CN101719140B - Figure retrieving method - Google Patents

Abstract

The invention discloses a figure retrieving method, comprising the steps of: (1) modeling a polygon; establishing a three-dimensional grid model base; (2) matching three-dimensional grid models with two-dimensional images or figures; (3) extracting frameworks of the three-dimensional grid models; (4) retrieving the three-dimensional models according to the frameworks; (5) extracting characteristic points of the three-dimensional grid models; (6) computing control points of the three-dimensional grid models; (7) computing frequency spectra of the control points of the three-dimensional grid models; and (8) computing the similarity of the frequency spectra and retrieving a corresponding figure according to the similarity. The technical scheme is more convenient for retrieval and supports multimode retrieval.

Description

A kind of figure retrieving method

Technical field

The present invention relates to the graph processing technique field, be specifically related to a kind of figure retrieving method.

Background technology

Network and computer graphics are infiltrated in the daily life gradually, and people no longer are satisfied with can only see two dimensional image on network.On the other hand, along with the develop rapidly of the increasingly mature and computer hardware technique of dimensional Modeling Technology, the quantity of three-dimensional model has had tremendous growth in nearest 10 years.With respect to two dimensional image, people can browse own part interested from arbitrarily angled, so more receive the people to like that purposes is more extensive.Online game and animation are technological, Web education is technological, all can not lack these medium of three-dimensional picture based on the information service gordian technique of Web and the research of product and hot fields such as database and data mining technology.

Make full use of existing three-dimensional modeling data resource, can alleviate the workload of design new model greatly, also can promote simultaneously the circulation of three-dimensional data and in the application in each field.Yet, how in the 3 d model library of magnanimity, to search own interested model fast, for 3 d model library is set up the problem that search engine is a difficulty.The figure retrieving method of prior art is according to how much contents three-dimensional model to be carried out systematic searching, and the user can not pass through the search interface retrieval request expression easily.

Summary of the invention

The technical matters that the present invention will solve provides a kind of figure retrieving method; Can overcome the deficiency of prior art; Realize the multi-modal retrieval of figure; Make that retrieval is more convenient, improve present multimedia search technology and the blank of filling up present network three-dimensional picture search engine, promote the realization of intelligent multi-modal search engine of future generation.

Technical scheme provided by the invention is following:

The present invention provides a kind of figure retrieving method, comprising:

1) sets up the three-dimensional grid model storehouse;

2) when user input be X-Y scheme or image the time; Mate with the profile of three-dimensional grid model in the three-dimensional grid model storehouse; According to matching parameter three-dimensional grid model is projected to two-dimensional space; Obtain the two dimensional image or the figure of projection, calculate two dimensional image or the figure of figure and input or the degree of correlation between the image that projection obtains then, retrieval obtains three-dimensional grid model according to the degree of correlation;

3) when user input be three-dimensional grid model the time, the three-dimensional grid model of input is carried out skeletal extraction, according to the three-dimensional grid model skeleton that extracts, preliminary search obtains three-dimensional grid model in the three-dimensional grid model storehouse;

The three-dimensional grid model of the three-dimensional grid model that 4) retrieval is obtained and user's input carries out feature point extraction; Replace original three-dimensional grid model; Carry out triangulation again; Cut-off rule behind the subdivision is carried out piecewise fitting, obtain the reference mark of original three-dimensional grid model, according to topological structure frequency domain transform is carried out at the reference mark then;

5) user who calculates imports the similarity between the reference mark frequency domain coordinate figure of reference mark frequency domain coordinate figure and the three-dimensional grid model in the three-dimensional grid model storehouse of three-dimensional grid model, goes out graph of a correspondence according to similarity retrieval;

Wherein, the topological structure according to three-dimensional grid model in the step 4) carries out frequency domain transform to the reference mark that obtains, and comprising:

(1) with the mould of reference mark sorted in the reference mark of coarse grids to the vector of grid element center;

