, the present invention has defined a kind of Boolean function and has represented syntople between tectonic element, and this Boolean function is expressed as
Wherein,
Be main shaft vector of unit length collection, if
Represented label L
1And L
2The tectonic element of representative is along vector of unit length
Can connect, otherwise, represent that corresponding tectonic element is along vector of unit length
Can not connect.Instance model M
EBoolean function computation process as follows:
Step 1: for each label L
1∈ U
L, label l
2∈ U
L, main shaft vector of unit length vector
Initialization
Step 2: travel through every sub spaces in three-dimensional grid space, suppose the parametrization coordinate of current subspace
Be [i, j, k], the label of the tectonic element that the subspace is corresponding is L
1, calculate successively this subspace along vector of unit length
Adjacent subspace, the parametrization coordinate of this adjacent subspace is
If this parametrization coordinate in field of definition, extracts tectonic element label L corresponding to this adjacent subspace
2, and arrange
With
According to above-mentioned steps, the present invention calculate Boolean function corresponding to instance model with expression between tectonic element in abutting connection with restriction relation, the new model that generates all satisfies this in abutting connection with constraint.
2. the parametric modeling of construction process
Because the present invention uses Artificial Immune Algorithm, modeling result is optimized, therefore need to set up a kind of process administrative mechanism to modeling, the present invention uses Petri net descriptive model building-up process, according to defined Petri web frame, can utilize the inference method based on the Petri net that the model building-up process is represented, and the corresponding mutation operator of definable is to realize the differentiation to reasoning process.
2.1Petri web frame
The Petri net is a kind of strong instrument of describing the asynchronism and concurrency system, its structure is a kind of oriented bipartite graph, mainly comprise two category nodes: storehouse institute's node and migration node, wherein, what storehouse institute node was described is system state, can use holder to agree the behavioral characteristics of library representation institute node, the migration node represents the conversion between state.Directed edge only can be from storehouse institute node to the migration node, and perhaps from the migration node to storehouse institute node, the generation of at every turn moving node need to consume the Tuo Ken of input magazine institute node, and produces corresponding holder agree in output storehouse institute node.Basic petri net can use tlv triple P
N={ S
P, T
R, F
LFormalization representation, tlv triple must satisfy following condition:
dom(F
L)∪cod(F
L)=S
P∪T
R
Wherein,
, S
PAnd T
RBe called storehouse institute's set of node and transition collection, F
LBe flow relation, the state of basic petri net only uses that holder agree number regulation storehouse institute node.If use the basic petri net realization to the increment filling process of three-dimensional element, holder is agree number and will be increased substantially along with the lifting of model complexity, causes network " node blast ".Therefore, the present invention uses the colored Petri network realization to the description of modeling process, colored Petri network is a kind of High Level Petri Net structure, distinguishing different types of holder by different colors agree, namely give holder and agree certain information content, thereby improved the descriptive power to system state on the basis that does not increase storehouse institute node.
