CN103218818B - Three-dimensional model dividing method and system - Google Patents
Three-dimensional model dividing method and system Download PDFInfo
- Publication number
- CN103218818B CN103218818B CN201310138385.3A CN201310138385A CN103218818B CN 103218818 B CN103218818 B CN 103218818B CN 201310138385 A CN201310138385 A CN 201310138385A CN 103218818 B CN103218818 B CN 103218818B
- Authority
- CN
- China
- Prior art keywords
- dimension
- pel
- mark
- marked
- region
- 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.)
- Active
Links
Abstract
A kind of three-dimensional model dividing method, pel mark is carried out to the two-dimension projection of three-dimensional model, calculate the confidence level of pel mark, mark according to the pel of the confidence level calculated to three-dimensional model, the figure element dividing of same mark will be noted as to same parts further in three-dimensional model, can split accurately the three-dimensional model of Non-Manifold, split the parts obtained corresponding with the building block of the entity that three-dimensional model represents.And, above-mentioned three-dimensional model dividing method will split the calculating of pel mark and the confidence level of being correlated be transformed into two-dimension projection to three-dimensional model, further according to splitting three-dimensional model the result of two-dimension projection, the data processing amount in three-dimensional model cutting procedure significantly can be reduced.In addition, a kind of three-dimensional model segmenting system is also provided.
Description
[technical field]
The present invention relates to graphics process field, particularly a kind of three-dimensional model dividing method and system.
[background technology]
The pel of three-dimensional model is carried out significant classification mark, and will be noted as other figure element dividing of same class is further same building block, thus three-dimensional model is divided into multiple independently building block, is the basis of carrying out three-dimensional model process.A lot of tasks in Geometric Modeling, industry manufacture, cartoon making and texture mapping all depend on the result of three-dimensional model segmentation.Such as, the pel of the three-dimensional model of a chair is labeled as leg, handrail, seat cushion, mark that backrest is corresponding, will the figure element dividing of same mark be marked with to same building block further, thus obtains the model data of each parts of chair.
Three-dimensional model segmentation is a basic problem of graphics.Although have already been proposed a lot of method, traditional cutting techniques mainly designs for the three-dimensional model flowing shape.Then common a lot of three-dimensional models are all the Non-Manifolds be spliced by multiple fritter.
[summary of the invention]
Based on this, be necessary to provide a kind of three-dimensional model dividing method and the system that can split the three-dimensional model of Non-Manifold.
A kind of three-dimensional model dividing method, comprises the following steps S1 ~ S4:
S1: projection three-dimensional model under the first predetermined number predetermined angle, obtains the first predetermined number two-dimension projection of three-dimensional model;
S2: pel mark is carried out to the first predetermined number two-dimension projection and calculate pel mark confidence level;
S3: mark according to the pel of the confidence level calculated to three-dimensional model;
S4: the figure meta-tag according to three-dimensional model is split three-dimensional model, will be noted as the figure element dividing of same mark to same parts in three-dimensional model.
Wherein in an embodiment, in step S2, pel mark is carried out to the two-dimension projection under each predetermined angle and the step of confidence level calculating pel mark comprises the following steps S21 ~ S27:
S21: obtain the two dimension of object under described predetermined angle represented by described three-dimensional model and mark picture, the described two dimension pel marked in picture has been marked with the mark of pel generic, and the quantity that the described two dimension of acquisition has marked picture is the second predetermined number;
S22: described two-dimension projection and described two dimension have been marked picture and has carried out Region dividing;
S23: the region that the region in described two-dimension projection and described two dimension have marked in picture is mated, the region obtained in two-dimension projection has marked region corresponding in picture in described two dimension;
S24: calculating described two-dimensional projection picture difference of aliging with the distortion marked between picture of described two dimension is the distance sum that described two-dimensional projection picture and described two dimension have marked corresponding region between picture;
S25: the two dimension obtaining the second predetermined number has marked distortion alignment difference in picture and marked picture by the sort two dimension of the 3rd forward predetermined number of order from small to large;
S26: the two dimension of the 3rd predetermined number obtained has been marked the mark of pel in picture and moved on the pel in described two-dimension projection by gradation;
S27: calculate each pel in two-dimension projection described in each mark migration and be marked with the confidence level that two dimension has marked the mark correspondence of pel in picture.
Wherein in an embodiment, in step S22, some X-Y schemes are carried out Region dividing and comprise the following steps, this X-Y scheme refers to described two-dimension projection or described two dimension marks picture:
S221: every a line pel of X-Y scheme or each row pel are divided into an independent region;
S222: according to the distance between following formulae discovery adjacent area, a region is merged into by apart from minimum adjacent area, repeat the step of " according to the distance between following formulae discovery adjacent area; merge into a region by apart from minimum adjacent area ", until the region quantity comprised in X-Y scheme is not more than preset value;
Wherein:
D
iand D
i+1be two adjacent regions, Distant (D
i, D
i+1) be D
iwith D
i+1between distance, h
iand h
i+1be respectively D
iand D
i+1width, R [D
i] and R [D
i+1] be respectively and be positioned at D
iand D
i+1the line at middle part;
Two independents variable in above-mentioned function BiSH () are represented, then BiSH (α, β)=max (SH (α, β), SH (α with α and β
c, β
c)), α
cand β
cbe respectively in α and β the pel representing cavity;
Two independents variable in above-mentioned function SH () are represented, then with A, B
wherein,
H(A,B)=max
a∈A(min
b∈Bdist(a,b)),H(B,A)=max
b∈B(min
a∈Adist(b,a));
(b a) is the distance norm between a and b to dist.
Wherein in an embodiment, step S23 comprises the following steps:
S231: the 1st region described two dimension having been marked picture is corresponding with the 1st region of described two-dimension projection;
S232: the region of searching successively in described two-dimension projection according to region putting in order in described two-dimension projection has marked region corresponding in picture in described two dimension, wherein, searches the some region TD in two-dimension projection
jcomprise below the step in region corresponding in described two dimension has marked sheet on a map:
Obtain TD in two-dimension projection
jprevious region TD
j-1region SD corresponding in figure has been marked in described two dimension
i, and obtain the described two dimension region SD of this correspondence in mark figure
ia rear region SD
i+1;
Calculate TD
jwith SD
idistance be h
j× BiSH (R [TD
j], R [SD
i]), and calculate TD
jwith SD
i+1distance be h
j× BiSH (R [TD
j], R [SD
i+1]),
Wherein, h
jfor the width of TDj, R [TD
j], R [SD
i] and R [SD
i+1] be respectively and be positioned at TD
j, SD
iand SD
i+1the line at middle part;
Get TD
iand TD
i+1in with TD
japart from little region as TD
jregion corresponding in picture has been marked in two dimension;
Wherein, the described two dimension tandem that marked the region of picture and described two-dimension projection by region the calculating that puts in order in the drawings.
Wherein in an embodiment, in step S26, a certain two dimension is marked on pel that the mark of pel in picture moves in described two-dimension projection and has comprised the following steps:
The mark of the line described two dimension marked in the region of picture moves on the line in the region of the described two-dimension projection corresponding with this region, remembers that the region that described two dimension has marked picture is SD
i, note SD
ithe quantity of the line comprised is n1, note and SD
ithe region of corresponding described two-dimension projection is TD
j, note TD
jthe quantity of the line comprised is n2,
If n2>n1, then use SD
iin each line on mark mark TD
jmiddle n2/n1 bar line,
If n2<n1, then get SD
iin n2 bar line, with on this n2 bar line mark mark TD
jin n2 bar line;
Above-mentioned line has marked putting in order in picture and two-dimension projection according to line in two dimension carried out to the mark migration of line, the mark that two dimension has marked the line be arranged in front in picture is migrated on the line that is arranged in front in corresponding two-dimension projection, wherein, a certain bar line SL in picture has been marked by two dimension
imark mark two-dimension projection in a certain bar line TL
jcomprise the following steps:
Step S261: search SL
iin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at TL
jon corresponding position by TL
jsegmentation, and search TL
jin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at SL
ion corresponding position by SL
isegmentation;
Step S262: according to non-empty line segment at SL
iand TL
jon put in order SL
ithe TL that moves to of the mark of non-space line segment
jon non-empty line segment, SL
iin the mark of non-empty line segment that is arranged in front be migrated to corresponding TL
jin on the non-empty line segment that is arranged in front.