(2) obtain the Kirchhoff matrix from the network topology relation

K＝D-A (6)

D is a diagonal matrix, and the element Dii on its diagonal line is corresponding with the valency of vertex v i, and A is the adjacency matrix of grid;

The Kirchhoff matrix is carried out n the proper vector w that characteristic value decomposition obtains _iCarry out ascending order and arrange the n*n mapping matrix W of composition;

(3) 3 vectors of the volume coordinate at sorted n reference mark structure before before:

X＝(x ₁，x ₂，…，x _n)，Y＝(y ₁，y ₂，…，y _n)，Z＝(z ₁，z ₂，…，z _n) (8)

These 3 vector projections are obtained frequency domain vector to the proper vector base W:

$\left\{\begin{array}{c}{X}_s=\mathrm{WX}\\ Y_s=\mathrm{WY}\\ Z_s=\mathrm{WZ}\end{array}\right.---\left(9\right)$

The amplitude S of the frequency spectrum that each summit is corresponding _iComputing formula is:

$S_i=\sqrt{{\left|\right|X_s\left|\right|}^2+{\left|\right|Y_s\left|\right|}^2+{\left|\right|Z_s\left|\right|}^2}---\left(11\right).$

Preferably, the three-dimensional grid model storehouse in the step 1) to various three-dimensional data forms reorganize with the polygon modeling after obtain.

Preferably; The skeletal extraction process is: at first set up progressive grid representation for the grid model of input; Then progressive grid is constantly carried out the limit conversion of collapsing, if a limit does not have adjacent triangle in the process of collapsing, then this limit is labeled as the skeleton limit; And remain into the end of collapsing, the skeleton of the final limit component model that obtains always.

Preferably, the three-dimensional grid model of the three-dimensional grid model that obtains for retrieval in the step 4) and user's input carries out feature point extraction according to its spatial form, replaces original three-dimensional grid model with umbilical cord point as unique point.

Preferably, according to unique point three-dimensional grid model is carried out triangulation in the step 4), the cut-off rule behind the subdivision is carried out piecewise fitting, obtain cut-point, with the reference mark of these cut-points as original three-dimensional grid model.

The present invention has following beneficial effect:

(1) can be according to single two dimensional image retrieval three-dimensional picture, the user can import the picture of common forms such as bmp, jpeg, tiff, can retrieve graph of a correspondence through the inventive method.

(2) method for expressing of the three-dimensional model of support input is more extensive, all is suitable for for triangular mesh data, cloud data, volume data or polygonal grid model.

(3) adopted hierarchical search, more accurate sooner.At first the skeleton according to three-dimensional picture carries out coarse search, adopts the further feature extractive technique accurately to search for to the three-dimensional picture in the Search Results, has guaranteed real-time and accuracy.

Description of drawings

In order to be illustrated more clearly in the embodiment of the invention or technical scheme of the prior art; To do to introduce simply to the accompanying drawing of required use in embodiment or the description of the Prior Art below; Obviously, the accompanying drawing in describing below only is some embodiments of the present invention, for those of ordinary skills; Under the prerequisite of not paying creative work, can also obtain other accompanying drawing according to these accompanying drawings.

Fig. 1 is that the user imports X-Y scheme or treatment of picture synoptic diagram;

Fig. 2 is a coupling process flow diagram three-dimensional and two dimension;

Fig. 3 is the processing flow chart when importing the 3D grid data for the user.

Embodiment

To combine the accompanying drawing in the embodiment of the invention below, the technical scheme in the embodiment of the invention is carried out clear, intactly description, obviously, described embodiment only is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment among the present invention, those of ordinary skills are not making all other embodiment that obtained under the creative work prerequisite, all belong to the scope of the present invention's protection.

