CN1340791A - Method and device for execute linear interpotation of three-dimensional pattern reestablishing - Google Patents
Method and device for execute linear interpotation of three-dimensional pattern reestablishing Download PDFInfo
- Publication number
- CN1340791A CN1340791A CN00131687A CN00131687A CN1340791A CN 1340791 A CN1340791 A CN 1340791A CN 00131687 A CN00131687 A CN 00131687A CN 00131687 A CN00131687 A CN 00131687A CN 1340791 A CN1340791 A CN 1340791A
- Authority
- CN
- China
- Prior art keywords
- imaging
- interpolation
- pixel
- series
- part imaging
- 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.)
- Pending
Links
Images
Landscapes
- Image Processing (AREA)
Abstract
A quick linear interpolation method for reproducing 3D image and its device are disclosed. The pixels of adjacent original image parts are paired to obtain the interpolated image part. The said original and interpolated image parts are used to reproduce a 3D object. An outline pixel can have zero, one, or more thain one pair. A pixel pair can be used to determine the coordinate of pixel in the said interpolated image part. Its advantage are linear interpolation, high speed and high adaptability.
Description
The present invention relates to the method and apparatus that carries out image is rebuild, relate in particular to the method and apparatus of three-dimensional image reconstruction being carried out the fast linear interpolation.
Three-dimensional image reconstruction is to develop in the later stage of the eighties, and in a lot of fields available, particularly medical science and biomedical field.Computer topography (CT) (CT) is the example of an application, wherein becomes two-dimension image rebuild objective to be used for diagnosis.Three-dimensional image reconstruction uses a series of two dimensional image parts, so that rebuild the original objective of an original imaging.Three-dimensional image reconstruction is the process through relatively selecting preferably, i.e. cutting is used for the organ-tissue of micro.
Two kinds of reconstruction algorithm are generally arranged, and one is used linear interpolation, and another uses nonlinear interpolation.E.Keppel, " by the approximate complex surface of the triangulation of isoline " literary composition (IBM researches and develops monthly magazine, volume 19,2-11 page or leaf, in January, 1975) a kind of initial linear interpolation three-dimensional reconstruction technology is disclosed.As shown in Figure 1, the technology of Keppel is according to triangulation, and wherein key is the triangle of this target surface of search expression in a large amount of candidate's triangles.Suppose that first two dimensional image partly has m isoline pixel, and second two dimensional image partly have n isoline pixel, and [(m-1)+(n+1)] then arranged! [(m-1)! (n-1)! ] individual triangle can be used for rebuilding the surface of original objective.And if m=n=12, then leg-of-mutton quantity is 10
7
Developed many algorithms after the technology of Keppel, to improve the interpolation result, some algorithm comprises the search at an optimal path.Yet most algorithm is according to triangulation.
Nonlinear interpositioning also is used to three-dimensional reconstruction, and than linear interpolation technology adaptability is arranged more usually.Example comprises W.C.Lin etc. " being used for " (IEEE communication from the dynamic elasticity interpolation method of continuous cross-section imaging reconstruction 3-D target, medical imaging, volume 3,225-232 page or leaf in September, 1988) and W.C.Lin et al. " being used for rebuilding from the end view drawing of series a kind of new surperficial interpositioning of 3D target " (computer graphical image is handled, volume 48, the 124-143 page or leaf, 1989) these two technology all obtain rather good reconstructed results; It can adapt to different application with " rebuilding shape from plane section " (computer graphical image is handled, volume 44,1-29 page or leaf, 1988) of J.D.Boissnnat.
Lin etc. " being used for " one literary composition from the dynamic elasticity interpolation method of continuous cross-section imaging reconstruction 3-D target a kind of nonlinear interpositioning is disclosed, the vacancy that is used between the profile of two continuous parts is filled, so that recapture the profile of an objective.Its total idea is that sign acts on a stress field of first profile, and attempts to be out of shape this first profile, so as with next-door neighbour's profile phase picture.The advantage of this method comprise it the ability of control branch problem and it diagram description method is applied to comprise the solid target of volume pixel parts.Be used for giving a hand, or improve the automatic operation and the analysis of this object construction from the technology of serial section reconstruction of three-dimensional (3-D) target or aspect the understanding of object construction.