Wherein in an embodiment, step S262 is by SL
iin the mark of a certain non-empty line segment be migrated to TL
jthe non-empty line segment of middle correspondence comprises the following steps:
Judge SL
iin this non-empty line segment and TL
jwhether the length of the non-empty line segment of middle correspondence is identical, if not, then by SL
iin this non-empty line segment stretching to TL
jthe length of the non-empty line segment of middle correspondence is identical, if so, then directly enters next step;
According to pel putting in order SL on line segment
iin the mark of pel of this non-empty line segment move to TL
jon the pel of the non-empty line segment of middle correspondence, the mark of the pel be arranged in front is migrated on the pel that is arranged in front.
Wherein in an embodiment, according to pel TP a certain in two-dimension projection described in following formulae discovery in step S27
jbe marked with described two dimension and mark corresponding pel SP in picture
ithe confidence level C of mark correspondence
ij, note TL
jfor TP in two-dimension projection
jthe line at place, SL
ifor marking SP in picture
ithe line at place:
C
ij=exp(-(ci+cs(i,j)+cp(i,j)
2/σ
2)
Ci is TP
jplace two-dimension projection and SP
iplace two dimension has marked the distortion alignment difference and TP between picture
jthe business of the width of place two-dimension projection;
Cs (i, j)=BiSH (TL
j, SL
i), wherein, the definition of BiSH () function is identical with the definition of above-mentioned BiSH () function;
Cp (i, j) is TP
jat TL
jin positional alignment sequence valve and SP
iat SL
iin the absolute value of difference of positional alignment sequence valve;
σ is that Gauss supports base.
Wherein in an embodiment, note V is the pel set of three-dimensional model, and the mark that described two dimension has marked pel generic in picture comprises B
1... B
k..., B
n, n is the categorical measure that described two dimension has marked pel generic in picture;
Step S3 comprises the following steps:
Statistics pel μ is marked with the mark B of each classification
kconfidence level C (B
k, μ), k=1 ..., n;
Calculate pel μ and be marked with mark B
kprobability p (B
k| μ) be
Search the mark B making following energy function ξ reach minimum value
μ, μ ∈ V, B
μ∈ { B
1... B
k..., B
n}:
If three-dimensional model has complete connectedness, then
Wherein, E
1for the set on the limit that pel adjacent and connected in three-dimensional model is formed, E
2for the set on the limit that pel adjacent in three-dimensional model is formed, μ v is the limit that in V, pel μ and v is formed, l
μ vfor the length on the limit that pel μ and v in V is formed, θ
μ vfor in V pel μ and v formed positive dihedral angle, d (μ, v) for the Euclidean distance of pel μ and v in V, ω and λ be preset value;
The mark found is marked on the pel of the correspondence of three-dimensional model.
A kind of three-dimensional model segmenting system, comprising:
Two-dimension projection acquisition module, for projection three-dimensional model under the first predetermined number predetermined angle, obtains the first predetermined number two-dimension projection of three-dimensional model;
Two-dimensional primitive mark and confidence level computing module, for pel mark is carried out to the first predetermined number two-dimension projection and calculate pel mark confidence level;
Three-dimensional pel labeling module, for marking according to the pel of confidence level to three-dimensional model calculated;
Segmentation module, splits three-dimensional model for the figure meta-tag according to three-dimensional model, will be noted as the figure element dividing of same mark to same parts in three-dimensional model.
Wherein in an embodiment, described pel mark and confidence level computing module comprise:
Mark picture acquisition module, two dimension for obtaining object represented by described three-dimensional model under the corresponding predetermined angle of described two-dimension projection marks picture, the described two dimension pel marked in picture has been marked with the mark of pel generic, and the quantity that the described two dimension of acquisition has marked picture is the second predetermined number;
Region dividing module, carries out Region dividing for described two-dimension projection and described two dimension have been marked picture;
Region Matching module, mates for the region region in described two-dimension projection and described two dimension marked in picture, and the region obtained in two-dimension projection has marked region corresponding in picture in described two dimension;
Difference computation module is the distance sum that described two-dimensional projection picture and described two dimension have marked corresponding region between picture for calculating described two-dimensional projection picture difference of aliging with the distortion marked between picture of described two dimension;
Mark transferring module, two dimension for obtaining the second predetermined number to have marked in picture distortion alignment difference and has marked picture by the sort two dimension of the 3rd forward predetermined number of order from small to large, and the two dimension of the 3rd predetermined number obtained has been marked the mark of pel in picture and moved on the pel in described two-dimension projection by gradation;
Confidence level computing module, is marked with for calculating each pel in two-dimension projection described in each mark migration the confidence level that two dimension has marked the mark correspondence of pel in picture.
Wherein in an embodiment, some described two-dimension projections or described two dimension have been marked the process that picture carries out Region dividing and have been by Region dividing module:
Every a line pel of X-Y scheme or each row pel are divided into an independent region, this X-Y scheme refers to described two-dimension projection or described two dimension marks picture, according to the distance between following formulae discovery adjacent area, a region is merged into by apart from minimum adjacent area, repeat the step of " according to the distance between following formulae discovery adjacent area; merge into a region by apart from minimum adjacent area ", until the region quantity comprised in X-Y scheme is not more than preset value;
Wherein:
D
iand D
i+1be two adjacent regions, Distant (D
i, D
i+1) be D
iwith D
i+1between distance, h
iand h
i+1be respectively D
iand D
i+1width, R [D
i] and R [D
i+1] be respectively and be positioned at D
iand D
i+1the line at middle part;
Two independents variable in above-mentioned function BiSH () are represented, then with α and β
α
cand β
cbe respectively in α and β the pel representing cavity;
Two independents variable in above-mentioned function SH () are represented, then with A, B
wherein,
H(A,B)=max
a∈A(min
b∈Bdist(a,b)),H(B,A)=max
b∈B(min
a∈Adist(b,a));
(b a) is the distance norm between a and b to dist.
Wherein in an embodiment, the 1st region that Region Matching module is used for described two dimension to mark picture is corresponding with the 1st region of described two-dimension projection, the region of searching successively in described two-dimension projection according to region putting in order in described two-dimension projection has marked region corresponding in picture in described two dimension, wherein, the some region TD in two-dimension projection are searched
jthe process in region corresponding in described two dimension has marked sheet on a map is:
Obtain TD in two-dimension projection
jprevious region TD
j-1region SD corresponding in figure has been marked in described two dimension
i, and obtain the described two dimension region SD of this correspondence in mark figure
ia rear region SD
i+1; Calculate TD
jwith SD
idistance be h
j× BiSH (R [TD
j], R [SD
i]), and calculate TD
jwith SD
i+1distance be h
j× BiSH (R [TD
j], R [SD
i+1]),
Wherein, h
jfor the width of TDj, R [TD
j], R [SD
i] and R [SD
i+1] be respectively and be positioned at TD
j, SD
iand SD
i+1the line at middle part;
Get TD
iand TD
i+1in with TD
japart from little region as TD
jregion corresponding in picture has been marked in two dimension;
Wherein, the described two dimension tandem that marked the region of picture and described two-dimension projection by region the calculating that puts in order in the drawings.
Wherein in an embodiment, the process that a certain two dimension has marked on pel that the mark of pel in picture moves in described two-dimension projection by mark transferring module is:
The mark of the line described two dimension marked in the region of picture moves on the line in the region of the described two-dimension projection corresponding with this region, remembers that the region that described two dimension has marked picture is SD
i, note SD
ithe quantity of the line comprised is n1, note and SD
ithe region of corresponding described two-dimension projection is TD
j, note TD
jthe quantity of the line comprised is n2,
If n2>n1, then use SD
iin each line on mark mark TD
jmiddle n2/n1 bar line,
If n2<n1, then get SD
iin n2 bar line, with on this n2 bar line mark mark TD
jin n2 bar line;
Above-mentioned line has marked putting in order in picture and two-dimension projection according to line in two dimension carried out to the mark migration of line, the mark that two dimension has marked the line be arranged in front in picture is migrated on the line that is arranged in front in corresponding two-dimension projection, wherein, a certain bar line SL in picture has been marked by two dimension
imark mark two-dimension projection in a certain bar line TL
jprocess be:
Search SL
iin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at TL
jon corresponding position by TL
jsegmentation, and search TL
jin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at SL
ion corresponding position by SL
isegmentation;
According to non-empty line segment at SL
iand TL
jon put in order SL
ithe TL that moves to of the mark of non-space line segment
jon non-empty line segment, SL
iin the mark of non-empty line segment that is arranged in front be migrated to corresponding TL
jin on the non-empty line segment that is arranged in front.