The inventive method is mainly studied the multi-modal retrieval of figure; Its principle is the characteristic according to three-dimensional picture; Through information such as two dimensional image, X-Y scheme, three-dimensional models, calculate the figure of user's input or the similarity between image and the three-dimensional picture, thereby realize the multi-modal retrieval of three-dimensional picture.Here multi-modal being meant supports that the user inquires about with two dimensional image, X-Y scheme and the three-dimensional model of common form.

Aspect of the present invention extracts very little data volume from three-dimensional data, and with its principal character as corresponding figure or image, can retrieve according to this principal character.This characteristic does not receive the influence of factors such as noise, similarity transformation, different resolution sampling basically.

The inventive method mainly comprises following 8 links:

(1) at first the three-dimensional modeling data of various data layouts is reorganized, carry out the polygon modeling, obtain three-dimensional grid model, and set up the three-dimensional grid model storehouse.

(2) when user input be X-Y scheme or image the time; Mate with the profile of three-dimensional grid model in the three-dimensional grid model storehouse; According to matching parameter three-dimensional grid model is projected to two-dimensional space then; Obtain the two dimensional image or the figure of projection, calculate two dimensional image or the figure of figure and input or the degree of correlation between the image that projection obtains then, retrieve three-dimensional grid model according to the degree of correlation.

(3) when user input be three-dimensional grid model the time, the three-dimensional grid model of input is carried out skeletal extraction.

(4) according to the three-dimensional grid model skeleton that extracts, in the three-dimensional grid model storehouse, carry out preliminary search fast, obtain the three-dimensional grid model that preliminary search goes out.

(5) three-dimensional grid model that preliminary search is gone out according to the graphics theory and the three-dimensional grid model of user's input carry out feature point extraction according to its spatial form, replace original three-dimensional grid model, significantly reduce data volume.

(6) according to the unique point of extracting three-dimensional grid model is carried out triangulation, the cut-off rule behind the subdivision is carried out piecewise fitting, obtain cut-point, the reference mark of these cut-points as original three-dimensional grid model.

(7) topological structure according to three-dimensional grid model carries out frequency domain transform to the reference mark, the frequency domain coordinate figure of controlled point.

(8) user who calculates imports the similarity between the reference mark frequency domain coordinate figure of reference mark frequency domain coordinate figure and the three-dimensional grid model in the three-dimensional grid model storehouse of three-dimensional grid model, goes out graph of a correspondence according to similarity retrieval.

Technical characterstic of the present invention mainly embodies as follows:

(1) system can be according to single two dimensional image retrieval three-dimensional picture.The user can import the picture of common forms such as bmp, jpeg, tiff, system will be from picture cutting object and extract profile information, mate with the three-dimensional picture in the three-dimensional grid model storehouse, return the three-dimensional grid model of coupling to the user.

(2) method for expressing of the three-dimensional model of system's support input is more extensive.Triangular mesh data, cloud data, volume data or polygonal grid model all are suitable for.

(3) system adopts hierarchical search.At first the skeleton according to three-dimensional picture carries out coarse search, adopts Feature Extraction Technology accurately to search for again to the three-dimensional picture in the Search Results, has guaranteed real-time and accuracy.

Explain further details in the face of the present invention down.

The present invention supports multi-modal figure retrieving method key step to comprise:

(1) carries out the polygon modeling, set up the three-dimensional grid model storehouse in advance;

(2) coupling of three-dimensional grid model and two dimensional image or figure;

(3) extraction of three-dimensional grid model skeleton;

(4) carry out three-dimensional model search according to skeleton;

(5) three-dimensional grid model feature point extraction;

(6) calculating at three-dimensional grid model reference mark;

(7) calculating of three-dimensional grid model reference mark frequency spectrum;

(8) calculate the frequency spectrum similarity, go out graph of a correspondence according to similarity retrieval.

Below respectively above-mentioned steps is described in detail.

(1) carries out the polygon modeling, set up the three-dimensional grid model storehouse in advance.

