CN102024036A - Three-dimensional object retrieval method and device based on hypergraphs - Google Patents
Three-dimensional object retrieval method and device based on hypergraphs Download PDFInfo
- Publication number
- CN102024036A CN102024036A CN 201010571681 CN201010571681A CN102024036A CN 102024036 A CN102024036 A CN 102024036A CN 201010571681 CN201010571681 CN 201010571681 CN 201010571681 A CN201010571681 A CN 201010571681A CN 102024036 A CN102024036 A CN 102024036A
- Authority
- CN
- China
- Prior art keywords
- hypergraph
- dimensional object
- views
- module
- cluster
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Abstract
The invention provides a three-dimensional object retrieval method and device based on hypergraphs, wherein the method comprises the following steps: computing a distance matrix among all views of a three-dimensional object in a database; clustering all the views according to the distance matrix to obtain a plurality of clustering results, and constructing a plurality of hypergraphs corresponding to the three-dimensional object according to the clustering results; fusing the hypergraphs to form a fused hypergraph, analyzing the fused hypergraph, and establishing the relationship between the fused hypergraph and the three-dimensional object according to the analytic result; and retrieving the three-dimensional object according to the relationship. By using the method provided by the invention, the problem of low retrieval accuracy caused by complex three-dimensional object information is well solved. By establishing a model through hypergraphs, the method can effectively analyze the relationship of the three-dimensional object, thereby obtaining more accurate and effective retrieval effect.
Description
Technical field
The present invention relates to three dimensional object and handle three dimensional object analysis technical field, particularly a kind of retrieving three-dimensional objects method based on hypergraph.
Background technology
The quick growth of 3 D stereo object data has been accelerated in the progress of computing machine and multimedia technology.In the last few years, three dimensional object was increasingly extensive in the application in multiple fields such as computer-aided manufacturing, virtual reality, medical science and amusement, therefore, and become all the more important of retrieving three-dimensional objects method fast and effectively.
The describing method of traditional three dimensional object mainly is based on dummy model, but uses the process that traditional three dimensional object describing method needing usually when the true three-dimension object represented to carry out three-dimensional reconstruction.Because the calculated amount of three-dimensional reconstruction is bigger, this makes traditional three dimensional object describing method can not well be applied to the analysis of true three-dimension object and handles.
Along with the fast development of camera technology, more method is paid close attention to and three dimensional object analysis based on many views.This method based on many views is described the information of three dimensional object by view more than a group, and then finishes the further work such as retrieval of 3 D stereo object.
Because a three dimensional object comprising a large amount of many views, how using therefore that relevance that many views carry out three dimensional object describes is problem of difficulty.(D.Y.Chen in the method that in European graphics meeting, proposed in 2003, X.P.Tian, Y.T.Shen, and M.Ouhyoung.On visual similarity based 3d model retrieval.Computer Graphics Forum) light field descriptor (Lighting Filed Descriptor) has been proposed, carry out data acquisition by camera array in 20 vertex position settings of regular dodecahedron, obtain many group views and describe original three dimensional object, these views are described the space structure information of three dimensional object from different angles, on the other hand, this method mates to come to the coupling between the 3 D stereo object at so many views array.With the Zernike square of the view of two-value and the feature that the fourier descriptor feature is used as view, yet, in this method camera array there is the fixing requirement that is provided with.2007 on international IEEE multimedia transactions (T.F.Ansary, M.Daoudi, and J.P.Vandeborre, " A bayesian 3-d search engine using adaptive views clustering, " IEEETransactions on Multimedia, vol.9, no.1, pp.78-88,2007.) a kind of 3 D stereo object search method based on Bayesian analysis is proposed, wherein view obtains and also is to use 320 fixing camera arrays.This method at first obtains 320 original images, here at original view, the selected characteristics of image of Zernike square of 49 dimensions, this method is at first carried out representational view and is selected from original view, by calculating to the overall similarity between the view, carry out K mean iterative cluster, wherein each step all attempts existing classification results is carried out cluster again, and wherein K is chosen for 2.Here, Bayes's information prepares to be used to judge the effect and the stop condition of cluster, in ensuing processing, only representational view just is applied in the concrete retrieval analysis, by the Bayesian probability analysis between the view being obtained the degree of correlation between the whole three dimensional object, thereby finish the retrieval work based on view of 3 D stereo object.