The present invention is in corresponding to every sub spaces in lattice space storehouse institute node at filling process state, and operating accordingly at the migration node definition, this operation are agree information according to the holder of input magazine institute node and are calculated the holder of output storehouse institute node and agree information.The present invention is considered as the model building-up process to insert to a three-dimensional grid space process of tectonic element, this process can realize in two steps: at first calculate the candidate's tectonic element set that can fill each subspace, then select a unit to fill from the set of candidate's tectonic element.Accordingly, the scale parameter that at first the present invention sets according to the user is set up the three-dimensional grid space, then at each trellis subspace place's definition two class libraries institute's node and two classes migration node, Fig. 4 is the Petri web frame of embodiment, is followed successively by by the flow relation direction: the candidate upgrades migration node T
R1, the candidate storehouse node S of institute
P1, the candidate selects to move node T
R2, the filling reservoir node S of institute
P2Flow relation between adjacent subspace upgrades the migration node from candidate storehouse institute's node or filling reservoir institute node-flow to the candidate, the expression candidate upgrades the information that the migration nodal operation is accepted candidate storehouse institute's node or the filling reservoir institute node of adjacent subspace, and calculates candidate's filler cells serial number information of subspace, place.The candidate selects to move that node is random from the candidate storehouse institute node of this subspace selects a certain unit to fill.If the state in the institute's node of the candidate storehouse in a certain subspace or filling reservoir institute node changes, notify adjacent subspace to carry out state and upgrade.In order to enrich the information storage capability of storehouse institute node, the present invention uses the colored Petri web frame, and to support the Construction of A Model process based on reasoning, it is T that representation is agree in defined holder by the control parameter of agreeing in holder record modeling process
Token={ s
Uflag, S
C, l
r, wherein, update mode mark s
Uflag∈ (f
u, f
d, f
l, f
r, f
f, f
b), f
u, f
d, f
l, f
r, f
f, f
bRepresent respectively whether these upper and lower, left and right, subspace, these six of front and rears have upgraded state separately in abutting connection with the subspace, if 1, expression is not also upgraded, on the contrary expression is upgraded; Candidate's label S set
C, for being filled into the tectonic element label set in this space; Final label l
r, for finally being filled into the tectonic element label in this space.
2.2 migration nodal operation definition
Information is agree in holder in the renewable storehouse of execution migration nodal operation institute node.Petri net in the present invention comprises two class migration nodes: the candidate upgrades the migration node and the candidate selects to move node, wherein, the preferential execution degree that the candidate upgrades the migration node is higher than the preferential execution degree that the candidate selects to move node, and the below will introduce this two classes migration nodal operation respectively in detail.
The candidate upgrades the migration node: this migration nodal operation upgrades candidate's label sets of subspace, place according to candidate's label sets of adjacent subspace, upgrade the adjacency constraint that result need satisfy the definition of instance model Boolean function, in case a certain in candidate's label sets of subspace, place be marked at do not exist in candidate's label sets of adjacent cells can be adjacent mark, namely clash with restriction relation, this mark is removed from candidate collection.By the definition of subspace as can be known, the candidate upgrades the migration node and has 12 input magazine institute nodes, and without loss of generality, the definition candidate upgrades migration node T
R1Input magazine institute set of node be:
{S
P1u,S
P2u,S
P1d,S
P2d,S
P1l,S
P2l,S
P1r,S
P2r,S
P1f,S
P2f,S
P1b,S
P2b},
Wherein, S
P1u, S
P2uBe respectively candidate storehouse institute's node and filling reservoir institute node in the subspace that is in subspace, place top, S
P1d, S
P2dBe respectively candidate storehouse institute's node and filling reservoir institute node in the subspace that is in subspace, place below, S
P1l, S
P2lBe respectively candidate storehouse institute's node and filling reservoir institute node in the subspace that is in subspace, place left, S
P1r, S
P2rBe respectively candidate storehouse institute's node of being in right-hand subspace, subspace, place and filling reservoir institute node, S
P1f, S
P2fBe respectively candidate storehouse institute's node and filling reservoir institute node in the subspace that is in the place ahead, subspace, place, S
P1b, S
P2bBe respectively candidate storehouse institute's node and filling reservoir institute node in the subspace that is in rear, subspace, place.(annotate: the input magazine institute set of node that the candidate who is arranged in the subspace, border upgrades the migration node is the subset of this set, only need therefrom remove non-existent input magazine institute node and get final product).Upgrade migration node T for the candidate
R1, its trigger condition sign F
R1Computing formula is as follows:
F
ud=S
P1u.f
u∨S
P2u.f
u∨S
P1d.f
d∨S
P2d.f
d
F
lr=S
P1l.p
l∨S
P2l.p
l∨S
P1r.p
r∨S
P2r.p
r
F
fb=S
P1f.p
f∨S
P2f.p
f∨S
P1b.p
b∨S
P2b.p
b,
F
R1=F
ud∨F
lr∨F
fb
If sign F
R1=1, to carry out this candidate and move node, its detailed operating process is as follows:
Step 1: each the input magazine node S of institute that travels through this migration node
P α β, wherein, α ∈ 1,2} library representation institute node type label, β ∈ u, and d, l, r, f, b} are the bearing mark of subspace, if S
P α β.p
β=0, return to the next input magazine of step 1 judgement institute node;
Step 2: calculate from moving subspace, node place to the input magazine node S of institute
P α βThe vector of unit length of subspace, place
, the parametrization coordinate that note is moved subspace, node place is [i
T, j
T, k
T], the parametrization coordinate of subspace, input magazine institute's node place be [i '
T, j '
T, k '
T],
Step 3: for the willing middle candidate's label S set of input magazine institute's node holder
CIn each label
The set of structure label
This set is satisfied in abutting connection with constrain set for all
The set of label l, merge all label set
Be S
U
Step 4: establish S
CoFor candidate's label set of this migration node input magazine institute node, calculate S '
C=S
Co∩ S
UFor the new candidate's label set of output storehouse institute's node, S is set
P α β.p
β=0, if S
Co≠ S '
C, export the node holder of storehouse institute agree in all update mode marks be set to 1, return to the next input magazine of step 1 judgement institute node.