The present invention carries out the polygon modeling according to the three-dimensional grid model data, structure polygon three-dimensional grid model.The polygon modeling is to utilize many polygon simulation curved surfaces to carry out, and polygon is many more, and then model approaches true curved surface more.The polygon modeling is to be easy to a kind of modeling technique of realizing the most again, and can obtain high-precision model, adopts the network of triangle case form usually.

For the convenience and the unification of following description, utilize mathematic sign to provide the definition of three-dimensional grid model.Grid model M={V, C} is made up of vertex set V and annexation set C, wherein gathers V and comprises N vertex v _iAnd the coordinate figure on each summit is by (x _i, y _i, z _i) confirm, promptly

V＝{v _i}，i＝0，1，…，N-1，v _i＝(x _i，y _i，z _i) (1)

And annexation set C is expressed as

C＝{{i _k，j _k}} _{k＝0，…m-1}，0≤i _k≤N-1，0≤j _k≤N-1 (2)

Here { i _k, j _kRepresent by i _kIndividual summit and j _kThe k bar limit that individual summit is confirmed.

(2) coupling of three-dimensional grid model and two dimensional image or figure

Fig. 1 is that the user imports X-Y scheme or treatment of picture synoptic diagram.

As shown in Figure 1, for single image/figure, carry out outline with the model bank figure, carry out the 2D projection again, carry out figure coupling relatedness computation then.

Under many circumstances, user's retrieval is input as two dimensional image or figure, and is as shown in Figure 2 with the coupling detailed process of three-dimensional grid model.At first, the present invention adopts the profile (or graph outline) of image and the three-dimensional grid model profile in the three-dimensional grid model storehouse to carry out initial matching earlier; After the outline, the three-dimensional grid model that obtains is projected to the 2D space according to matching parameter, obtain projected image, carry out further accurately coupling with image relevant matches method then.

The profile that all grid models in the three-dimensional grid model storehouse have been accomplished X, Y, three positive dirctions of Z extracts and projection calculating.

For three-dimensional grid model, the inventive method is the profile that extract on each limit in the traversal grid, and concrete grammar is following:

1. if only be connected with a triangle when the front, it belongs to profile so;

2. if as front and two triangle F ₁And F ₂, then define its normal vector and be respectively

With

, current lens location and the vector of working as between the summit in front do

If , promptly With

Axle with respect to camera lens is in different directions, and F is described ₁And F ₂One face mirror first back to camera lens, therefore when the front be silhouette edge, otherwise, when the front is not significant silhouette edge.

For figure, do not need the profile leaching process.

For image, can adopt operator extraction profiles such as Sobel, Prewitt.

Outline process of the present invention can adopt based on the correspondingly-shaped matching process, like the Hausdoff distance; Perhaps polygon decomposes matching process; As carry out the line segment processing, and be that triangulation is carried out at the center with the profile center of gravity, confirm border vertices; Couple together and just become a convex polygon; With the former figure of this convex polygon approximate representation,, carry out the graph outline coupling thereby form a series of triangles with polygon center of gravity and summit line.The outline process is simply efficient, but the Rapid Realization initial retrieval.

Image relevant matches method of the present invention is two dimensional image or the figure of figure and input or the degree of correlation between the image that the projection of Calculation of Three Dimensional grid model obtains, as utilizes each similarity to pixel in the normalization correlated measure computed image zone.For two width of cloth image I to be matched ₁(x, y) and I ₂(x, y), mapping to be checked position (i, j) go up the normalization correlated measure and be defined as:

$\mathrm{\ρ}(i,j)=\frac{\underset{u=1}{\overset{m}{\mathrm{\Σ}}}\underset{v=1}{\overset{n}{\mathrm{\Σ}}}(I_1(u,v)-{\stackrel{\‾}{I}}_1)\·(I_2(u+i,v+j)-{\stackrel{\‾}{I}}_2)}{\sqrt{\underset{u=1}{\overset{m}{\mathrm{\Σ}}}\underset{v=1}{\overset{n}{\mathrm{\Σ}}}{(I_1(u,v)-{\stackrel{\‾}{I}}_1)}^2\underset{u=1}{\overset{m}{\mathrm{\Σ}}}\underset{v=1}{\overset{n}{\mathrm{\Σ}}}{(I_2(u+i,v+j)-{\stackrel{\‾}{I}}_2)}^2}}$

When the degree of correlation greater than a setting threshold, promptly retrieve the three-dimensional grid model of coupling.