These traditional three dimensional object analytical approachs based on view are mainly carried out direct or indirect methods such as coupling by the many views to three dimensional object and are carried out comparison between the three dimensional object.But because the complicacy of three dimensional object information, this makes that the method for views registered of direct applying three-dimensional object can not the effectively analysis of carrying out the three dimensional object correlativity.
Summary of the invention
Purpose of the present invention is intended to solve at least one of above-mentioned technological deficiency.
For achieving the above object, one aspect of the present invention has proposed a kind of retrieving three-dimensional objects method based on hypergraph, and this method is carried out modeling by using hypergraph to 3-D view, thereby to carrying out correlation analysis between the three dimensional object.
For this reason, the present invention proposes a kind of retrieving three-dimensional objects method, may further comprise the steps: the distance matrix in the computational data storehouse between all views of three dimensional object based on hypergraph; According to described distance matrix described all views are carried out cluster obtaining a plurality of cluster results, and make up a plurality of hypergraphs of described three dimensional object correspondence according to described a plurality of cluster results; Described a plurality of hypergraphs are merged forming a hypergraph after the fusion, and the hypergraph after the described fusion is analyzed, and set up relevance between the described three dimensional object according to analysis result; With according to the described three dimensional object of described correlation retrieval.
In one embodiment of the invention, distance matrix in the described computational data storehouse between all views of three dimensional object further comprises: with Zernike Moments is that characteristics of image carries out feature extraction to obtain described feature extraction result to described all views; Use Euclidean distance according to described feature extraction result and calculate distance between any two views, the distance calculation between described all views finishes, and obtains the distance matrix between described all views.
In one embodiment of the invention, describedly described all views are carried out cluster to obtain a plurality of cluster results according to distance matrix, further comprise: adopt K mean cluster method that described all views are carried out cluster, wherein, described cluster result changes according to the difference of described K value.
In one embodiment of the invention, the described a plurality of hypergraphs that make up described three dimensional object correspondence according to a plurality of cluster results, further comprise: the corresponding views set with described three dimensional object is the super limit of described hypergraph, connection is the summit of corresponding hypergraph with each described three dimensional object, to form a plurality of hypergraphs.
In one embodiment of the invention, described a plurality of hypergraphs are merged to form a hypergraph after the fusion, and the hypergraph after the described fusion analyzed, and set up relevance between the described three dimensional object according to analysis result, further comprise: described a plurality of hypergraphs are averaged fusion, to form a hypergraph after the fusion; Analyze according to the label between the summit of the hypergraph of default objective function after,, wherein, satisfy the described summit that relevance requires and have similar label to obtain in the database relevance between the three dimensional object arbitrarily to described fusion.
Another aspect of the present invention has also proposed a kind of device based on retrieving three-dimensional objects, comprising: distance matrix computing module, described distance matrix computing module are used for the distance matrix between all views of computational data storehouse three dimensional object; Hypergraph makes up module, and described hypergraph makes up module and is used for according to described distance matrix described all views being carried out cluster obtaining a plurality of cluster results, and makes up a plurality of hypergraphs of described three dimensional object correspondence according to described a plurality of cluster results; Relating module, described relating module are used for described a plurality of hypergraphs are merged forming a hypergraph after the fusion, and the hypergraph after the described fusion is analyzed, and set up relevance between the described three dimensional object according to analysis result; And retrieval module, described retrieval module is used for according to the described three dimensional object of described correlation retrieval.
In one embodiment of the invention, distance matrix in the described computational data storehouse between all views of three dimensional object further comprises: with Zernike Moments is that characteristics of image carries out feature extraction to obtain described feature extraction result to described all views; Use Euclidean distance according to described feature extraction result and calculate distance between any two views.
In one embodiment of the invention, described hypergraph makes up module and comprises the cluster module and make up module that wherein, described cluster module is used to adopt K mean cluster method that described all views are carried out cluster, wherein, described cluster result changes according to the difference of described K value; Described structure module is used for being the super limit of described hypergraph with the corresponding views set of described three dimensional object that connecting with each described three dimensional object is the summit of corresponding hypergraph, to form a plurality of hypergraphs.
In one embodiment of the invention, described relating module comprises Fusion Module and the related module of setting up, and wherein, described Fusion Module is used for described a plurality of hypergraphs are averaged fusion, to form a hypergraph after the fusion; Described association is set up module and is used for analyzing according to the label between the summit of the hypergraph of default objective function after to described fusion, to obtain the relevance between any three dimensional object in the database, wherein, satisfy the described summit that relevance requires and have similar label.