So far can complete the candidate upgrades the migration node and the state of candidate storehouse institute node is upgraded realizing.
The candidate selects to move node: this migration nodal operation selects tectonic element corresponding to any mark to be filled in migration subspace, node place from the set of the input magazine node candidate of institute label, carry out trigger condition that the candidate selects to move node and be the node holder of filling reservoir institute agree in final label not by assignment.The execution that the candidate selects to move node comprises three kinds of situations:
(1). the candidate's label set-inclusion zero element during the node holder of candidate storehouse institute is agree;
(2). the element of candidate's label set-inclusion during the node holder of candidate storehouse institute is agree;
(3). the candidate's label set-inclusion during the node holder of candidate storehouse institute is agree is more than an element.
The execution relative importance value is situation (1)〉situation (2)〉situation (3).If candidate's label set-inclusion zero element, reasoning construction process failure is if the reasoning construction process again that begins most that this situation dates back reasoning process occurs, until reasoning is successfully constructed; If the set of candidate's label only comprises an element, in should the migration node directly the filling reservoir institute node holder of output being agree, final label assignment is this element; If candidate's label set-inclusion is more than an element, this migration node is from the random final label of selecting a label to give filling reservoir institute node of candidate's label set, and set the candidate label set of filling reservoir institute node holder in agreeing and only comprise final label, set simultaneously all update mode marks and be set to 1.
2.3 the variation of tectonic element anabolic process
Described reasoning process has the two random selection courses in place, and first random selection course is that the candidate of situation (3) selects to move the node selection, and when namely not having other feasible classes migration nodes, the present invention selects an executable migration node to carry out at random; Second random selection course is that the candidate under implementation status (3) selects to move node, selects a mark as final mark at random from the set of candidate's label.
According to above two kinds of random selection situations, the present invention has defined two kinds of mutation operators to realize the differentiation to construction process: node selects variation and label to select variation.
Node is selected variation: this variation is selected other executable candidates to select to move node and is carried out when the selection candidate selects to move node and selects at random;
Label is selected variation: when this variation is selected to move node the same candidate of execution, select another label in the set of candidate's label.
By above two kinds of mutation operators, can satisfy instance model in abutting connection with the traversal of constraint set of feasible solution with realization to all by the differentiation to construction process, thereby can support to use Artificial Immune Algorithm to concentrate selection to meet the solution of instance model structure and shape facility most from separating.
3. based on the optimal combination of artificial immunity
After the parametrization that has built the model building-up process represented, the present invention used clonal selection algorithm to realize the Combinatorial Optimization synthetic to model.Clonal selection algorithm is an important algorithm in artificial immunity field, the fields such as network security, viral detection, pattern-recognition, Combinatorial Optimization have been widely used in, its basic thought is to select those can identify the cell of antigen as the immunocyte of a new generation, and increases.The present invention adopts parameterized Petri net as the coding of immunocyte, selects variation and label to select variation as mutation operator node, affinity calculating take the number of unit ratio of model-composing as feature to weigh the similarity of two models.