(3) extraction of three-dimensional grid model skeleton

Skeleton is the good descriptive geometry characteristic of a kind of character, claims axis (Medial Axis) again, is a kind of effective pattern description means.As its name suggests, skeleton is a kind of solid of line style, occupy the symcenter of figure, and the topological structure identical with former figure arranged, and is keeping the shape information of former figure.

Fig. 3 is the processing flow chart when importing the 3D grid data for the user.

The process of Fig. 3 comprises following process, i.e. the extraction of (3) three-dimensional grid model skeleton; (4) carry out three-dimensional model search according to skeleton; (5) three-dimensional grid model feature point extraction; (6) calculating at three-dimensional grid model reference mark; (7) calculating of three-dimensional grid model reference mark frequency spectrum; (8) calculate the frequency spectrum similarity, go out graph of a correspondence according to similarity retrieval.Description below particular content is participated in.

The present invention has designed a kind of fast based on the skeletal extraction algorithm of multi-resolution grid: at first set up progressive grid representation for three-dimensional grid model; Then progressive grid is constantly carried out the limit conversion of collapsing; If a limit does not have adjacent triangle in the process of collapsing; This edge is labeled as the skeleton limit so, and remains into the end of collapsing always, and the final limit that obtains has just constituted the skeleton of grid model.All grid models in the three-dimensional grid model storehouse have been accomplished skeletal extraction and have been calculated.When importing grid model, the user need extract skeleton.

(4) carry out three-dimensional model search according to skeleton

Extracting skeleton is the process that the 3D figure is converted into the 3D line segment, and the data volume of 3D line segment significantly reduces with respect to original 3D figure, therefore can accelerate retrieval rate.

The three-dimensional grid model of user's input and the grid model in the storehouse carry out skeleton relatively; Relatively skeleton adopts major component PCA analytic approach; With the skeleton location, the distance that skeleton is calculated in segmentation directly adopts Euclidean distance to sort then earlier; According to the range information comparative result, the three-dimensional grid model that is preliminary search that the Euclidean distance difference is minimum.

(5) three-dimensional grid model feature point extraction

According to the three-dimensional grid model of graphics theory with user's input, carry out feature point extraction according to its spatial form, replace original three-dimensional grid model, significantly reduce data volume.All grid models in the three-dimensional grid model storehouse have been accomplished feature point extraction, the reference mark frequency spectrum calculates.

The present invention uses the unique point of umbilical cord point as the arbitrary triangle three-dimensional grid model.With the unique point of umbilical cord point as three-dimensional grid model surface, for noise, shearing, rotation, translation, convergent-divergent, different resolution sampling etc. because the influence that different acquisition equipment causes has very strong robustness.Owing to the curvature on the limit that strides across grid model is bigger, so curvature tensor can be defined as each point on the limit of grid model.In any net region B, the curvature tensor of fixed point v can be estimated with following formula:

$T\left(v\right)=\frac{1}{\left|B\right|}\underset{e\∈B}{\mathrm{\Σ}}\mathrm{\β}\left(e\right)|e\∩B|\stackrel{\‾}{e}{\stackrel{\‾}{e}}^T---\left(3\right)$

Wherein, | B| is the area of the neighborhood of v; β (e) is the angle between two leg-of-mutton normal vectors of adjacency of limit e; | e ∩ B| is the length of the limit e in the area B, and

is the unit normal vector of the e along the limit.The neighboring region B of v is by being the centre of sphere with v, is that the spheroid of radius and the crossing circle of grid model define with r.Radius r is the scale parameter of specifying Curvature Estimation.