Wherein in an embodiment, mark transferring module is by SL
iin the mark of a certain non-empty line segment be migrated to TL
jprocess on the non-empty line segment of middle correspondence is:
Judge SL
iin this non-empty line segment and TL
jwhether the length of the non-empty line segment of middle correspondence is identical, if not, then by SL
iin this non-empty line segment stretching to TL
jthe length of the non-empty line segment of middle correspondence is identical, if so, then directly enters next step;
According to pel putting in order SL on line segment
iin the mark of pel of this non-empty line segment move to TL
jon the pel of the non-empty line segment of middle correspondence, the mark of the pel be arranged in front is migrated on the pel that is arranged in front.
Wherein in an embodiment, according to pel TP a certain in two-dimension projection described in following formulae discovery in confidence level computing module
jbe marked with described two dimension and mark corresponding pel SP in picture
ithe confidence level C of mark correspondence
ij, note TL
jfor TP in two-dimension projection
jthe line at place, SL
ifor marking SP in picture
ithe line at place:
C
ij=exp(-(ci+cs(i,j)+cp(i,j)
2/σ
2)
Ci is TP
jplace two-dimension projection and SP
iplace two dimension has marked the distortion alignment difference and TP between picture
jthe business of the width of place two-dimension projection;
Cs (i, j)=BiSH (TL
j, SL
i), wherein, the definition of BiSH () function is identical with the definition of above-mentioned BiSH () function;
Cp (i, j) is TP
jat TL
jin positional alignment sequence valve and SP
iat SL
iin the absolute value of difference of positional alignment sequence valve;
σ is that Gauss supports base.
Wherein in an embodiment, note V is the pel set of three-dimensional model, and the mark that described two dimension has marked pel generic in picture comprises B
1... B
k..., B
n, n is the categorical measure that described two dimension has marked pel generic in picture;
Three-dimensional pel labeling module is marked with the mark B of each classification for adding up pel μ
kconfidence level C (B
k, μ), k=1 ..., n,
Calculate pel μ and be marked with mark B
kprobability p (B
k| μ) be
Search the mark B making following energy function ξ reach minimum value
μ, μ ∈ V, B
μ∈ { B
1... B
k..., B
n}:
If three-dimensional model has complete connectedness, then
Wherein, E
1for the set on the limit that pel adjacent and connected in three-dimensional model is formed, E
2for the set on the limit that pel adjacent in three-dimensional model is formed, μ v is the limit that in V, pel μ and v is formed, l
μ vfor the length on the limit that pel μ and v in V is formed, θ
μ vfor in V pel μ and v formed positive dihedral angle, d (μ, v) for the Euclidean distance of pel μ and v in V, ω and λ be preset value,
The mark found is marked on the pel of the correspondence of three-dimensional model.
Above-mentioned three-dimensional model dividing method and system, pel mark is carried out to the two-dimension projection of three-dimensional model, calculate the confidence level of pel mark, mark according to the pel of the confidence level calculated to three-dimensional model, the figure element dividing of same mark will be noted as to same parts further in three-dimensional model, can split accurately the three-dimensional model of Non-Manifold, split the parts obtained corresponding with the building block of the entity that three-dimensional model represents.And, said method and system will split the calculating of pel mark and the confidence level of being correlated be transformed into two-dimension projection to three-dimensional model, further according to splitting three-dimensional model the result of two-dimension projection, the data processing amount in three-dimensional model cutting procedure significantly can be reduced.
[accompanying drawing explanation]
Fig. 1 is the schematic flow sheet of the three-dimensional model dividing method in an embodiment;
Fig. 2 is the schematic flow sheet carrying out pel mark to the two-dimension projection under each predetermined angle in an embodiment and calculate the step of the confidence level of pel mark;
Fig. 3 is the schematic flow sheet of the step of in an embodiment, some X-Y schemes being carried out Region dividing;
Fig. 4 is the schematic flow sheet of the step S23 of Fig. 2 in an embodiment;
Fig. 5 has marked a certain bar line SL in picture by two dimension in an embodiment
imark mark two-dimension projection in a certain bar line TL
jthe schematic flow sheet of step;
Fig. 6 is the structural representation of the three-dimensional model segmenting system in an embodiment;
Fig. 7 is the structural representation of two-dimensional primitive mark and confidence level computing module in an embodiment.
[embodiment]
As shown in Figure 1, in one embodiment, a kind of three-dimensional model dividing method, comprises the following steps S1 ~ S4:
S1: projection three-dimensional model under the first predetermined number predetermined angle, obtains the first predetermined number two-dimension projection of three-dimensional model.
S2: pel mark is carried out to the first predetermined number two-dimension projection and calculate pel mark confidence level.
As shown in Figure 2, wherein, pel mark is carried out to the two-dimension projection under each predetermined angle and the step of confidence level calculating pel mark comprises the following steps S21 ~ S27:
S21: obtain the two dimension of object under described predetermined angle represented by described three-dimensional model and mark picture, the described two dimension pel marked in picture has been marked with the mark of pel generic, and the quantity that the described two dimension of acquisition has marked picture is the second predetermined number.
S22: described two-dimension projection and described two dimension have been marked picture and has carried out Region dividing.
As shown in Figure 3, in step S22, some X-Y schemes are carried out Region dividing and comprise the following steps S221 and S222, this X-Y scheme refers to described two-dimension projection or described two dimension marks picture:
S221: every a line pel of X-Y scheme or each row pel are divided into an independent region;
S222: calculate the distance between adjacent area, a region is merged into by apart from minimum adjacent area, repeat the step of " calculating the distance between adjacent area; merge into a region by apart from minimum adjacent area ", until the region quantity comprised in X-Y scheme is not more than preset value.Concrete, S222 is according to the distance between following formulae discovery adjacent area:
Wherein:
D
iand D
i+1be two adjacent regions, Distant (D
i, D
i+1) be D
iwith D
i+1between distance, h
iand h
i+1be respectively D
iand D
i+1width, R [D
i] and R [D
i+1] be respectively and be positioned at D
iand D
i+1the line at middle part;
Two independents variable in above-mentioned function BiSH () are represented, then with α and β
α
cand β
cbe respectively in α and β the pel representing cavity;
Two independents variable in above-mentioned function SH () are represented, then with A, B
wherein,
H(A,B)=max
a∈A(min
b∈Bdist(a,b)),H(B,A)=max
b∈B(min
a∈Adist(b,a));
(b a) is the distance norm between a and b to dist.
S23: the region that the region in described two-dimension projection and described two dimension have marked in picture is mated, the region obtained in two-dimension projection has marked region corresponding in picture in described two dimension;
As shown in Figure 4, step S23 comprises the following steps:
S231: the 1st region described two dimension having been marked picture is corresponding with the 1st region of described two-dimension projection;
S232: the region of searching successively in described two-dimension projection according to region putting in order in described two-dimension projection has marked region corresponding in picture in described two dimension, wherein, searches the some region TD in two-dimension projection
jcomprise below the step in region corresponding in described two dimension has marked sheet on a map:
(1) TD in two-dimension projection is obtained
jprevious region TD
j-1region SD corresponding in figure has been marked in described two dimension
i, and obtain the described two dimension region SD of this correspondence in mark figure
ia rear region SD
i+1;
(2) TD is calculated
jwith SD
idistance be h
j× BiSH (R [TD
j], R [SD
i]), and calculate TD
jwith SD
i+1distance be h
j× BiSH (R [TD
j], R [SD
i+1]),
Wherein, h
jfor TD
jwidth, R [TD
j], R [SD
i] and R [SD
i+1] be respectively and be positioned at TD
j, SD
iand SD
i+1the line at middle part;
(3) TD is got
iand TD
i+1in with TD
japart from little region as TD
jregion corresponding in picture has been marked in two dimension;
Wherein, the described two dimension tandem that marked the region of picture and described two-dimension projection by region the calculating that puts in order in the drawings.
Concrete, in one embodiment, step S23 can calculate the Region Matching matrix that described two dimension has marked the m × n of picture and described two-dimension projection, m is that described two dimension has been marked on a map the quantity in region that sheet comprises, n is the quantity in the region that described two-dimension projection comprises, the element M in Region Matching matrix
ijfor described two dimension has marked the distance in a jth region in i-th region and described two-dimension projection in picture,
M
ij=h
j×BiSH(R[TD
j],R[SD
i]),
Wherein, SD
iand TD
jbe respectively described two dimension and marked in picture that in i-th region and described two-dimension projection a jth region, R [TD
j] and R [SD
i] be respectively and be positioned at TD
jand SD
ithe line at middle part;
Further, by M
11as the point of impact, to turn left along the point of impact in Region Matching matrix or the next point of impact is searched in the lower left corner, such as, M
ijfor the point of impact, then compare M
i (j+1)and M
(i+1) (j+1)value, get wherein less one as the point of impact.All the point of impact is found at each row of Region Matching matrix.Obtain row subscript and the row subscript of matrix element corresponding to the point of impact, then in two-dimension projection, this row subscript corresponding region has marked region corresponding in picture in two dimension and has been region corresponding to this row subscript.