Model synthetic method detailed step based on artificial immunity optimization of the present invention is described below:
The architectural feature calculation procedure of instance model:
The example scenario model of note input is M
E, carry out the architectural feature of instance model and calculate, obtain the tectonic element number vector of instance model
Wherein,
Expression example scenario model M
EContained label is the tectonic element number of L;
The initialization of population step:
According to the Petri web frame that step 2 is set up, the inference method that Reusability transmits based on signal selects operation to generate N by the random label that the candidate selects to move node
AisIndividual different new model is designated as initial population Abs, comprises the following steps:
Step 1: candidate's label S set that in the storehouse institute node of all nodes of initialization, holder is agree
CBe label set U
L, all update mode marks are set to 0;
Step 2: select at random a certain candidate to select to move node and carry out its operation;
Step 3: the executable candidate of search upgrades the migration node and carries out its operation in Petri net, until have executable candidate to upgrade the migration node;
Step 4: a certain executable candidate of random search selects to move node and carries out its operation in the Petri net, does not have the candidate who carries out to select to move node if exist, and returns to step 2;
Step 5: the tectonic element corresponding according to the label combination of Petri net builds final mask.
Because reasoning process exists stray parameter, different stochastic processes produces different results, and the present invention repeatedly carries out the inference method that transmits based on signal and generates N
Ais(the present invention gets N to individual different new model
Ais=100), be designated as initial population Abs;
The population step of updating:
Carry out the individual configurations feature calculation, obtain each individual tectonic element number vector in initial population
Wherein,
Represent the tectonic element number that each individual contained label is L;
Calculated examples model of place M
EWith each individual M
OBetween affinity, its computing formula is:
Wherein, aff (M
E, M
O) expression example scenario model M
EWith individual M
OBetween affinity;
Select K the highest individuality of affinity as new population A bs
2(the present invention gets K=30);
Calculate new population A bs
2In each individual M '
OConcentration:
Wherein, density (M '
O) be individual M '
OConcentration, h is the sum formula ordinal number, M '
hBe new population A bs
2In h individual, aff (M '
O, M '
h) be individual M '
OAnd M '
hBetween affinity;
For population A bs
2In each individuality, the clone obtains new individuality according to concentration and affinity, to consist of new population set Abs
3, higher with the affinity of instance model, lower clone is more for concentration;
According to the mutation operator of step 2 definition, each individual execution is made a variation, and obtain new population Abs in conjunction with initial population
3∪ Abs;
Calculate new population Abs
3Each individual and affinity instance model in ∪ Abs;
Select the N of affinity maximum
AisIndividuality returns to the population step of updating as new initial population, until the gained population no longer changes.
Through optimizing process, the present invention selects the individuality of affinity maximum as new results model, thereby it is synthetic to complete the model of optimizing based on artificial immunity.
The invention provides a kind of thinking and method of the large scale scene synthetic method based on example; method and the approach of this technical scheme of specific implementation are a lot; the above is only the preferred embodiment of the present invention; should be understood that; for those skilled in the art; under the prerequisite that does not break away from the principle of the invention, can also make some improvements and modifications, these improvements and modifications also should be considered as protection scope of the present invention.In the present embodiment not clear and definite each ingredient all available prior art realized.
Claims (4)
1. the large scale scene synthetic method based on example, is characterized in that, comprises the following steps:
Step 1, instance analysis: a given example scenario model, the tectonic element set of calculated examples model of place and the restriction relation between tectonic element;
Step 2, the parametric modeling of object module construction process: the scale of given object module, generate the anabolic process that the Petri web frame represents the object module tectonic element, and the descriptive modelling process;
Step 3 is based on the Combinatorial Optimization of artificial immunity: use Artificial Immune Algorithm to be optimized to obtain the model the most similar to the architectural feature of example scenario model to parameterized construction process.