Obtain after the curvature tensor on each summit, in the enterprising line linearity interpolation of each triangular mesh to obtain continuous curvature tensor field.The normal vector direction on summit is corresponding with the eigenwert of the amplitude of curvature tensor minimum; Other two eigenwerts are minimum curvature and the maximum curvature of corresponding vertex v respectively; When these two eigenwerts equated, vertex v was called as umbilical cord point, the unique point of the three-dimensional grid model that promptly uses among the present invention.

Obviously the key of Curvature Estimation is a scale parameter.The present invention adopts different yardsticks to carry out the curvature tensor estimation, so that the umbilical cord point under level and smooth tensor field and the estimation different scale is for the robustness of factors such as noise and affined transformation.For fear of losing the surperficial local message of grid model and reducing computation complexity, scale parameter can not be obtained excessive.The present invention selects for use self-adapted genetic algorithm as the optimization searching technology when seeking the highest scale parameter of robustness.In the highest several scales parameter of robustness, we select the maximum point of mean curvature that makes in the area B, the i.e. maximum point of the mark of curvature tensor for use.

(6) calculating at three-dimensional grid model reference mark

Unique point according to extracting is carried out triangulation to three-dimensional grid model, and the cut-off rule behind the subdivision is carried out piecewise fitting, obtains cut-point, with the reference mark of these cut-points as original three-dimensional grid model.

Obtain after the unique point, next step task is to reduce data volume, and this retrieval for characteristic registration and three-dimensional grid model is all extremely important.

Obtain after the unique point, the first step is grid to be carried out triangle cut apart.In the 2D space, the Delaunay triangle is cut apart can produce shape triangle uniformly, and has uniqueness.In the 3D surface, the Delaunay triangle is cut apart and is not to use Euclidean distance and is to use geodesic distance (geodesic distance is meant two summit bee-lines surfacewise of surface mesh in the 3d space).The present invention adopts the wave front method that whole Voronoi figure and double Delaunay triangle thereof are cut apart to estimate.The benefit that adopts this method is to make triangle cut apart the influence of the size that does not receive sampling rate.Wave front is to be the centre of sphere through calculating with the seed points, the crossing circle acquisition on the ever-increasing ball of radius and 3D grid surface.Employing is better than the result that the wave front method that adopts based on the network topology relation obtains based on the wave front method of the different radiuses of a ball.

(7) calculating of three-dimensional grid model reference mark frequency spectrum

Topological structure according to three-dimensional grid model carries out frequency domain transform to the reference mark, the frequency domain coordinate figure of controlled point.

Because the 3D grid is a figure and non-image,, therefore the 3D grid is carried out transform domain and handle and at first will construct an image function, so that the transform domain Processing Algorithm of classics is applied to the 3D Mesh Processing so each apex coordinate does not have its intrinsic image function.

At first, sorted in the reference mark.The present invention sorts to the reference mark of coarse grids with the mould of reference mark to the vector of grid element center.

The coordinate of definition grid element center does

$v_g=(x_g,y_g,z_g)=\underset{i=0}{\overset{N-1}{\mathrm{\Σ}}}{v}_i---\left(4\right)$

Each reference mark defines as follows to the mould of the vector of grid element center

${\mathrm{\ρ}}_i=\sqrt{{(x_i-x_g)}^2+{(y_i-y_g)}^2+{(z_i-z_g)}^2}---\left(5\right)$

The reference mark is carried out beginning to carry out frequency domain transform after ascending order arranges according to the mould size.At first from the network topology relation, obtain the combination Laplace operator or be called the Kirchhoff matrix.This defined matrix is following:

K＝D-A (6)

Wherein, D is a diagonal matrix, the element Dii on its diagonal line corresponding with the valency of vertex v i (valency promptly is the number on the limit of radiating from the summit), and A is the adjacency matrix of grid, its element definition is following:

For the grid model that n reference mark arranged, the size of matrix A, D and K is n*n.The Kirchhoff matrix is carried out characteristic value decomposition obtain n eigenvalue _iWith n proper vector w _iThis n proper vector is carried out ascending order arrange, can obtain its characteristic of correspondence vector, the proper vector after these orderings is the function base that frequency constantly increases.This function base only depends on the topological structure of grid model, and irrelevant with the geometrical property of grid model.The n*n mapping matrix note that proper vector after this n the ordering is formed is W.

3 vectors of the volume coordinate at sorted n reference mark structure before before:

X＝(x ₁，x ₂，…，x _n)，Y＝(y ₁，y ₂，…，y _n)，Z＝(z ₁，z ₂，…，z _n) (8)

These 3 vector projections to proper vector base W, can be obtained the frequency domain branch solution vector of spatial domain coordinate:

$\left\{\begin{array}{c}{X}_s=\mathrm{WX}\\ Y_s=\mathrm{WY}\\ Z_s=\mathrm{WZ}\end{array}\right.---\left(9\right)$

Volume coordinate also can be obtained by the frequency coordinate reduction:

$\left\{\begin{array}{c}X=W^{-1}{X}_s\\ Y=W^{-1}{Y}_s\\ Z=W^{-1}{Z}_s\end{array}\right.---\left(10\right)$

The amplitude S of the frequency spectrum that each summit is corresponding _iCan be by computes:

$S_i=\sqrt{{\left|\right|X_s\left|\right|}^2+{\left|\right|Y_s\left|\right|}^2+{\left|\right|Z_s\left|\right|}^2}---\left(11\right)$

(8) calculate the frequency spectrum similarity, go out graph of a correspondence according to similarity retrieval.

The degree of correlation between the reference mark frequency domain coordinate figure of the three-dimensional grid model of calculating user input and the reference mark frequency domain coordinate figure of the three-dimensional grid model in the storehouse retrieves graph of a correspondence according to the degree of correlation.

When comparing the similarity of two grid models, then judge through the wave-form similarity that compares frequency coefficient.Criterion when comparing similarity can adopt normalized correlation coefficient:

Wherein is the spectrum value that is calculated by the three-dimensional grid model that the user imports, and is the 3D grid spectrum value in the storehouse.NC is big more, and then spectral coefficient is relevant more.

A given threshold value T, if NC＞T thinks that then 3D grid that the user imports and the 3D grid in the storehouse are complementary, retrieval flow finishes.

The present invention has following beneficial effect:

(1) can be according to single two dimensional image retrieval three-dimensional picture, the user can import the picture of common forms such as bmp, jpeg, tiff, can retrieve graph of a correspondence through the inventive method.

(2) method for expressing of the three-dimensional model of support input is more extensive, all is suitable for for triangular mesh data, cloud data, volume data or polygonal grid model.

(3) adopted hierarchical search, more accurate sooner.At first the skeleton according to three-dimensional picture carries out coarse search, adopts the further feature extractive technique accurately to search for to the three-dimensional picture in the Search Results, has guaranteed real-time and accuracy.

One of ordinary skill in the art will appreciate that all or part of step in the whole bag of tricks of the foregoing description is to instruct relevant hardware to accomplish through program; This program can be stored in the computer-readable recording medium; Storage medium can comprise: ROM (read-only memory) (ROM; Read Only Memory), RAS (RAM, Random Access Memory), disk or CD etc.

More than to a kind of figure retrieving method that the embodiment of the invention provided; Carried out detailed introduction; Used concrete example among this paper principle of the present invention and embodiment are set forth, the explanation of above embodiment just is used for helping to understand method of the present invention and core concept thereof; Simultaneously, for one of ordinary skill in the art, according to thought of the present invention, the part that on embodiment and range of application, all can change, in sum, this description should not be construed as limitation of the present invention.