By method of the present invention, it is low effectively to solve the retrieval accuracy that the complexity of three dimensional object information brings, shortcomings such as computation complexity height.This method is carried out modeling by hypergraph, can carry out the correlation analysis of three dimensional object effectively, thus can obtain more accurate, more efficiently retrieving three-dimensional objects effect.
Aspect that the present invention adds and advantage part in the following description provide, and part will become obviously from the following description, or recognize by practice of the present invention.
Description of drawings
Above-mentioned and/or additional aspect of the present invention and advantage are from obviously and easily understanding becoming the description of embodiment below in conjunction with accompanying drawing, wherein:
Fig. 1 is the process flow diagram based on the retrieving three-dimensional objects method of hypergraph of the embodiment of the invention;
Fig. 2 for the comparison diagram of the result for retrieval of the method result for retrieval of using the embodiment of the invention and other three kinds of search methods and
Fig. 3 is the retrieving three-dimensional objects structure drawing of device based on hypergraph of the embodiment of the invention.
Embodiment
Describe whole embodiment of the present invention below in detail, the example of described embodiment is shown in the drawings, and wherein identical from start to finish or similar label is represented identical or similar elements or the element with identical or similar functions.Below by the embodiment that is described with reference to the drawings is exemplary, only is used to explain the present invention, and can not be interpreted as limitation of the present invention.
The present invention be directed to the poor accuracy of existing retrieving three-dimensional objects, the deficiency that is difficult to analyze and a kind of retrieving three-dimensional objects method of proposing based on hypergraph, this method is carried out modeling by using hypergraph to three dimensional object, from according to correlation analysis result between the three dimensional object three dimensional object being retrieved.
Below in conjunction with accompanying drawing the retrieving three-dimensional objects method based on hypergraph that the embodiment of the invention proposes is described in detail.
As shown in Figure 1, be the process flow diagram based on the retrieving three-dimensional objects method of hypergraph of the embodiment of the invention, in an embodiment of the present invention, this method may further comprise the steps:
Step S101, the distance matrix in the computational data storehouse between all views of three dimensional object.
Particularly, adopt the characteristics of image of Zernike Moments as three dimensional object view in the database, according to feature extraction algorithm view is carried out Zernike Moments feature extraction, thereby obtain the feature extraction result of all views, then, use the Zernike Moments feature extraction result who obtains, and adopt Euclidean distance as computed range, view applications distances computing method are calculated distance between any two views, thereby calculate the distance value between any two views.Similarly,, adopt this distance calculating method that the distance calculation between all views is finished, thereby obtain the distance matrix between all views for all views.
Step S102 carries out cluster obtaining a plurality of cluster results according to described distance matrix to described all views, and makes up a plurality of hypergraphs of described three dimensional object correspondence according to described a plurality of cluster results.
Particularly, use K mean cluster method that all views are carried out cluster, wherein by setting different K values, generate different cluster results, and on the basis of every group of cluster result, with the summit of each three dimensional object as hypergraph, and each the view set that obtains is as a super limit of this hypergraph, be used for connecting the summit of hypergraph, thereby form a plurality of hypergraphs.
In specific embodiments of the invention, as: G
1=(V
1, E
1, w
1), G
2=(V
2, E
2, w
2) ... G
κ=(V
κ, E
κ, w
κ) be used for representing K hypergraph obtaining, wherein, in specific embodiment, the K value can be got the positive integer greater than 1.According to instantiation of the present invention, at the definition of hypergraph, H
1, H
2, L H
κ, be used for representing G
1=(V
1, E
1, w
1), G
2=(V
2, E
2, w
2) ... G
κ=(V
κ, E
κ, w
κ) correlation matrix of hypergraph, D
V1, D
V2, L D
V κBe used for representing the summit degree matrix of above-mentioned hypergraph, and D
E1, D
E2, L D
E κBe used for representing the limit degree matrix of above-mentioned hypergraph.
Wherein, the H matrix is (with H
iBe example) method for building up as follows:
Wherein, according to above-mentioned formula as can be known, when a summit belonged to a super limit, H matrix correspondence position was 1, otherwise correspondence position is 0.
According to embodiments of the invention, the method for building up of two other matrix is as follows:
d(v
i)＝∑
e∈Eh(v
i，e)；
Step S103 merges forming a hypergraph after the fusion described a plurality of hypergraphs, and the hypergraph after the described fusion is analyzed, and sets up relevance between the described three dimensional object according to analysis result.