A literary composition disclosed dynamic elasticity profile interpositioning, spline is theoretical " being used for rebuilding from the end view drawing of series a kind of new surperficial interpositioning of 3D target " of Lin etc. and based on the combination of the surperficial interpositioning of quadratic variation.To be by carrying out an elasticity interpolation algorithm form so that produce a series of intermediate profile between the continuous cross section of every a couple in the initial description of objective.Subsequently, by this profile map being become a surface function territory, using spline function to calculate the rough handling that this initial table face amount execution is used for surface calculating then.According to the output of this rough handling, be used to the calculating of final surface demonstration based on the surperficial interpolation algorithm of quadratic variation.This method is utilized the continuity of higher derivative so that produce the level and smooth and complete surface of this objective.
Volume of article structure of Boissonnat, its border is the polyhedron with gore, intersects this cutting planes along given profile.Pursue by the Delaunay triangle that between two adjacent sections, forms of deletion and partly to obtain this volume.Can use an algorithm to calculate the Delaunay triangle efficiently, this algorithm only uses two-dimensional operation to optimize the size of input and output.This shape method for reconstructing is from the extract maximum successive value of no singular point of this triangulation, and the variation of quantity from a kind of section to the another one section of profile effectively, therefore limit the profile that is branched with porose target.
Yet aforesaid non-linear interpolation technology is normally complicated, and needs a large amount of calculating.
The result is, needs adaptable and not according to the linear interpolation of the high speed and the high adaptivity of triangulation, is used for 3-D view and reproduces.
The present invention is intended to a kind of method and apparatus, is used for 3-D view is reproduced execution fast linear interpolation.Total design of the present invention comprises: the pixel in the adjacent initial protion imaging is connected, so that obtain the part imaging of an interpolation, and use the part imaging of this original part imaging and this interpolation to recover this three-dimensional target.
In a most preferred embodiment, the present invention is intended to a kind of method that comprises three key steps: a pre-treatment step, an interpolation step and an objective recovering step.In this pre-treatment step, the part imaging of input is prepared interpolation.This can comprise that the adjacent importation imaging of assurance is in correct associated orientation.Can realize correct orientation by sign being added to the importation imaging, it can be arranged on before the interpolation, so that guarantee correct associated orientation.This pretreatment steps can also comprise detection roughly profile, the feature of extracting and contour segmentation point (manually or automatically) is set.
In the interpolation step, definite each contour pixel from first's imaging is to the distance of each contour pixel in the adjacent part imaging.The contour pixel with distance each contour pixel bee-line in this first's imaging in adjacent part imaging is coupled, promptly with this contour pixel pairing.One contour pixel can have zero, one or more than one pairing.The pairing of this pixel is used for determining the coordinate in the pixel of the part imaging of interpolation.This interpolation step can select to comprise a branch process substep, and wherein uniting of all possible branch is used to determine a plurality of branches.
As the result of the connection features of interpolation step, the surface that generates objective is level and smooth and complete.This interpolation step is linear and not according to triangulation.And because the low computational complexity of this interpolation step, method of the present invention and device have fast and very high adaptability.
Fig. 1 illustrates the linear interpolation technology according to a prior art of triangulation.
Fig. 2 illustrates the method for the present invention's one most preferred embodiment.
Fig. 3 illustrate the present invention's one most preferred embodiment in two adjacent importation imagings and the part imaging of an interpolation.
Fig. 4-6 illustrates the solution for bifurcation problem in most preferred embodiment of the present invention.