Claims (5)

1. a figure retrieving method is characterized in that, comprising:

1) sets up the three-dimensional grid model storehouse;

2) when user input be X-Y scheme or image the time; Mate with the profile of three-dimensional grid model in the three-dimensional grid model storehouse; According to matching parameter three-dimensional grid model is projected to two-dimensional space; Obtain the two dimensional image or the figure of projection, calculate two dimensional image or the figure of figure and input or the degree of correlation between the image that projection obtains then, retrieval obtains three-dimensional grid model according to the degree of correlation;

3) when user input be three-dimensional grid model the time, the three-dimensional grid model of input is carried out skeletal extraction, according to the three-dimensional grid model skeleton that extracts, preliminary search obtains three-dimensional grid model in the three-dimensional grid model storehouse;

The three-dimensional grid model of the three-dimensional grid model that 4) retrieval is obtained and user's input carries out feature point extraction; Replace original three-dimensional grid model; Carry out triangulation again; Cut-off rule behind the subdivision is carried out piecewise fitting, obtain the reference mark of original three-dimensional grid model, according to topological structure frequency domain transform is carried out at the reference mark then;

5) user who calculates imports the similarity between the reference mark frequency domain coordinate figure of reference mark frequency domain coordinate figure and the three-dimensional grid model in the three-dimensional grid model storehouse of three-dimensional grid model, goes out graph of a correspondence according to similarity retrieval;

Wherein, the topological structure according to three-dimensional grid model in the step 4) carries out frequency domain transform to the reference mark that obtains, and comprising:

(1) with the mould of reference mark sorted in the reference mark of coarse grids to the vector of grid element center;

(2) obtain the Kirchhoff matrix from the network topology relation

K＝D-A (6)

D is a diagonal matrix, and the element Dii on its diagonal line is corresponding with the valency of vertex v i, and A is the adjacency matrix of grid;

The Kirchhoff matrix is carried out n the proper vector w that characteristic value decomposition obtains _iCarry out ascending order and arrange the n*n mapping matrix W of composition;

(3) 3 vectors of the volume coordinate at sorted n reference mark structure before before:

X＝(x ₁，x ₂，…，x _n)，Y＝(y ₁，y ₂，…，y _n)，Z＝(z ₁，z ₂，…，z _n)?(8)

These 3 vector projections are obtained frequency domain vector to the proper vector base W:

$\left\{\begin{array}{c}{X}_s=\mathrm{WX}\\ Y_s=\mathrm{WY}\\ Z_s=\mathrm{WZ}\end{array}\right.---\left(9\right)$

The amplitude S of the frequency spectrum that each summit is corresponding _iComputing formula is:

$S_i=\sqrt{{\left|\right|X_s\left|\right|}^2+{\left|\right|Y_s\left|\right|}^2+{\left|\right|Z_s\left|\right|}^2}---\left(11\right).$

2. figure retrieving method according to claim 1 is characterized in that:

Three-dimensional grid model storehouse in the step 1) to various three-dimensional data forms reorganize with the polygon modeling after obtain.

3. figure retrieving method according to claim 1 and 2 is characterized in that:

The skeletal extraction process is: at first set up progressive grid representation for the grid model of input; Then progressive grid is constantly carried out the limit conversion of collapsing; If a limit does not have adjacent triangle in the process of collapsing; Then this limit is labeled as the skeleton limit, and remains into the end of collapsing always, the skeleton of the final limit component model that obtains.

4. figure retrieving method according to claim 1 and 2 is characterized in that:

The three-dimensional grid model of the three-dimensional grid model that obtains for retrieval in the step 4) and user's input carries out feature point extraction according to its spatial form, replaces original three-dimensional grid model with umbilical cord point as unique point.

5. figure retrieving method according to claim 1 and 2 is characterized in that:

According to unique point three-dimensional grid model is carried out triangulation in the step 4), the cut-off rule behind the subdivision is carried out piecewise fitting, obtain cut-point, with the reference mark of these cut-points as original three-dimensional grid model.

Images