Particularly, at first all hypergraphs that obtain are merged, form the hypergraph after the fusion.Furtherly, at first all hypergraphs are averaged fusion, that is to say that each hypergraph all has identical weight, use the objective function that is provided with then the label between each node on the hypergraph is analyzed, thereby obtain the relevance between any three dimensional object in the database.In an embodiment of the present invention, this retrieving three-dimensional objects method requires have the summit of more correlativitys to have similar label.
In specific embodiments of the invention, the vector that makes label is f, and the objective function on the hypergraph after definition is merged is:
Wherein
Wherein,
Order
Wherein
Thus.Can obtain Ω (f)=f
TΔ f.
R
Emp(f) be empiric risk, it done as giving a definition:
Wherein, y is existing label vector, and the label vector of f for optimizing.
Therefore, analyze on this hypergraph to optimizing following formula:
Φ (f)=f
TΔ f+ λ || f-y||
2, λ＞0 wherein.
By computing, can obtain following result:
As can be seen from the results, at a given three dimensional object that is used to retrieve, corresponding element is set to 1 among the y, and other elements are set to 0.The final correlativity that obtains f as other three dimensional objects and this searching object in the database is used for retrieval.
Step S104 is according to the described three dimensional object of described correlation retrieval.
Understand for method of the present invention being had more clearly, below just search method of the present invention is described in detail in conjunction with instantiation.
[embodiment]
Present embodiment is for selecting the 3D object database based on image collection of National Taiwan University, and each object has been chosen 500 objects altogether as experimental data base by 20 graphical representations.In test, respectively with each object as object to be retrieved, retrieve, and analyze last integrated retrieval effect.
At first, adopt the characteristics of image of Zernike Moments as three dimensional object view in the 3D object database, according to feature extraction algorithm view is carried out Zernike Moments feature extraction, thereby obtain the feature extraction result of all views, then, use the Zernike Moments feature extraction result who obtains, and adopt Euclidean distance as computed range, view applications distances computing method are calculated distance between any two views, thereby calculate the distance value between any two views.Similarly,, adopt this distance calculating method that the distance calculation between all views is finished, thereby obtain the distance matrix between all views for all views.
Then, use K mean cluster method that all are attempted to carry out cluster, wherein, generate different cluster results by setting different K values.In this example, K respectively value be 50,100,200,400,600,1000,1500,2000 and 3000.And on the basis of every group of cluster result, with the summit of each object as hypergraph, and each the view set that obtains is used for connecting the summit of hypergraph, thereby forms a plurality of hypergraphs as a super limit of this hypergraph.Here G
1=(V
1, E
1, w
1), G
2=(V
2, E
2, w
2) ... G
κ=(V
κ, E
κ, w
κ) be used for representing κ hypergraph obtaining.At the definition of hypergraph, H
1, H
2, L H
κ, D
V1, D
V2, L D
V κ... D
E1, D
E2, L D
E κBe used for representing correlation matrix, summit degree matrix and the Bian Du matrix of these hypergraphs.
Wherein, the H matrix is (with H
iBe example) method for building up as follows:
Wherein, according to above-mentioned formula as can be known, when a summit belonged to a super limit, H matrix correspondence position was 1, otherwise correspondence position is 0.
According to embodiments of the invention, the method for building up of two other matrix is as follows:
d(v
i)＝∑
e∈Eh(v
i，e)；
At last, all hypergraphs that obtain are merged, form the hypergraph after the fusion.Furtherly, at first all hypergraphs are averaged fusion, that is to say that each hypergraph all has identical weight, use the objective function that is provided with then the label between each node on the hypergraph is analyzed, thereby obtain the relevance between any three dimensional object in the database.In an embodiment of the present invention, this retrieving three-dimensional objects method requires have the summit of more correlativitys to have similar label.
Concrete step describes in conjunction with formula:
The vector that makes label is f, and the objective function on the hypergraph after definition is merged is:
Wherein
Wherein,
Order
Wherein
Thus.We can obtain Ω (f)=f
TΔ f.
R
Emp(f) be empiric risk, it done as giving a definition:
Wherein, y is existing label vector, and the label vector of f for optimizing.
Therefore, analyze on this hypergraph to optimizing following formula:
Φ (f)=f
TΔ f+ λ || f-y||
2, λ＞0 wherein.