2. a kind of large scale scene synthetic method based on example according to claim 1, is characterized in that, in step 1, comprises the following steps:
Use the symmetrical cell detection algorithm to extract the self similarity unit of example scenario model, according to the size of self similarity unit and the equally spaced three groups of orthogonal plane groups of positional information structure to cut apart the example scenario model, each organizes the plane respectively take X-axis, Y-axis and Z-direction as the method direction, and its number of planes is respectively N
x, N
yAnd N
z, set up thus the tectonic element set { M of example scenario model
[i, j, k]| 1≤i≤N
x, 1≤j≤N
y, 1≤k≤N
z, wherein, M
[i, j, k]Be a tectonic element, i, j and k represent that respectively this tectonic element is along the ordinal number of X-axis, Y-axis and Z axis;
Use the triangular mesh dough sheet that each tectonic element is expressed as: M
[i, j, k]={ V
[i, j, k], F
[i, j, k], l
[i, j, k], wherein, V
[i, j, k]Be vertex set, F
[i, j, k]Be tri patch collection, l
[i, j, k]Be label;
Calculate each tectonic element M
[i, j, k]Shape facility, comprise grid table area S, rectangular parallelepiped convex closure feature { x
0, x
1, y
0, y
1, z
0, z
1And grid vertex overlay marks { f
uvw| 0≤u, v, w≤1}, wherein, subscript u, v and w represent grid vertex overlay marks f
uvwThe rectangular parallelepiped convex closure summit sequence number of indication, described grid table area S=∑ A
F, A
FBe the area of a triangle surface on grid, described rectangular parallelepiped convex closure feature { x
0, x
1, y
0, y
1, z
0, z
1Be convex closure at X, the coordinate extreme value on three main shafts of Y and Z, grid vertex overlay marks f
uvwExpression rectangular parallelepiped convex closure summit [x
u, y
v, z
w] whether belong to vertex set V
[i, j, k]If belong to f
uvw=1, on the contrary f
uvw=0;
Use shape facility to the self similarity relation of tectonic element cluster with the extraction tectonic element, if two tectonic elements
With
Feature in full accord (wherein,
With
Expression ordinal number group is the reduced representation form of [i, j, k]), use same label L to these two tectonic elements
With
Label
With
Carry out assignment, namely
Incorporate label L used into label set U
LIn;
Use following Boolean function represent example scenario Construction of A Model unit in abutting connection with constrain set:
3. a kind of large scale scene synthetic method based on example according to claim 2, it is characterized in that, in step 2, the parametrization that adopts colored Petri network to set up the tectonic element anabolic process of object module represents, and then complete description to construction process, comprise the following steps:
Scale parameter S according to user's setting
x, S
yAnd S
z, set up a three-dimensional grid space, S
x, S
yAnd S
zRepresent respectively this lattice space along the subspace number of X-axis, Y-axis and Z axis, span is positive integer;
Set up local Petri net in each lattice space in three-dimensional grid space, each local Petri net comprises candidate and two storehouse institute's nodes of filling and the candidate upgrades and the candidate selects two migration nodes, being followed successively by the candidate along flow relation upgrades migration node, candidate storehouse institute node, candidate and selects to move node and filling reservoir institute node, and connect arbitrary candidate storehouse institute node and upgrade the migration node with filling reservoir institute node with the candidate of adjacent tectonic element and be connected, thereby set up the Petri web frame of object module;
The operation of migration node in the Petri web frame of objective definition model:
The candidate upgrades the migration node: agree information and calculate the Tuo Ken of this migration node in abutting connection with constrain set according to the holder of candidate storehouse institute's node of adjacent tectonic element and filling reservoir institute node;
The candidate selects to move node: select a label L to make up from candidate storehouse institute node is random, the execution priority that the candidate selects to move node is lower than the execution priority that the candidate upgrades the migration node;
Define two class mutation operators and realize variation to the tectonic element anabolic process:
Label is selected variation: this variation does not change the node sequence of reasoning, only changes the candidate and selects to move the label that node is selected;
Node is selected variation: this variation selects other feasible migration nodes to carry out carrying out the candidate when selecting to move the node execution.
4. a kind of large scale scene synthetic method based on example according to claim 3, is characterized in that, comprises following steps in step 3:
The architectural feature calculation procedure of instance model:
The example scenario model of note input is M
E, carry out the architectural feature of instance model and calculate, obtain the tectonic element number vector of instance model
Wherein,
Expression example scenario model M
EContained label is the tectonic element number of L;
The initialization of population step:
According to the Petri web frame that step 2 is set up, the inference method that Reusability transmits based on signal selects operation to generate N by the random label that the candidate selects to move node
AisIndividual different new model is designated as initial population Abs;
The population step of updating:
Carry out the individual configurations feature calculation, obtain each individual tectonic element number vector in initial population
Wherein,
Represent the tectonic element number that each individual contained label is L;
Calculated examples model of place M
EWith each individual M
OBetween affinity, its computing formula is:
Wherein, aff (M
E, M
O) expression example scenario model M
EWith individual M
OBetween affinity;
Select K the highest individuality of affinity as new population A bs
2
Calculate new population A bs
2In each individual M '
OConcentration:
Wherein, density (M '
O) be individual M '
OConcentration, h is the sum formula ordinal number, M '
hBe new population A bs
2In h individual, aff (M '
O, M '
h) be individual M '
OAnd M '
hBetween affinity;
For population A bs
2In each individuality, the clone obtains new individuality according to concentration and affinity, to consist of new population set Abs
3
According to the mutation operator of step 2 definition, each individual execution is made a variation, and obtain new population Abs in conjunction with initial population
3∪ Abs;
Calculate new population Abs
3Each individual and affinity instance model in ∪ Abs;
Select the N of affinity maximum
AisIndividuality returns to the population step of updating as new initial population, until the gained population no longer changes.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310075955.9A CN103177473B (en) | 2013-03-11 | 2013-03-11 | A kind of large scale scene synthetic method of Case-based Reasoning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310075955.9A CN103177473B (en) | 2013-03-11 | 2013-03-11 | A kind of large scale scene synthetic method of Case-based Reasoning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103177473A true CN103177473A (en) | 2013-06-26 |
CN103177473B CN103177473B (en) | 2015-08-19 |
Family
ID=48637299
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310075955.9A Expired - Fee Related CN103177473B (en) | 2013-03-11 | 2013-03-11 | A kind of large scale scene synthetic method of Case-based Reasoning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103177473B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105957132A (en) * | 2016-04-21 | 2016-09-21 | 北京大学 | Three dimensional scene high-performance drawing optimization method containing height complex drawing elements |
CN110309612A (en) * | 2019-07-08 | 2019-10-08 | 西安交通大学 | Dynamical system fault handling method based on fuzzy fault Petri network |
CN111582163A (en) * | 2020-05-07 | 2020-08-25 | 北京理工大学 | Large-scale crowd position transformation scheme generation method |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101249019A (en) * | 2008-03-11 | 2008-08-27 | 微创医疗器械(上海)有限公司 | Human body organ three-dimensional surface rebuilding method and system |
CN102012976A (en) * | 2010-11-09 | 2011-04-13 | 刘丽 | Artificial immunity network-based positron emission tomography (PET) molecular image dynamics modeling method |
-
2013
- 2013-03-11 CN CN201310075955.9A patent/CN103177473B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101249019A (en) * | 2008-03-11 | 2008-08-27 | 微创医疗器械(上海)有限公司 | Human body organ three-dimensional surface rebuilding method and system |
CN102012976A (en) * | 2010-11-09 | 2011-04-13 | 刘丽 | Artificial immunity network-based positron emission tomography (PET) molecular image dynamics modeling method |
Non-Patent Citations (6)
Title |
---|