S24: calculating described two-dimensional projection picture difference of aliging with the distortion marked between picture of described two dimension is the distance sum that described two-dimensional projection picture and described two dimension have marked corresponding region between picture; This distance is the distance between the region of calculating in the step (2) of step S232.
S25: the two dimension obtaining the second predetermined number has marked distortion alignment difference in picture and marked picture by the sort two dimension of the 3rd forward predetermined number of order from small to large.
S26: the two dimension of the 3rd predetermined number obtained has been marked the mark of pel in picture and moved on the pel in described two-dimension projection by gradation.
In step S26, a certain two dimension is marked on pel that the mark of pel in picture moves in described two-dimension projection and comprised the following steps: the mark of the line described two dimension marked in the region of picture has moved on the line in the region of the described two-dimension projection corresponding with this region.
Concrete, remember that the region that described two dimension has marked picture is SD
i, note SD
ithe quantity of the line comprised is n1, note and SD
ithe region of corresponding described two-dimension projection is TD
j, note TD
jthe quantity of the line comprised is n2,
If n2>n1, then use SD
iin each line on mark mark TD
jmiddle n2/n1 bar line.If n2/n1 is not integer, then n2/n1 can be carried out round.
If n2<n1, then get SD
iin n2 bar line, with on this n2 bar line mark mark TD
jin n2 bar line.In one embodiment, desirable SD
iin the 1st bar of line, 1+n1/n2 bar line, 1+2n1/n2 article line mark TD
jin first, second and third line, the rest may be inferred.
Above-mentioned line has marked putting in order in picture and two-dimension projection according to line in two dimension carried out to the mark migration of line, and the mark that two dimension has marked the line be arranged in front in picture is migrated on the line that is arranged in front in corresponding two-dimension projection.As shown in Figure 5, wherein, a certain bar line SL in picture has been marked by two dimension
imark mark two-dimension projection in a certain bar line TL
jcomprise the following steps:
Step S261: search SL
iin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at TL
jon corresponding position by TL
jsegmentation, and search TL
jin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at SL
ion corresponding position by SL
isegmentation.
In advance two dimension can be marked picture and two-dimension projection is adjusted to unified size and form.Thus two dimension to have marked the length of the line in picture equal with the length of the line in two-dimension projection.Above-mentioned middle pel is at TL
jon correspondence position refer to middle pel at SL
iin arrangement position at TL
jon correspondence position.Such as, middle pel is at SL
iin arrangement position be 3, then at TL
jon arrangement position be that the place of 3 is by TL
jsegmentation.
Step S262: according to non-empty line segment at SL
iand TL
jon put in order SL
ithe TL that moves to of the mark of non-space line segment
jon non-empty line segment, SL
iin the mark of non-empty line segment that is arranged in front be migrated to corresponding TL
jin on the non-empty line segment that is arranged in front.
Concrete, by SL
iin the mark of a certain non-empty line segment be migrated to TL
jthe non-empty line segment of middle correspondence comprises the following steps:
(1) SL is judged
iin this non-empty line segment and TL
jwhether the length of the non-empty line segment of middle correspondence is identical, if not, then by SL
iin this non-empty line segment stretching to TL
jthe length of the non-empty line segment of middle correspondence is identical, if so, then directly enters next step;
(2) according to pel putting in order SL on line segment
iin the mark of pel of this non-empty line segment move to TL
jon the pel of the non-empty line segment of middle correspondence, the mark of the pel be arranged in front is migrated on the pel that is arranged in front.
S27: calculate each pel in two-dimension projection described in each mark migration and be marked with the confidence level that two dimension has marked the mark correspondence of pel in picture;
Concrete, according to pel TP a certain in two-dimension projection described in following formulae discovery in step S27
jbe marked with described two dimension and mark corresponding pel SP in picture
ithe confidence level C of mark correspondence
ij, note TL
jfor TP in two-dimension projection
jthe line at place, SL
ifor marking SP in picture
ithe line at place:
C
ij=exp(-(ci+cs(i,j)+cp(i,j)
2/σ
2)
Ci is TP
jplace two-dimension projection and SP
iplace two dimension has marked the distortion alignment difference and TP between picture
jthe business of the width of place two-dimension projection;
Cs (i, j)=BiSH (TL
j, SL
i), wherein, the definition of BiSH () function is identical with the definition of above-mentioned BiSH () function;
Cp (i, j) is TP
jat TL
jin positional alignment sequence valve and SP
iat SL
iin the absolute value of difference of positional alignment sequence valve;
σ is that Gauss supports base.
S3: mark according to the pel of the confidence level calculated to three-dimensional model.
Concrete, note V is the pel set of three-dimensional model, and the mark that described two dimension has marked pel generic in picture comprises B
1... B
k..., B
n, n is the categorical measure that described two dimension has marked pel generic in picture;
Step S3 comprises the following steps,
(1) the mark B that pel μ is marked with each classification is added up
kconfidence level C (B
k, μ), k=1 ..., n.
Concrete, if pel TP in two-dimension projection in step S26
jbe marked with two dimension and mark pel SP in picture
imark, pel TP
jpel corresponding is in the three-dimensional model μ, and pel SP
ibe labeled as B
k, then the pel TP will calculated in step S27
jbe marked with pel SP
ithe confidence level C of mark correspondence
ijbe added to C (B
k, μ).
(2) calculate pel μ and be marked with mark B
kprobability p (B
k| μ) be
(3) the mark B making following energy function ξ reach minimum value is searched
μ, μ ∈ V, B
μ∈ { B
1... B
k..., B
n}:
If three-dimensional model has complete connectedness, all connected between namely adjacent in three-dimensional model pel, then
Otherwise,
Wherein, E
1for the set on the limit that pel adjacent and connected in three-dimensional model is formed, E
2for the set on the limit that pel adjacent in three-dimensional model is formed, μ v is the limit that in V, pel μ and v is formed, l
μ vfor the length on the limit that pel μ and v in V is formed, θ
μ vfor the positive dihedral angle that pel μ and v in V is formed, d (μ, v) is the Euclidean distance of pel μ and v in V;
ω and λ is preset value, ω and λ is two energy being used for equilibrium criterion energetic portions, smooth based on connectedness
with the energy of the smooth based on distance
affect restrictive condition.
(4) mark found is marked on the pel of correspondence of three-dimensional model.
S4: the figure meta-tag according to three-dimensional model is split three-dimensional model, will be noted as the figure element dividing of same mark to same parts in three-dimensional model.
As shown in Figure 6, in one embodiment, a kind of three-dimensional model segmenting system, comprises two-dimension projection acquisition module 10, two-dimensional primitive mark and confidence level computing module 20, three-dimensional pel labeling module 30 and segmentation module 40, wherein:
Two-dimension projection acquisition module 10, for projection three-dimensional model under the first predetermined number predetermined angle, obtains the first predetermined number two-dimension projection of three-dimensional model.
Two-dimensional primitive mark and confidence level computing module 20 for pel mark is carried out to the first predetermined number two-dimension projection and calculate pel mark confidence level.
As shown in Figure 7, two-dimensional primitive mark and confidence level computing module 20 comprise and mark picture acquisition module 210, Region dividing module 220, Region Matching module 230, difference computation module 240, mark transferring module 250 and confidence level computing module 270, wherein:
Mark picture acquisition module 210 and mark picture for obtaining the two dimension of object under described predetermined angle represented by described three-dimensional model, the described two dimension pel marked in picture has been marked with the mark of pel generic, and the quantity that the described two dimension of acquisition has marked picture is the second predetermined number.
Described two-dimension projection and described two dimension have been marked picture and have carried out Region dividing by Region dividing module 220.
Concrete, some described two-dimension projections or described two dimension have been marked the process that picture carries out Region dividing by Region dividing module 220: every a line pel of X-Y scheme or each row pel are divided into an independent region, and this X-Y scheme refers to described two-dimension projection or described two dimension marks picture; Further, calculate the distance between adjacent area, a region is merged into by apart from minimum adjacent area, repeat the step of " calculating the distance between adjacent area; merge into a region by apart from minimum adjacent area ", until the region quantity comprised in X-Y scheme is not more than preset value.Concrete, Region dividing module 220 is according to the distance between following formulae discovery adjacent area:
Wherein:
D
iand D
i+1be two adjacent regions, Distant (D
i, D
i+1) be D
iwith D
i+1between distance, h
iand h
i+1be respectively D
iand D
i+1width, R [D
i] and R [D
i+1] be respectively and be positioned at D
iand D
i+1the line at middle part;
Two independents variable in above-mentioned function BiSH () are represented, then with α and β
α
cand β
cbe respectively in α and β the pel representing cavity;
Two independents variable in above-mentioned function SH () are represented, then with A, B
wherein,
H(A,B)=max
a∈A(min
b∈Bdist(a,b)),H(B,A)=max
b∈B(min
a∈Adist(b,a));
(b a) is the distance norm between a and b to dist.
Region Matching module 230 is mated for the region region in described two-dimension projection and described two dimension marked in picture, and the region obtained in two-dimension projection has marked region corresponding in picture in described two dimension.
Concrete, Region Matching module 230 is corresponding for the 1st region in the 1st region with described two-dimension projection that described two dimension have been marked picture; Further, the region of searching successively in described two-dimension projection according to region putting in order in described two-dimension projection has marked region corresponding in picture in described two dimension, wherein, searches the some region TD in two-dimension projection
jthe process in region corresponding in described two dimension has marked sheet on a map is:
(1) TD in two-dimension projection is obtained
jprevious region TD
j-1region SD corresponding in figure has been marked in described two dimension
i, and obtain the described two dimension region SD of this correspondence in mark figure
ia rear region SD
i+1;
(2) TD is calculated
jwith SD
idistance be h
j× BiSH (R [TD
j], R [SD
i]), and calculate TD
jwith SD
i+1distance be h
j× BiSH (R [TD
j], R [SD
i+1]),
Wherein, h
jfor TD
jwidth, R [TD
j], R [SD
i] and R [SD
i+1] be respectively and be positioned at TD
j, SD
iand SD
i+1the line at middle part;
(3) TD is got
iand TD
i+1in with TD
japart from little region as TD
jregion corresponding in picture has been marked in two dimension;
Wherein, the described two dimension tandem that marked the region of picture and described two-dimension projection by region the calculating that puts in order in the drawings.
Concrete, in one embodiment, Region Matching module 230 can calculate the Region Matching matrix that described two dimension has marked the m × n of picture and described two-dimension projection, m is that described two dimension has been marked on a map the quantity in region that sheet comprises, n is the quantity in the region that described two-dimension projection comprises, the element M in Region Matching matrix
ijfor described two dimension has marked the distance in a jth region in i-th region and described two-dimension projection in picture,
M
ij=h
j×BiSH(R[TD
j],R[SD
i]),
Wherein, SD
iand TD
jbe respectively described two dimension and marked in picture that in i-th region and described two-dimension projection a jth region, R [TD
j] and R [SD
i] be respectively and be positioned at TD
jand SD
ithe line at middle part;
Further, Region Matching module 230 can by M
11as the point of impact, to turn left along the point of impact in Region Matching matrix or the next point of impact is searched in the lower left corner, such as, M
ijfor the point of impact, then compare M
i (j+1)and M
(i+1) (j+1)value, get wherein less one as the point of impact.All the point of impact is found at each row of Region Matching matrix.Obtain row subscript and the row subscript of matrix element corresponding to the point of impact, then in two-dimension projection, this row subscript corresponding region has marked region corresponding in picture in two dimension and has been region corresponding to this row subscript.
Difference computation module 240 is the distance sum that described two-dimensional projection picture and described two dimension have marked corresponding region between picture for calculating described two-dimensional projection picture difference of aliging with the distortion marked between picture of described two dimension; This distance is the distance between the region of Region Matching module 230 calculating.
Mark transferring module 250 to have marked in picture distortion alignment difference for the two dimension obtaining the second predetermined number and has marked picture by the sort two dimension of the 3rd forward predetermined number of order from small to large, further, the two dimension of the 3rd predetermined number obtained has been marked the mark of pel in picture and has moved on the pel in described two-dimension projection by gradation.
Concrete, the process that a certain two dimension has marked on pel that the mark of pel in picture moves in described two-dimension projection by mark transferring module 250 is: the mark of the line described two dimension marked in the region of picture moves on the line in the region of the described two-dimension projection corresponding with this region.
Concrete, remember that the region that described two dimension has marked picture is SD
i, note SD
ithe quantity of the line comprised is n1, note and SD
ithe region of corresponding described two-dimension projection is TD
j, note TD
jthe quantity of the line comprised is n2,
If n2>n1, then use SD
iin each line on mark mark TD
jmiddle n2/n1 bar line.If n2/n1 is not integer, then n2/n1 can be carried out round.
If n2<n1, then get SD
iin n2 bar line, with on this n2 bar line mark mark TD
jin n2 bar line.In one embodiment, desirable SD
iin the 1st bar of line, 1+n1/n2 bar line, 1+2n1/n2 article line mark TD
jin first, second and third line, the rest may be inferred.
Above-mentioned line has marked putting in order in picture and two-dimension projection according to line in two dimension carried out to the mark migration of line, and the mark that two dimension has marked the line be arranged in front in picture is migrated on the line that is arranged in front in corresponding two-dimension projection.Wherein, a certain bar line SL in picture has been marked by two dimension
imark mark two-dimension projection in a certain bar line TL
jprocess be:
(1) SL is searched
iin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at TL
jon corresponding position by TL
jsegmentation, and search TL
jin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at SL
ion corresponding position by SL
isegmentation.
In advance two dimension can be marked picture and two-dimension projection is adjusted to unified size and form.Thus two dimension to have marked the length of the line in picture equal with the length of the line in two-dimension projection.Above-mentioned middle pel is at TL
jon correspondence position refer to middle pel at SL
iin arrangement position at TL
jon correspondence position.Such as, middle pel is at SL
iin arrangement position be 3, then at TL
jon arrangement position be that the place of 3 is by TL
jsegmentation.
(2) according to non-empty line segment at SL
iand TL
jon put in order SL
ithe TL that moves to of the mark of non-space line segment
jon non-empty line segment, SL
iin the mark of non-empty line segment that is arranged in front be migrated to corresponding TL
jin on the non-empty line segment that is arranged in front.
Concrete, mark transferring module 250 is by SL
iin the mark of a certain non-empty line segment be migrated to TL
jprocess on the non-empty line segment of middle correspondence is:
(1) SL is judged
iin this non-empty line segment and TL
jwhether the length of the non-empty line segment of middle correspondence is identical, if not, then by SL
iin this non-empty line segment stretching to TL
jthe length of the non-empty line segment of middle correspondence is identical, if so, then directly enters next step;
(2) according to pel putting in order SL on line segment
iin the mark of pel of this non-empty line segment move to TL
jon the pel of the non-empty line segment of middle correspondence, the mark of the pel be arranged in front is migrated on the pel that is arranged in front.
Confidence level computing module 270 is marked with for calculating each pel in two-dimension projection described in each mark migration the confidence level that two dimension has marked the mark correspondence of pel in picture;
Concrete, confidence level computing module 270 is according to pel TP a certain in two-dimension projection described in following formulae discovery
jbe marked with described two dimension and mark corresponding pel SP in picture
ithe confidence level C of mark correspondence
ij, note TL
jfor TP in two-dimension projection
jthe line at place, SL
ifor marking SP in picture
ithe line at place:
C
ij=exp(-(ci+cs(i,j)+cp(i,j)
2/σ
2)
Ci is TP
jplace two-dimension projection and SP
iplace two dimension has marked the distortion alignment difference and TP between picture
jthe business of the width of place two-dimension projection;
Cs (i, j)=BiSH (TL
j, SL
i), wherein, the definition of BiSH () function is identical with the definition of above-mentioned BiSH () function;
Cp (i, j) is TP
jat TL
jin positional alignment sequence valve and SP
iat SL
iin the absolute value of difference of positional alignment sequence valve;
σ is that Gauss supports base.
Three-dimensional pel labeling module 30 is for marking according to the pel of confidence level to three-dimensional model calculated.
Concrete, note V is the pel set of three-dimensional model, and the mark that described two dimension has marked pel generic in picture comprises B
1... B
k..., B
n, n is the categorical measure that described two dimension has marked pel generic in picture;
The process that three-dimensional pel labeling module 30 marks according to the pel of the confidence level calculated to three-dimensional model is:
(1) the mark B that pel μ is marked with each classification is added up
kconfidence level C (B
k, μ), k=1 ..., n.