By computing, can obtain following result:
As can be seen from the results, at a given three dimensional object that is used to retrieve, corresponding element is set to 1 among the y, and other elements are set to 0.The final correlativity that obtains f as other three dimensional objects and this searching object in the database is used for retrieval.
The retrieving three-dimensional objects result of present embodiment is the comparison diagram of the result for retrieval of the method result for retrieval of using the embodiment of the invention and other three kinds of search methods as shown in Figure 2.Provided recall ratio-precision ratio curve among Fig. 2, first method 203 is calculated the Hausdorff distance (HAUS) in two groups of three dimensional object views, second method 204 calculate all images distance in two groups of views and average (MEAN), the third method 202 is at first calculated the minor increment in each view of each view of searching object and another object, again to all apart from summation (SumMin).And the search method 201 (Hypergraph) of the embodiment of the invention is on the retrieval effectiveness of three dimensional object, with classic method mutually specific energy obtain better effect.
In an embodiment of the present invention, also proposed a kind of retrieving three-dimensional objects device, as shown in Figure 3, be the structural drawing based on the retrieving three-dimensional objects device of hypergraph of the embodiment of the invention based on hypergraph.Should comprise that distance matrix computing module 310, hypergraph made up module 320, relating module 330 and retrieval module 340 based on the retrieving three-dimensional objects device 300 of hypergraph.Wherein, distance matrix computing module 310 is used for the distance matrix between all views of computational data storehouse three dimensional object; Hypergraph makes up module 320 and is used for according to described distance matrix described all views being carried out cluster obtaining a plurality of cluster results, and makes up a plurality of hypergraphs of described three dimensional object correspondence according to described a plurality of cluster results; Relating module 330 is used for described a plurality of hypergraphs are merged forming a hypergraph after the fusion, and the hypergraph after the described fusion is analyzed, and sets up relevance between the described three dimensional object according to analysis result; Retrieval module 340 is used for according to the described three dimensional object of described correlation retrieval.
Further, hypergraph structure module 320 comprises cluster module 321 and makes up module 322.Wherein, cluster module 321 is used to adopt K mean cluster method that described all views are carried out cluster, and wherein, described cluster result changes according to the difference of described K value; Make up module 322 and be used for being the super limit of described hypergraph that connecting with each described three dimensional object is the summit of corresponding hypergraph, to form a plurality of hypergraphs with the corresponding views set of described three dimensional object.
Relating module 330 comprises Fusion Module 331 and the related module 332 of setting up.Wherein, Fusion Module 331 is used for described a plurality of hypergraphs are averaged fusion, to form a hypergraph after the fusion; Association is set up module 332 and is used for analyzing according to the label between the summit of the hypergraph of default objective function after to described fusion, to obtain the relevance between any three dimensional object in the database, wherein, satisfy the described summit that relevance requires and have similar label.
Retrieving three-dimensional objects method and apparatus by the present invention's proposition based on hypergraph, can change traditional direct three dimensional object correlation method based on views registered, correlativity by bottom level view, use hypergraph and carry out modeling, and further use the correlativity of the higher level of the study acquisition three dimensional object on the hypergraph.And this method can effectively be avoided following two kinds of problems, first kind problem is when having only the minority view similar between the similar three dimensional object, application can produce wrong result based on the method for the direct coupling of view, and the second class problem is to produce wrong result when having individual image closely similar between two non-homogeneous object.Certainly, this method can the more effective correlation analysis that carries out three dimensional object, thereby obtains the better retrieval result.In addition, the method simplicity of design that the present invention proposes is easy to realize.
Although illustrated and described embodiments of the invention, for the ordinary skill in the art, be appreciated that without departing from the principles and spirit of the present invention and can carry out multiple variation, modification, replacement and modification that scope of the present invention is by claims and be equal to and limit to these embodiment.
Claims (9)
1. the retrieving three-dimensional objects method based on hypergraph is characterized in that, may further comprise the steps:
Distance matrix in the computational data storehouse between all views of three dimensional object;
According to described distance matrix described all views are carried out cluster obtaining a plurality of cluster results, and make up a plurality of hypergraphs of described three dimensional object correspondence according to described a plurality of cluster results;
Described a plurality of hypergraphs are merged forming a hypergraph after the fusion, and the hypergraph after the described fusion is analyzed, and set up relevance between the described three dimensional object according to analysis result; With
According to the described three dimensional object of described correlation retrieval.