A most preferred embodiment of the present invention is shown in Figure 2.The present invention includes three key steps: pre-treatment step 100, interpolation step 200 and objective recovering step 300.This pre-treatment step 100 is prepared and directed a series of profile or part imaging 110, and it constitutes the input of this method; This interpolation step 200 is relevant the pixel of adjacent part imaging, so that produce the part imaging 210 of interpolation; And target recovering step 300 that should three-dimensional produces these objectives according to the part imaging 110 and the interpolation part imaging 210 of input.
In order to guarantee the correct execution of interpolation step, the part imaging 110 of input must be located on the right, so that they can be by relative orientation correctly each other.A kind ofly be used to realize that the technology of correct associated orientation is, when each part imaging 110 is cutting, in each part imaging 110, place sign.Fig. 3 illustrates two adjacent part imagings 1102,1104, and each comprises three signs 112.In pre-treatment step 100 processes, sign 112 is used to correctly make part imaging 1102 and 1104 relative orientations.Though Fig. 3 illustrates three signs 112, can use any according to the degree of accuracy of expecting greater than two number.
Pre-treatment step 100 also may comprise detection profile roughly and extract.These two functions can be carried out by the known conventional art of any those skilled in the art.This pre-treatment step 100 can also comprise contour segmentation point (manually or automatically) is set, and use the known any conventional art of those skilled in the art to follow the trail of each part imaging 110, so that obtain the single pixel of a series of continuous, level and smooth these part imagings 110 of composition.The conventional method of a demonstration is the Seitz algorithm, but also can use other ordinary skill.
The target of interpolation step 200 is the part imagings 210 that produce interpolation.Illustrate as Fig. 3, suppose C
1And C
2Be that part imaging 1102 and 1104 is respectively at plane Z
1And Z
2In contour curve, then
C
1＝{P
I，l≤I≤n
1} (1)
C
2＝{P
i，l≤j≤n
2} (2)
If n
1〉=n
2And k=n
2/ n
1, then the pixel of this interpolation step 200 pairing substep 202 can be carried out as follows.If the pixel pairing is from pixel P
1Beginning, first step are to calculate at P
1And C
2In each contour pixel between distance, such as Dj (l≤j≤n
2).If Dm=min (Dj) is then at C
2In pixel Qm be P
1The pairing of pixel.Carry out pixel pairing step 200, so that make other pixel P
J+1, P
J+2, P
N1, P
1, P
J+1With pixel Qm, Qm+1 ..., Q
N2Pairing.Among Fig. 3, Ql ..., Qm-1, P
J+1With Qint (m+k) pairing, P
J+2With Qint (m+2k) pairing etc.
As above set forth, C1 is at plane Z=Z
1In and C
2At Z=Z
2In.If C ' is the part imaging 210 (in order clearly not illustrate) of the interpolation between part imaging 1102 and 1104, then C ' is in the Z=Z ' of plane.As shown in Figure 3, P
1Coordinate be (x
i, y
i, Z), the coordinate of Qm is (x
m, y
m, z
2), and the coordinate of the interpolated pixel S of the part imaging C ' of interpolation be (x, y, z).Thus:
X=x
i-(x
i-x
m) x[(z
1-z)/(z
1-z
2)] and (3)
x＝y
i－(y
i－y
m)x[(z
1－z)/(z
1－z
2)] (4)
If Δ x=x
i-x
m, Δ y=y
i-y
m, Δ z=z
1-z
2, Δ z '=z
1-z, then
X=x
i-Δ xx (Δ z '/Δ z) and (5)
y＝y
i－Δxx(Δz′/Δz)， (6)
It defines the coordinate of each pixel in the part imaging 210 of the interpolation between importation imaging 1102 and 1104.
Equation (5) and (6) can be repeated, so that obtain the coordinate of the part imaging 210 of the interpolation between each phase adjacency pair of part imaging 110.
As above discuss, the processing of branch often is a problem in the application of many three-dimensional image reconstructions.Two good examples are the nerve in this biomedical sector or the imaging of tracheae.As shown in Figure 4, part imaging C has three branches, C
1, C
2And C
3At part imaging C and three C of branch
1, C
2And C
3Between the part imaging 210 of interpolation can be overlapping, as shown in Figure 5.