Concrete, if pel TP in two-dimension projection
jbe labeled the upper two dimension of transferring module 250 mark and mark pel SP in picture
imark, pel TP
jpel corresponding is in the three-dimensional model μ, and pel SP
ibe labeled as B
k, then pel TP confidence level computing module 270 calculated
jbe marked with pel SP
ithe confidence level C of mark correspondence
ijbe added to C (B
k, μ).
(2) calculate pel μ and be marked with mark B
kprobability p (B
k| μ) be
(3) the mark B making following energy function ξ reach minimum value is searched
μ, μ ∈ V, B
μ∈ { B
1... B
k..., B
n}:
If three-dimensional model has complete connectedness, all connected between namely adjacent in three-dimensional model pel, then
Otherwise,
Wherein, E
1for the set on the limit that pel adjacent and connected in three-dimensional model is formed, E
2for the set on the limit that pel adjacent in three-dimensional model is formed, μ v is the limit that in V, pel μ and v is formed, l
μ vfor the length on the limit that pel μ and v in V is formed, θ
μ vfor the positive dihedral angle that pel μ and v in V is formed, d (μ, v) is the Euclidean distance of pel μ and v in V;
ω and λ is preset value, ω and λ is two energy being used for equilibrium criterion energetic portions, smooth based on connectedness
with the energy of the smooth based on distance
affect restrictive condition.
(4) mark found is marked on the pel of correspondence of three-dimensional model.
Segmentation module 40 is split three-dimensional model for the figure meta-tag according to three-dimensional model, will be noted as the figure element dividing of same mark to same parts in three-dimensional model.
Above-mentioned three-dimensional model dividing method and system, pel mark is carried out to the two-dimension projection of three-dimensional model, calculate the confidence level of pel mark, mark according to the pel of the confidence level calculated to three-dimensional model, the figure element dividing of same mark will be noted as to same parts further in three-dimensional model, can split accurately the three-dimensional model of Non-Manifold, split the parts obtained corresponding with the building block of the entity that three-dimensional model represents.And, said method and system will split the calculating of pel mark and the confidence level of being correlated be transformed into two-dimension projection to three-dimensional model, further according to splitting three-dimensional model the result of two-dimension projection, the data processing amount in three-dimensional model cutting procedure significantly can be reduced.
The above embodiment only have expressed several embodiment of the present invention, and it describes comparatively concrete and detailed, but therefore can not be interpreted as the restriction to the scope of the claims of the present invention.It should be pointed out that for the person of ordinary skill of the art, without departing from the inventive concept of the premise, can also make some distortion and improvement, these all belong to protection scope of the present invention.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.
Claims (14)
1. a three-dimensional model dividing method, comprises the following steps S1 ~ S4:
S1: projection three-dimensional model under the first predetermined number predetermined angle, obtains the first predetermined number two-dimension projection of three-dimensional model;
S2: pel mark is carried out to the first predetermined number two-dimension projection and calculate pel mark confidence level;
S3: mark according to the pel of the confidence level calculated to three-dimensional model;
S4: the figure meta-tag according to three-dimensional model is split three-dimensional model, will be noted as the figure element dividing of same mark to same parts in three-dimensional model;
Wherein, in step S2, pel mark is carried out to the two-dimension projection under each predetermined angle and the step of confidence level calculating pel mark comprises the following steps S21 ~ S27:
S21: obtain the two dimension of object under described predetermined angle represented by described three-dimensional model and mark picture, the described two dimension pel marked in picture has been marked with the mark of pel generic, and the quantity that the described two dimension of acquisition has marked picture is the second predetermined number;
S22: described two-dimension projection and described two dimension have been marked picture and has carried out Region dividing;
S23: the region that the region in described two-dimension projection and described two dimension have marked in picture is mated, the region obtained in two-dimension projection has marked region corresponding in picture in described two dimension;
S24: calculating described two-dimensional projection picture difference of aliging with the distortion marked between picture of described two dimension is the distance sum that described two-dimensional projection picture and described two dimension have marked corresponding region between picture;
S25: the two dimension obtaining the second predetermined number has marked distortion alignment difference in picture and marked picture by the sort two dimension of the 3rd forward predetermined number of order from small to large;
S26: the two dimension of the 3rd predetermined number obtained has been marked the mark of pel in picture and moved on the pel in described two-dimension projection by gradation;
S27: calculate each pel in two-dimension projection described in each mark migration and be marked with the confidence level that two dimension has marked the mark correspondence of pel in picture.
2. three-dimensional model dividing method according to claim 1, is characterized in that, in step S22, some X-Y schemes is carried out Region dividing and comprises the following steps, and this X-Y scheme refers to described two-dimension projection or described two dimension marks picture:
S221: every a line pel of X-Y scheme or each row pel are divided into an independent region;
S222: according to the distance between following formulae discovery adjacent area, a region is merged into by apart from minimum adjacent area, repeat the step of " according to the distance between following formulae discovery adjacent area; merge into a region by apart from minimum adjacent area ", until the region quantity comprised in X-Y scheme is not more than preset value;
Wherein:
D
iand D
i+1be two adjacent regions, Distant (D
i, D
i+1) be D
iwith D
i+1between distance, h
iand h
i+1be respectively D
iand D
i+1width, R [D
i] and R [D
i+1] be respectively and be positioned at D
iand D
i+1the line at middle part;
Two independents variable in above-mentioned function BiSH () are represented, then BiSH (α, β)=max (SH (α, β), SH (α with α and β
c, β
c)), α
cand β
cbe respectively in α and β the pel representing cavity;
Two independents variable in above-mentioned function SH () are represented, then SH (A, B)=max (H (A, B), H (B, A)) with A, B, wherein,
H(A,B)=max
a∈A(min
b∈Bdist(a,b)),H(B,A)=max
b∈B(min
a∈Adist(b,a));
(b a) is the distance norm between a and b to dist.
3. three-dimensional model dividing method according to claim 1, it is characterized in that, step S23 comprises the following steps:
S231: the 1st region described two dimension having been marked picture is corresponding with the 1st region of described two-dimension projection;
S232: the region of searching successively in described two-dimension projection according to region putting in order in described two-dimension projection has marked region corresponding in picture in described two dimension, wherein, searches the some region TD in two-dimension projection
jcomprise below the step in region corresponding in described two dimension has marked sheet on a map:
Obtain TD in two-dimension projection
jprevious region TD
j-1region SD corresponding in figure has been marked in described two dimension
i, and obtain the described two dimension region SD of this correspondence in mark figure
ia rear region SD
i+1;
Calculate TD
jwith SD
idistance be h
j× BiSH (R [TD
j], R [SD
i]), and calculate TD
jwith SD
i+1distance be h
j× BiSH (R [TD
j], R [SD
i+1]),
Wherein, h
jfor TD
jwidth, R [TD
j], R [SD
i] and R [SD
i+1] be respectively and be positioned at TD
j, SD
iand SD
i+1the line at middle part;
Get TD
iand TD
i+1in with TD
japart from little region as TD
jregion corresponding in picture has been marked in two dimension;
Wherein, the described two dimension tandem that marked the region of picture and described two-dimension projection by region the calculating that puts in order in the drawings.
4. three-dimensional model dividing method according to claim 1, is characterized in that, a certain two dimension has been marked on pel that the mark of pel in picture moves in described two-dimension projection and comprise the following steps in step S26:
The mark of the line described two dimension marked in the region of picture moves on the line in the region of the described two-dimension projection corresponding with this region, remembers that the region that described two dimension has marked picture is SD
i, note SD
ithe quantity of the line comprised is n1, note and SD
ithe region of corresponding described two-dimension projection is TD
j, note TD
jthe quantity of the line comprised is n2,
If n2>n1, then use SD
iin each line on mark mark TD
jmiddle n2/n1 bar line,
If n2<n1, then get SD
iin n2 bar line, with on this n2 bar line mark mark TD
jin n2 bar line;
Above-mentioned line has marked putting in order in picture and two-dimension projection according to line in two dimension carried out to the mark migration of line, the mark that two dimension has marked the line be arranged in front in picture is migrated on the line that is arranged in front in corresponding two-dimension projection, wherein, a certain bar line SL in picture has been marked by two dimension
imark mark two-dimension projection in a certain bar line TL
jcomprise the following steps:
Step S261: search SL
iin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at TL
jon corresponding position by TL
jsegmentation, and search TL
jin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at SL
ion corresponding position by SL
isegmentation;
Step S262: according to non-empty line segment at SL
iand TL
jon put in order SL
ithe TL that moves to of the mark of non-space line segment
jon non-empty line segment, SL
iin the mark of non-empty line segment that is arranged in front be migrated to corresponding TL
jin on the non-empty line segment that is arranged in front.
5. three-dimensional model dividing method according to claim 4, it is characterized in that, step S262 is by SL
iin the mark of a certain non-empty line segment be migrated to TL
jthe non-empty line segment of middle correspondence comprises the following steps:
Judge SL
iin this non-empty line segment and TL
jwhether the length of the non-empty line segment of middle correspondence is identical, if not, then by SL
iin this non-empty line segment stretching to TL
jthe length of the non-empty line segment of middle correspondence is identical, if so, then directly enters next step;
According to pel putting in order SL on line segment
iin the mark of pel of this non-empty line segment move to TL
jon the pel of the non-empty line segment of middle correspondence, the mark of the pel be arranged in front is migrated on the pel that is arranged in front.
6. three-dimensional model dividing method according to claim 1, is characterized in that, according to pel TP a certain in two-dimension projection described in following formulae discovery in step S27
jbe marked with described two dimension and mark corresponding pel SP in picture
ithe confidence level C of mark correspondence
ij, note TL
jfor TP in two-dimension projection
jthe line at place, SL
ifor marking SP in picture
ithe line at place:
C
ij=exp(-(ci+cs(i,j)+cp(i,j)
2/σ
2)
Ci is TP
jplace two-dimension projection and SP
iplace two dimension has marked the distortion alignment difference and TP between picture
jthe business of the width of place two-dimension projection;
Cs (i, j)=BiSH (TL
j, SL
i), wherein, the definition of BiSH () function is identical with the definition of above-mentioned BiSH () function;
Cp (i, j) is TP
jat TL
jin positional alignment sequence valve and SP
iat SL
iin the absolute value of difference of positional alignment sequence valve;
σ is that Gauss supports base.
7. three-dimensional model dividing method according to claim 6, is characterized in that, note V is the pel set of three-dimensional model, and the mark that described two dimension has marked pel generic in picture comprises B
1... B
k..., B
n, n is the categorical measure that described two dimension has marked pel generic in picture;
Step S3 comprises the following steps:
Statistics pel μ is marked with the mark B of each classification
kconfidence level C (B
k, μ), k=1 ..., n;
Calculate pel μ and be marked with mark B
kprobability p (B
k| μ) be
k=1 ..., n;
Search the mark B making following energy function ξ reach minimum value
μ, μ ∈ V, B
μ∈ { B
1... B
k..., B
n}:
If three-dimensional model has complete connectedness, then
otherwise,
Wherein, E
1for the set on the limit that pel adjacent and connected in three-dimensional model is formed, E
2for the set on the limit that pel adjacent in three-dimensional model is formed, μ v is the limit that in V, pel μ and v is formed, l
μ vfor the length on the limit that pel μ and v in V is formed, θ
μ vfor in V pel μ and v formed positive dihedral angle, d (μ, v) for the Euclidean distance of pel μ and v in V, ω and λ be preset value;
The mark found is marked on the pel of the correspondence of three-dimensional model.
8. a three-dimensional model segmenting system, is characterized in that, comprising:
Two-dimension projection acquisition module, for projection three-dimensional model under the first predetermined number predetermined angle, obtains the first predetermined number two-dimension projection of three-dimensional model;
Two-dimensional primitive mark and confidence level computing module, for pel mark is carried out to the first predetermined number two-dimension projection and calculate pel mark confidence level;
Three-dimensional pel labeling module, for marking according to the pel of confidence level to three-dimensional model calculated;
Segmentation module, splits three-dimensional model for the figure meta-tag according to three-dimensional model, will be noted as the figure element dividing of same mark to same parts in three-dimensional model;
Wherein, described pel mark and confidence level computing module comprise:
Mark picture acquisition module, two dimension for obtaining object represented by described three-dimensional model under the corresponding predetermined angle of described two-dimension projection marks picture, the described two dimension pel marked in picture has been marked with the mark of pel generic, and the quantity that the described two dimension of acquisition has marked picture is the second predetermined number;
Region dividing module, carries out Region dividing for described two-dimension projection and described two dimension have been marked picture;
Region Matching module, mates for the region region in described two-dimension projection and described two dimension marked in picture, and the region obtained in two-dimension projection has marked region corresponding in picture in described two dimension;
Difference computation module is the distance sum that described two-dimensional projection picture and described two dimension have marked corresponding region between picture for calculating described two-dimensional projection picture difference of aliging with the distortion marked between picture of described two dimension;
Mark transferring module, two dimension for obtaining the second predetermined number to have marked in picture distortion alignment difference and has marked picture by the sort two dimension of the 3rd forward predetermined number of order from small to large, and the two dimension of the 3rd predetermined number obtained has been marked the mark of pel in picture and moved on the pel in described two-dimension projection by gradation;
Confidence level computing module, is marked with for calculating each pel in two-dimension projection described in each mark migration the confidence level that two dimension has marked the mark correspondence of pel in picture.
9. three-dimensional model segmenting system according to claim 8, is characterized in that, some described two-dimension projections or described two dimension have been marked the process that picture carries out Region dividing and be by Region dividing module:
Every a line pel of X-Y scheme or each row pel are divided into an independent region, this X-Y scheme refers to described two-dimension projection or described two dimension marks picture, according to the distance between following formulae discovery adjacent area, a region is merged into by apart from minimum adjacent area, repeat the step of " according to the distance between following formulae discovery adjacent area; merge into a region by apart from minimum adjacent area ", until the region quantity comprised in X-Y scheme is not more than preset value;
Wherein:
D
iand D
i+1be two adjacent regions, Distant (D
i, D
i+1) be D
iwith D
i+1between distance, h
iand h
i+1be respectively D
iand D
i+1width, R [D
i] and R [D
i+1] be respectively and be positioned at D
iand D
i+1the line at middle part;
Two independents variable in above-mentioned function BiSH () are represented, then BiSH (α, β)=max (SH (α, β), SH (α with α and β
c, β
c)), α
cand β
cbe respectively in α and β the pel representing cavity;
Two independents variable in above-mentioned function SH () are represented, then SH (A, B)=max (H (A, B), H (B, A)) with A, B, wherein,
H(A,B)=max
a∈A(min
b∈Bdist(a,b)),H(B,A)=max
b∈B(min
a∈Adist(b,a));
(b a) is the distance norm between a and b to dist.
10. three-dimensional model segmenting system according to claim 8, it is characterized in that, the 1st region that Region Matching module is used for described two dimension to mark picture is corresponding with the 1st region of described two-dimension projection, the region of searching successively in described two-dimension projection according to region putting in order in described two-dimension projection has marked region corresponding in picture in described two dimension, wherein, the some region TD in two-dimension projection are searched
jthe process in region corresponding in described two dimension has marked sheet on a map is:
Obtain TD in two-dimension projection
jprevious region TD
j-1region SD corresponding in figure has been marked in described two dimension
i, and obtain the described two dimension region SD of this correspondence in mark figure
ia rear region SD
i+1; Calculate TD
jwith SD
idistance be h
j× BiSH (R [TD
j], R [SD
i]), and calculate TD
jwith SD
i+1distance be h
j× BiSH (R [TD
j], R [SD
i+1]),
Wherein, h
jfor the width of TDj, R [TD
j], R [SD
i] and R [SD
i+1] be respectively and be positioned at TD
j, SD
iand SD
i+1the line at middle part;
Get TD
iand TD
i+1in with TD
japart from little region as TD
jregion corresponding in picture has been marked in two dimension;
Wherein, the described two dimension tandem that marked the region of picture and described two-dimension projection by region the calculating that puts in order in the drawings.
11. three-dimensional model segmenting systems according to claim 8, is characterized in that, the process that a certain two dimension has marked on pel that the mark of pel in picture moves in described two-dimension projection by mark transferring module is:
The mark of the line described two dimension marked in the region of picture moves on the line in the region of the described two-dimension projection corresponding with this region, remembers that the region that described two dimension has marked picture is SD
i, note SD
ithe quantity of the line comprised is n1, note and SD
ithe region of corresponding described two-dimension projection is TD
j, note TD
jthe quantity of the line comprised is n2,
If n2>n1, then use SD
iin each line on mark mark TD
jmiddle n2/n1 bar line,
If n2<n1, then get SD
iin n2 bar line, with on this n2 bar line mark mark TD
jin n2 bar line;
Above-mentioned line has marked putting in order in picture and two-dimension projection according to line in two dimension carried out to the mark migration of line, the mark that two dimension has marked the line be arranged in front in picture is migrated on the line that is arranged in front in corresponding two-dimension projection, wherein, a certain bar line SL in picture has been marked by two dimension
imark mark two-dimension projection in a certain bar line TL
jprocess be:
Search SL
iin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at TL
jon corresponding position by TL
jsegmentation, and search TL
jin the middle pel of empty line segment folded by two non-empty line segments, at this middle pel at SL
ion corresponding position by SL
isegmentation;
According to non-empty line segment at SL
iand TL
jon put in order SL
ithe TL that moves to of the mark of non-space line segment
jon non-empty line segment, SL
iin the mark of non-empty line segment that is arranged in front be migrated to corresponding TL
jin on the non-empty line segment that is arranged in front.
12. three-dimensional model segmenting systems according to claim 11, is characterized in that, mark transferring module is by SL
iin the mark of a certain non-empty line segment be migrated to TL
jprocess on the non-empty line segment of middle correspondence is:
Judge SL
iin this non-empty line segment and TL
jwhether the length of the non-empty line segment of middle correspondence is identical, if not, then by SL
iin this non-empty line segment stretching to TL
jthe length of the non-empty line segment of middle correspondence is identical, if so, then directly enters next step;
According to pel putting in order SL on line segment
iin the mark of pel of this non-empty line segment move to TL
jon the pel of the non-empty line segment of middle correspondence, the mark of the pel be arranged in front is migrated on the pel that is arranged in front.
13. three-dimensional model segmenting systems according to claim 8, is characterized in that, according to pel TP a certain in two-dimension projection described in following formulae discovery in confidence level computing module
jbe marked with described two dimension and mark corresponding pel SP in picture
ithe confidence level C of mark correspondence
ij, note TL
jfor TP in two-dimension projection
jthe line at place, SL
ifor marking SP in picture
ithe line at place:
C
ij=exp(-(ci+cs(i,j)+cp(i,j)
2/σ
2)
Ci is TP
jplace two-dimension projection and SP
iplace two dimension has marked the distortion alignment difference and TP between picture
jthe business of the width of place two-dimension projection;
Cs (i, j)=BiSH (TL
j, SL
i), wherein, the definition of BiSH () function is identical with the definition of above-mentioned BiSH () function;
Cp (i, j) is TP
jat TL
jin positional alignment sequence valve and SP
iat SL
iin the absolute value of difference of positional alignment sequence valve;
σ is that Gauss supports base.
14. three-dimensional model segmenting systems according to claim 13, is characterized in that, note V is the pel set of three-dimensional model, and the mark that described two dimension has marked pel generic in picture comprises B
1... B
k..., B
n, n is the categorical measure that described two dimension has marked pel generic in picture;
Three-dimensional pel labeling module is marked with the mark B of each classification for adding up pel μ
kconfidence level C (B
k, μ), k=1 ..., n,
Calculate pel μ and be marked with mark B
kprobability p (B
k| μ) be
k=1 ..., n,
Search the mark B making following energy function ξ reach minimum value
μ, μ ∈ V, B
μ∈ { B
1... B
k..., B
n}:
If three-dimensional model has complete connectedness, then
otherwise,
Wherein, E
1for the set on the limit that pel adjacent and connected in three-dimensional model is formed, E
2for the set on the limit that pel adjacent in three-dimensional model is formed, μ v is the limit that in V, pel μ and v is formed, l
μ vfor the length on the limit that pel μ and v in V is formed, θ
μ vfor in V pel μ and v formed positive dihedral angle, d (μ, v) for the Euclidean distance of pel μ and v in V, ω and λ be preset value,
The mark found is marked on the pel of the correspondence of three-dimensional model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310138385.3A CN103218818B (en) | 2013-04-19 | 2013-04-19 | Three-dimensional model dividing method and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310138385.3A CN103218818B (en) | 2013-04-19 | 2013-04-19 | Three-dimensional model dividing method and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103218818A CN103218818A (en) | 2013-07-24 |
CN103218818B true CN103218818B (en) | 2016-01-20 |
Family
ID=48816561
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310138385.3A Active CN103218818B (en) | 2013-04-19 | 2013-04-19 | Three-dimensional model dividing method and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103218818B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP3361958B1 (en) | 2015-10-13 | 2023-01-25 | Mazor Robotics Ltd. | Global spinal alignment planning method |
CN106204749B (en) * | 2016-07-07 | 2019-03-29 | 北京航空航天大学 | The threedimensional model of sparse low-rank feature representation is divided into segmentation method |
CN106910252B (en) * | 2017-01-20 | 2018-05-22 | 东北石油大学 | A kind of online mask method of threedimensional model based on semantic space projective transformation and system |
CN111340963B (en) * | 2020-02-26 | 2024-02-09 | 陕西理工大学 | Element ordering system in space image layer |
CN111340956B (en) * | 2020-02-26 | 2024-02-06 | 陕西理工大学 | Space graph drawing method |
CN111968240B (en) * | 2020-09-04 | 2022-02-25 | 中国科学院自动化研究所 | Three-dimensional semantic annotation method of photogrammetry grid based on active learning |
CN117274883B (en) * | 2023-11-20 | 2024-01-26 | 南昌工程学院 | Target tracking method and system based on multi-head attention optimization feature fusion network |
-
2013
- 2013-04-19 CN CN201310138385.3A patent/CN103218818B/en active Active
Non-Patent Citations (5)
Title |
---|
A 2D driven 3D vessel segmentation algorithm for 3D digital subtraction angiography data;M Spiegel et al;《PHYSICS IN MEDICINE AND BIOLOGY》;20111231;第56卷;摘要,第3节 * |
Active Co-Analysis of a Set of Shapes;Yunhai Wang et al;《ACM Transactions on Graphics》;20121130;第31卷(第6期);第3、5节,图3、8 * |
Efficient Volume Exploration Using the Gaussian Mixture Model;Yunhai Wang et al;《IEEE TRANSACTIONS ON VISUALIZATION AND COMPUTER GRAPHICS》;20111130;第17卷(第11期);全文 * |
On Visual Similarity Based 3D Model Retrieval;Ding-Yun Chen et al;《EUROGRAPHICS 2003》;20031231;第22卷(第3期);全文 * |
The visual perception of 3D shape;James T. Todd;《TRENDS in Cognitive Sciences》;20040331;第8卷(第3期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN103218818A (en) | 2013-07-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103218818B (en) | Three-dimensional model dividing method and system | |
CN104036544B (en) | A kind of building roof method for reconstructing based on on-board LiDAR data | |
CN101719286B (en) | Multiple viewpoints three-dimensional scene reconstructing method fusing single viewpoint scenario analysis and system thereof | |
CN110188228B (en) | Cross-modal retrieval method based on sketch retrieval three-dimensional model | |
CN109615698A (en) | Multiple no-manned plane SLAM map blending algorithm based on the detection of mutual winding | |
CN104615638B (en) | A kind of distributed Density Clustering method towards big data | |
CN104866862A (en) | Strip steel surface area type defect identification and classification method | |
CN110378239A (en) | A kind of real-time traffic marker detection method based on deep learning | |
CN101882150B (en) | Three-dimensional model comparison and search method based on nuclear density estimation | |
CN101526944A (en) | Image retrieving comparison method | |
CN108305289B (en) | Three-dimensional model symmetry characteristic detection method and system based on least square method | |
CN103247225A (en) | Instant positioning and map building method and equipment | |
CN107067471A (en) | A kind of adaptive scanning speed method for improving pendant body model forming quality | |
CN107194984A (en) | Mobile terminal real-time high-precision three-dimensional modeling method | |
CN102903111B (en) | Large area based on Iamge Segmentation low texture area Stereo Matching Algorithm | |
CN109493344A (en) | A kind of semantic segmentation method of large-scale city three-dimensional scenic | |
CN113111978A (en) | Three-dimensional target detection system and method based on point cloud and image data | |
CN104574517A (en) | Processing method and device for boundary surface grid cell of three-dimensional model | |
CN110909778B (en) | Image semantic feature matching method based on geometric consistency | |
CN115861247A (en) | High-resolution remote sensing image contour multistage regularization method, system and application | |
Zhang et al. | RI-Fusion: 3D object detection using enhanced point features with range-image fusion for autonomous driving | |
Zhang et al. | Detecting, fitting, and classifying surface primitives for infrastructure point cloud data | |
CN107203759A (en) | A kind of branch's recursion road restructing algorithm based on two view geometries | |
CN102521877A (en) | Method for reconstructing Chinese ancient building meaning model and component gallery from single image | |
CN110634149B (en) | Non-rigid target characteristic point matching method for optical motion capture system |
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 |