2. the retrieving three-dimensional objects method based on hypergraph as claimed in claim 1 is characterized in that, the distance matrix in the described computational data storehouse between all views of three dimensional object further comprises:
With Zernike Moments is that characteristics of image carries out feature extraction to obtain described feature extraction result to described all views;
Use Euclidean distance according to described feature extraction result and calculate distance between any two views, the distance calculation between described all views finishes, and obtains the distance matrix between described all views.
3. the retrieving three-dimensional objects method based on hypergraph as claimed in claim 1 is characterized in that, describedly according to distance matrix described all views is carried out cluster to obtain a plurality of cluster results, further comprises:
Adopt K mean cluster method that described all views are carried out cluster, wherein, described cluster result changes according to the difference of described K value.
4. the retrieving three-dimensional objects method based on hypergraph as claimed in claim 3 is characterized in that, describedly makes up a plurality of hypergraphs of described three dimensional object correspondence according to a plurality of cluster results, further comprises:
Corresponding views set with described three dimensional object is the super limit of described hypergraph, and connecting with each described three dimensional object is the summit of corresponding hypergraph, to form a plurality of hypergraphs.
5. the retrieving three-dimensional objects method based on hypergraph as claimed in claim 1, it is characterized in that, described a plurality of hypergraphs are merged to form a hypergraph after the fusion, and the hypergraph after the described fusion analyzed, and set up relevance between the described three dimensional object according to analysis result, further comprise:
Described a plurality of hypergraphs are averaged fusion, to form a hypergraph after the fusion;
Analyze according to the label between the summit of the hypergraph of default objective function after,, wherein, satisfy the described summit that relevance requires and have similar label to obtain in the database relevance between the three dimensional object arbitrarily to described fusion.
6. the retrieving three-dimensional objects device based on hypergraph is characterized in that, comprising:
Distance matrix computing module, described distance matrix computing module are used for the distance matrix between all views of computational data storehouse three dimensional object;
Hypergraph makes up module, and described hypergraph makes up module and is used for according to described distance matrix described all views being carried out cluster obtaining a plurality of cluster results, and makes up a plurality of hypergraphs of described three dimensional object correspondence according to described a plurality of cluster results;
Relating module, described relating module are used for described a plurality of hypergraphs are merged forming a hypergraph after the fusion, and the hypergraph after the described fusion is analyzed, and set up relevance between the described three dimensional object according to analysis result; With
Retrieval module, described retrieval module are used for according to the described three dimensional object of described correlation retrieval.
7. the retrieving three-dimensional objects device based on hypergraph as claimed in claim 6 is characterized in that, the distance matrix in the described computational data storehouse between all views of three dimensional object further comprises:
With Zernike Moments is that characteristics of image carries out feature extraction to obtain described feature extraction result to described all views;
Calculate distance between any two views according to described feature extraction result and Euclidean distance.
8. the retrieving three-dimensional objects device based on hypergraph as claimed in claim 6, it is characterized in that, described hypergraph makes up module and comprises the cluster module and make up module, wherein, described cluster module is used to adopt K mean cluster method that described all views are carried out cluster, wherein, described cluster result changes according to the difference of described K value; Described structure module is used for being the super limit of described hypergraph with the corresponding views set of described three dimensional object that connecting with each described three dimensional object is the summit of corresponding hypergraph, to form a plurality of hypergraphs.