This interpolation step 200 has and selectively comprises a branch process substep 204.In pixel pairing substep 202, (C, C
1), (C, C
2), (C, C
3) each matched respectively, and the part imaging 210 of this interpolation comprises three independent part imaging fragment I
1, I
2, I
3As shown in Figure 6, this interpolation part imaging 210 is restricted to the associating of three independent part imaging fragments:
I
1UI
2UI
3
In case the part imaging 110 according to input produces the part imaging 210 of enough interpolations and decomposed any branch, then objective recovering step 300 will produce three-dimensional target by any known routine techniques.As the result of the part imaging 210 of interpolation, the surface that generates objective is level and smooth with complete.This interpolation step 200 is linear and not according to triangulation.And because the low computational complexity of this interpolation step, the demonstration methods of Fig. 3 has high-speed and very high adaptability.
Though above-mentioned minimum spacing function can also use known other technology of those skilled in the art, characteristic or parameter as the technology of the adjacent part imaging of pairing.
Though above-mentioned the present invention describes in conjunction with contour pixel, the combination of the grouping of any pixel can be used to interpolation.
Though method of the present invention is described in conjunction with Fig. 2-6, device of the present invention can comprise the processor of any kind, realizes with hardware or the software of carrying out above-mentioned functions.
Therefore description of the invention significantly can change in many aspects.This variation is not considered to depart from the spirit and scope of the invention, and all this modifications all will be conspicuous to those skilled in the art.All will be included within the scope of following claim.
Claims (14)
1. execution linear interpolation method that is used for three-dimensional image reconstruction comprises step:
Prepare a series of part imagings that are used for the input of interpolation; And
By at least two adjacent pixels of pairing in the imaging series of said importation, determine the part imaging of at least one interpolation.
2. according to the method for claim 1, also comprise step:
Produce the target of a three-dimensional according to the part imaging of the part imaging series of this input and this at least one interpolation.
3. match this pixel according to the process of claim 1 wherein according to minimum spacing.
4. be contour pixel according to the pixel that the process of claim 1 wherein.
5. according to the method for claim 1, said preparation process comprises,
Each the pixel of series of ground directed importation imaging is relative to each other.
6. according to the method for claim 5, wherein the pixel of the part imaging series of each input is used the sign of the series that is added to this importation imaging and is orientated with being relative to each other.
7. according to the method for claim 1, also comprise step:
Handle the branch in the series of part imaging of input according to associative function, so that be created in the branch at least one interpolation part imaging.
8. be programmed a kind of device of carrying out linear interpolation, being used for three-dimensional image reconstruction, comprise:
A processor is prepared a series of part imagings that are used for the input of interpolation, and by at least two adjacent pixels of pairing in the imaging series of said importation, determines the part imaging of at least one interpolation.
9. device according to Claim 8, wherein said processor is exported part imaging series and at least one interpolation part imaging of this input, to allow to produce the target of a three-dimensional.
10. device according to Claim 8 wherein matches this pixel according to minimum spacing.
11. device according to Claim 8, wherein this pixel is a contour pixel.
12. according to right. require 8 device, the pixel of each of the importation imaging series that the further orientation of wherein said processor is relative to each other.
13. according to the device of claim 12, wherein the pixel of the part imaging series of each input is used the sign of the series that is added to this importation imaging and is orientated with being relative to each other.