LIANGLIANG NAN,ET AL: "SmartBoxes for Interactive Urban Reconstruction", 《ACM TRANSACTIONS ON GRAPHICS》, vol. 29, no. 4, 31 July 2010 (2010-07-31), pages 93, XP058041110, DOI: 10.1145/1778765.1778830 * |
MARK PAULY,ET AL: "Discovering structural regularity in 3D geometry", 《ALGORITHMEN UND MATHEMATISCHE MODELLIERUNG》, 31 August 2008 (2008-08-31) * |
PAUL MERRELL: "Example-based model synthesis", 《PROCEEDINGS OFTHE2007SYMPOSIUM ON INTERACTIVE3D GRAPHICS AND GAMES》, 30 December 2007 (2007-12-30), pages 105 - 112 * |
PAULY M,ET AL: "Procedural modeling of buildings", 《ACM TRANSACTION OF GRAPHICS》, vol. 26, no. 3, 31 July 2007 (2007-07-31), pages 614 - 623 * |
刘凯,等: "采用草绘轮廓的3维人脸建模方法", 《中国图像图形学报》, vol. 16, no. 6, 30 June 2011 (2011-06-30), pages 1102 - 1110 * |
宋沫飞,等: "采用多幅草图的正交多面体模型生成方法", 《计算机辅助设计与图形学学报》, vol. 24, no. 1, 31 January 2012 (2012-01-31), pages 50 - 59 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105957132A (en) * | 2016-04-21 | 2016-09-21 | 北京大学 | Three dimensional scene high-performance drawing optimization method containing height complex drawing elements |
CN105957132B (en) * | 2016-04-21 | 2019-03-08 | 北京大学 | Optimization method is drawn comprising the highly complex three-dimensional scenic high-performance for drawing element |
CN110309612A (en) * | 2019-07-08 | 2019-10-08 | 西安交通大学 | Dynamical system fault handling method based on fuzzy fault Petri network |
CN111582163A (en) * | 2020-05-07 | 2020-08-25 | 北京理工大学 | Large-scale crowd position transformation scheme generation method |
Also Published As
Publication number | Publication date |
---|---|
CN103177473B (en) | 2015-08-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101807308B (en) | Three-dimensional model segmenting device and method | |
CN101944239B (en) | Three-dimensional model dividing method, device and comprise the image processing system of this device | |
CN107016175B (en) | It is applicable in the Automation Design method, apparatus and optimization method of neural network processor | |
Gyulassy et al. | A practical approach to Morse-Smale complex computation: Scalability and generality | |
CN103400372B (en) | A kind of three-dimensional topology information extracting method described based on Reeb figure | |
CN106203506A (en) | A kind of pedestrian detection method based on degree of depth learning art | |
CN103065354A (en) | Device and method for point cloud optimization | |
CN109190461B (en) | A kind of dynamic gesture identification method and system based on gesture key point | |
CN106056606A (en) | Image processing method and device | |
CN103080941A (en) | Apparatus for generating computational data, method for generating computational data, and program for generating computational data | |
CN105354208A (en) | Big data information mining method | |
CN102708327A (en) | Network community discovery method based on spectrum optimization | |
CN103065009B (en) | Intelligent design system and method of traffic sign lines | |
CN106845536A (en) | A kind of parallel clustering method based on image scaling | |
CN103177473A (en) | Instance-based large-scale scene composition method | |
CN107229234A (en) | The distributed libray system and method for Aviation electronic data | |
CN110853120B (en) | Network layout method, system and medium based on segmentation drawing method | |
CN104090945B (en) | Geographic space entity constructing method and system | |
CN105868478A (en) | Rotating mechanical equipment virtual assembly model and method based on context awareness | |
Korotov et al. | Longest-edge n-section algorithms: properties and open problems | |
CN109492522A (en) | Specific objective detection model training program, equipment and computer readable storage medium | |
CN109325092A (en) | Merge the nonparametric parallelization level Di Li Cray process topic model system of phrase information | |
Huang et al. | A matching algorithm between precursory 3D process model and 2D working procedure drawing based on subgraph isomorphism | |
Tugurlan | Fast marching methods-parallel implementation and analysis | |
CN107133639A (en) | Merge the adaptively sampled method of non-equilibrium data of Boost models |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150819 Termination date: 20180311 |
|
CF01 | Termination of patent right due to non-payment of annual fee |