9. the retrieving three-dimensional objects device based on hypergraph as claimed in claim 6, it is characterized in that described relating module comprises Fusion Module and the related module of setting up, wherein, described Fusion Module is used for described a plurality of hypergraphs are averaged fusion, to form a hypergraph after the fusion; Described association is set up module and is used for analyzing according to the label between the summit of the hypergraph of default objective function after to described fusion, to obtain the relevance between any three dimensional object in the database, wherein, satisfy the described summit that relevance requires and have similar label.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010571681A CN102024036B (en) | 2010-11-29 | 2010-11-29 | Three-dimensional object retrieval method and device based on hypergraphs |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201010571681A CN102024036B (en) | 2010-11-29 | 2010-11-29 | Three-dimensional object retrieval method and device based on hypergraphs |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102024036A true CN102024036A (en) | 2011-04-20 |
| CN102024036B CN102024036B (en) | 2012-09-05 |
Family
ID=43865333
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201010571681A Active CN102024036B (en) | 2010-11-29 | 2010-11-29 | Three-dimensional object retrieval method and device based on hypergraphs |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102024036B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708589A (en) * | 2012-05-15 | 2012-10-03 | 北京航空航天大学 | Three-dimensional target multi-viewpoint view modeling method on basis of feature clustering |
| CN103258061A (en) * | 2013-06-09 | 2013-08-21 | 北京工商大学 | Area enclosure hypergraph visualization method based on interpolation algorithm |
| CN103955524A (en) * | 2014-05-09 | 2014-07-30 | 合肥工业大学 | Event-related socialized image searching algorithm based on hypergraph model |
| CN104820992A (en) * | 2015-05-19 | 2015-08-05 | 北京理工大学 | hypergraph model-based remote sensing image semantic similarity measurement method and device |
| CN105279524A (en) * | 2015-11-04 | 2016-01-27 | 盐城工学院 | High-dimensional data clustering method based on unweighted hypergraph segmentation |
| CN103870845B (en) * | 2014-04-08 | 2017-02-15 | 重庆理工大学 | Novel K value optimization method in point cloud clustering denoising process |
| CN107145519A (en) * | 2017-04-10 | 2017-09-08 | 浙江大学 | A kind of image retrieval and mask method based on hypergraph |
| CN107731283A (en) * | 2017-10-23 | 2018-02-23 | 清华大学 | A kind of image radio system based on more subspace modelings |
| CN109215741A (en) * | 2018-08-20 | 2019-01-15 | 中国矿业大学 | Oncogene based on double hypergraph regularizations expresses modal data double focusing class method |
| CN109446385A (en) * | 2018-11-14 | 2019-03-08 | 中国科学院计算技术研究所 | A kind of method of equipment map that establishing Internet resources and the application method of the equipment map |
| CN109829413A (en) * | 2019-01-25 | 2019-05-31 | 清华大学 | A kind of stereoscopic vision object recognition system based on quickly more figure fusion study |
| CN109857895A (en) * | 2019-01-25 | 2019-06-07 | 清华大学 | Stereoscopic vision search method and system based on polycyclic road view convolutional neural networks |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1181557A (en) * | 1996-10-30 | 1998-05-13 | 三菱电机株式会社 | Information searching device |
| US20090040225A1 (en) * | 2007-07-20 | 2009-02-12 | Fujitsu Limited | Apparatus and method for three-dimensional model retrieval |
-
2010
- 2010-11-29 CN CN201010571681A patent/CN102024036B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1181557A (en) * | 1996-10-30 | 1998-05-13 | 三菱电机株式会社 | Information searching device |
| US20090040225A1 (en) * | 2007-07-20 | 2009-02-12 | Fujitsu Limited | Apparatus and method for three-dimensional model retrieval |
Non-Patent Citations (1)
| Title |
|---|
| 《Electronics Letters》 20081230 Xiao,Q.K et al 3d object retrieval approach based on directed acyclic graph lightfield feature 847-848 1-9 第44卷, 第14期 2 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102708589A (en) * | 2012-05-15 | 2012-10-03 | 北京航空航天大学 | Three-dimensional target multi-viewpoint view modeling method on basis of feature clustering |
| CN102708589B (en) * | 2012-05-15 | 2014-10-29 | 北京航空航天大学 | Three-dimensional target multi-viewpoint view modeling method on basis of feature clustering |
| CN103258061A (en) * | 2013-06-09 | 2013-08-21 | 北京工商大学 | Area enclosure hypergraph visualization method based on interpolation algorithm |
| CN103258061B (en) * | 2013-06-09 | 2016-03-02 | 北京工商大学 | A kind of region enclosed hypergraph method for visualizing based on interpolation algorithm |
| CN103870845B (en) * | 2014-04-08 | 2017-02-15 | 重庆理工大学 | Novel K value optimization method in point cloud clustering denoising process |
| CN103955524A (en) * | 2014-05-09 | 2014-07-30 | 合肥工业大学 | Event-related socialized image searching algorithm based on hypergraph model |
| CN103955524B (en) * | 2014-05-09 | 2017-02-22 | 合肥工业大学 | Event-related socialized image searching algorithm based on hypergraph model |
| CN104820992A (en) * | 2015-05-19 | 2015-08-05 | 北京理工大学 | hypergraph model-based remote sensing image semantic similarity measurement method and device |
| CN104820992B (en) * | 2015-05-19 | 2017-07-18 | 北京理工大学 | A kind of remote sensing images Semantic Similarity measure and device based on hypergraph model |
| CN105279524A (en) * | 2015-11-04 | 2016-01-27 | 盐城工学院 | High-dimensional data clustering method based on unweighted hypergraph segmentation |
| CN107145519A (en) * | 2017-04-10 | 2017-09-08 | 浙江大学 | A kind of image retrieval and mask method based on hypergraph |
| CN107145519B (en) * | 2017-04-10 | 2020-07-03 | 浙江大学 | Image retrieval and annotation method based on hypergraph |
| CN107731283A (en) * | 2017-10-23 | 2018-02-23 | 清华大学 | A kind of image radio system based on more subspace modelings |
| CN109215741A (en) * | 2018-08-20 | 2019-01-15 | 中国矿业大学 | Oncogene based on double hypergraph regularizations expresses modal data double focusing class method |
| CN109215741B (en) * | 2018-08-20 | 2022-02-08 | 中国矿业大学 | Tumor gene expression profile data double clustering method based on double hypergraph regularization |
| CN109446385A (en) * | 2018-11-14 | 2019-03-08 | 中国科学院计算技术研究所 | A kind of method of equipment map that establishing Internet resources and the application method of the equipment map |
| CN109446385B (en) * | 2018-11-14 | 2022-06-14 | 中国科学院计算技术研究所 | Method for establishing network resource equipment map and using method of equipment map |
| CN109829413A (en) * | 2019-01-25 | 2019-05-31 | 清华大学 | A kind of stereoscopic vision object recognition system based on quickly more figure fusion study |
| CN109857895A (en) * | 2019-01-25 | 2019-06-07 | 清华大学 | Stereoscopic vision search method and system based on polycyclic road view convolutional neural networks |
| CN109857895B (en) * | 2019-01-25 | 2020-10-13 | 清华大学 | Stereo vision retrieval method and system based on multi-loop view convolutional neural network |
| CN109829413B (en) * | 2019-01-25 | 2020-11-06 | 清华大学 | Stereoscopic vision object identification system based on rapid multi-image fusion learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102024036B (en) | 2012-09-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102024036B (en) | Three-dimensional object retrieval method and device based on hypergraphs | |
| Assfalg et al. | Content-based retrieval of 3-D objects using spin image signatures | |
| Archambault et al. | Topolayout: Multilevel graph layout by topological features | |
| CN101201845B (en) | Method for searching three-dimensional model based on axis point-set hierarchical helix information | |
| Tao et al. | Partial retrieval of CAD models based on the gradient flows in Lie group | |
| CN105205135B (en) | A kind of 3D model retrieval methods and its retrieval device based on topic model | |
| CN106844620B (en) | View-based feature matching three-dimensional model retrieval method | |
| CN103699578B (en) | Image retrieval method based on spectrum analysis | |
| CN111027140B (en) | Airplane standard part model rapid reconstruction method based on multi-view point cloud data | |
| CN101477529A (en) | Three-dimensional object retrieval method and apparatus | |
| Zhou et al. | 2D compressive sensing and multi-feature fusion for effective 3D shape retrieval | |
| CN111078916A (en) | Cross-domain three-dimensional model retrieval method based on multi-level feature alignment network | |
| CN108595631B (en) | Three-dimensional CAD model double-layer retrieval method based on graph theory | |
| CN108537887A (en) | Sketch based on 3D printing and model library 3-D view matching process | |
| CN111460193B (en) | Three-dimensional model classification method based on multi-mode information fusion | |
| CN102708589B (en) | Three-dimensional target multi-viewpoint view modeling method on basis of feature clustering | |
| CN115082716A (en) | Multi-source point cloud rough matching algorithm for road fine reconstruction | |
| Tabia et al. | Local visual patch for 3D shape retrieval | |
| Su et al. | 3d-assisted image feature synthesis for novel views of an object | |
| CN101894267B (en) | Three-dimensional object characteristic view selection method | |
| CN102073738B (en) | Intelligent retrieval view selection-based three-dimensional object retrieval method and device | |
| CN102110159A (en) | CAD three-dimensional model retrieval method and system | |
| Gao et al. | Representative views re-ranking for 3d model retrieval with multi-bipartite graph reinforcement model | |
| CN109272013B (en) | Similarity measurement method based on learning | |
| CN107766562B (en) | Hierarchical measurement method for similarity of complex planar entity scene with holes |
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 |