14. device according to Claim 8, said processor are also handled branch in the series of part imaging of input according to associative function, so that be created in the branch at least one interpolation part imaging.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN00131687A CN1340791A (en) | 2000-08-29 | 2000-08-29 | Method and device for execute linear interpotation of three-dimensional pattern reestablishing |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN00131687A CN1340791A (en) | 2000-08-29 | 2000-08-29 | Method and device for execute linear interpotation of three-dimensional pattern reestablishing |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN1340791A true CN1340791A (en) | 2002-03-20 |
Family
ID=4594777
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN00131687A Pending CN1340791A (en) | 2000-08-29 | 2000-08-29 | Method and device for execute linear interpotation of three-dimensional pattern reestablishing |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1340791A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100369584C (en) * | 2003-09-29 | 2008-02-20 | Ge医疗***环球技术有限公司 | Image reconstruction method and X ray CT device |
| CN101976462A (en) * | 2010-10-29 | 2011-02-16 | 中国测绘科学研究院 | Three-dimensional reconstruction method |
| CN101266690B (en) * | 2007-03-15 | 2012-02-29 | 华南农业大学 | Plant root species form 3-D image reconstruction system and method |
| CN101617342B (en) * | 2007-01-16 | 2012-06-13 | 汤姆科技成像***有限公司 | A method and a system for graphic representation of dynamic information |
| CN101627410B (en) * | 2007-03-15 | 2012-11-28 | Gvbb控股股份有限公司 | Method and apparatus for automated aesthetic transitioning between scene graphs |
-
2000
- 2000-08-29 CN CN00131687A patent/CN1340791A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100369584C (en) * | 2003-09-29 | 2008-02-20 | Ge医疗***环球技术有限公司 | Image reconstruction method and X ray CT device |
| CN101617342B (en) * | 2007-01-16 | 2012-06-13 | 汤姆科技成像***有限公司 | A method and a system for graphic representation of dynamic information |
| CN101266690B (en) * | 2007-03-15 | 2012-02-29 | 华南农业大学 | Plant root species form 3-D image reconstruction system and method |
| CN101627410B (en) * | 2007-03-15 | 2012-11-28 | Gvbb控股股份有限公司 | Method and apparatus for automated aesthetic transitioning between scene graphs |
| CN101976462A (en) * | 2010-10-29 | 2011-02-16 | 中国测绘科学研究院 | Three-dimensional reconstruction method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP1303839B1 (en) | System and method for median fusion of depth maps | |
| Remondino et al. | Turning images into 3-D models | |
| CN1295650C (en) | Object-correspondence identification without full volume registration | |
| Han et al. | Stereo matching using genetic algorithm with adaptive chromosomes | |
| JPH11502335A (en) | Apparatus and method for providing depth modeling and depth information of moving object | |
| CN111882668A (en) | Multi-view three-dimensional object reconstruction method and system | |
| Holzmann et al. | Semantically aware urban 3d reconstruction with plane-based regularization | |
| Liu et al. | High-quality textured 3D shape reconstruction with cascaded fully convolutional networks | |
| Sugiura et al. | 3D surface reconstruction from point-and-line cloud | |
| CN109215118B (en) | Incremental motion structure recovery optimization method based on image sequence | |
| CN1340791A (en) | Method and device for execute linear interpotation of three-dimensional pattern reestablishing | |
| Liu et al. | Integrating cross-sectional imaging based reverse engineering with rapid prototyping | |
| CN110390336B (en) | Method for improving feature point matching precision | |
| CN118052828A (en) | Modifying a two-dimensional image using a segmented three-dimensional object grid of the two-dimensional image | |
| KR20030054360A (en) | Apparatus and Method for Converting 2D Images to 3D Object | |
| Zakaria et al. | Interactive stylized silhouette for point-sampled geometry | |
| CN113593007A (en) | Single-view three-dimensional point cloud reconstruction method and system based on variational self-coding | |
| Bilodeau et al. | Generic modeling of 3d objects from single 2d images | |
| KR20030017888A (en) | Triangular mesh segmentation apparatus and method based on surface normal | |
| CN116740300B (en) | Multi-mode-based prime body and texture fusion furniture model reconstruction method | |
| Salman et al. | High resolution surface reconstruction from overlapping multiple-views | |
| Xu et al. | Animated 3d line drawings with temporal coherence | |
| CN112116710B (en) | Curved surface reconstruction method based on trend constraint | |
| CN111862139B (en) | Dynamic object parametric modeling method based on color-depth camera | |
| Chou et al. | Multi-level 3d reconstruction with visibility constraints |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |